Multi-agent collaborative perception (CP) is a promising paradigm for improving autonomous driving safety, particularly for vulnerable road users like pedestrians, via robust 3D perception. However, existing CP approaches often optimize for vehicle detection performance metrics, underperforming on smaller, safety-critical objects such as pedestrians, where detection failures can be catastrophic. Furthermore, previous CP methods rely on full feature exchange rather than communicating only salient features that help reduce false negatives. To this end, we present FocalComm, a novel collaborative perception framework that focuses on exchanging hard-instance-oriented features among connected collaborative agents. FocalComm consists of two key novel designs, (1) a learnable progressive hard instance mining (HIM) module to extract hard instances-oriented features per agent, and (2) a query-based feature-level (intermediate) fusion technique that dynamically weights these identified features during collaboration. We show that FocalComm outperforms state-of-the-art collaborative perception methods on two challenging real world datasets (V2X-Real and DAIR-V2X) across both vehicle-centric and infrastructure-centric collaborative setups. FocalComm also shows strong performance gain in pedestrian detection in V2X-Real. Code and model checkpoints are available at https://github.com/scdrand23/FocalComm.

Accurate prediction of pedestrian trajectories is crucial as autonomous vehicles become more prevalent on roads. The dynamic nature of urban environments and the less predictable behavior of pedestrians present significant challenges in developing reliable prediction models. Earlier methods relying on recurrent neural networks (RNNs) and long-shortterm memory (LSTM) networks have shown promise, but often fail to fully take advantage of the rich visual and contextual information available in real-world scenarios. Recent advances in vision-language models (VLMs) offer new opportunities to improve pedestrian trajectory prediction by incorporating multimodal reasoning capabilities. This paper introduces a novel approach that uses a powerful pre-trained VLM to improve the estimation of pedestrian trajectories. Specifically, we first enable learning of semantically useful scene context and high-level reasoning features via vision-language model finetuning on specific prompts using road scenes with pedestrians. Next, with the learned VLM features and the pedestrian's past trajectory history, we predict future trajectories using an encoder-decoder head. Through experiments with first-person datasets JAAD and PIE, we show that utilizing visual-linguistic semantics via a pretrained vision-language model outperforms previous methods in both deterministic and stochastic trajectory prediction setups.

Predictive modeling of unannotated spatiotemporal data presents inherent challenges, primarily due to the highly entangled visual dynamics in real-world scenes. To tackle these complexities, we introduce a novel insight through Disentangling Deterministic and Probabilistic (DDP) modeling. We note a key observation in spatiotemporal data where low-level details typically remain stable, whereas high-level motion frequently exhibits dynamic variations. The core motivation involves constructing two distinct pathways in the latent space: a deterministic path and a probabilistic path. The probabilistic path begins by defining the motion flow, which explicitly describes complex many-to-many motion patterns between patches, and models its probabilistic distribution using a motion diffuser. The deterministic path incorporates a spectral-aware enhancer to retain and amplify visual details in the frequency domain. These designs ensure visual consistency while also capturing intricate long-term motion dynamics. Extensive experiments demonstrate the superiority of DDP across diverse scenario evaluations.

The remarkable progress in neural-network-driven visual data generation, especially with neural rendering techniques like Neural Radiance Fields and 3D Gaussian splatting, offers a powerful alternative to GANs and diffusion models. These methods can generate high-fidelity images and lifelike avatars, highlighting the need for robust detection methods. However, the lack of any large dataset containing images from neural rendering methods becomes a bottleneck for the detection of such sophisticated fake images. To address this limitation, we introduce NeuroRenderedFake, a comprehensive benchmark for evaluating emerging fake image detection methods. Our key contributions are threefold: (1) A large-scale dataset of fake images synthesized using state-of-the-art neural rendering techniques, significantly expanding the scope of fake image detection beyond generative models; (2) A cross-domain evaluation protocol designed to assess the domain gap and common artifacts between generative and neural rendering-based fake images; and (3) An in-depth spectral energy analysis that reveals how frequency domain characteristics influence the performance of fake image detectors. We train representative detectors, based on spatial, spectral, and multimodal architectures, on fake images generated by both generative and neural rendering models. We evaluate these detectors on 15 groups of fake images synthesized by cutting-edge neural rendering models, generative models, and combined methods that can exhibit artifacts from both domains. Additionally, we provide insightful findings through detailed experiments on degraded fake image detection and the impact of spectral features, aiming to advance research in this critical area.

The use of vehicle-to-vehicle (V2V) and vehicle-to-everything (V2X) communications for the exchange of sensor data on roads is becoming an important strategy to improve the safety of both vehicles and other road users. Termed collaborative perception (CP), this method offers potential solutions to overcome limitations inherent to single-agent vehicle perception systems, including occlusions, non-line-of-sight scenarios, and the poor detection of distant objects. However, the need for exchanging sensor data and extracted information also presents communication challenges, particularly due to latency and bandwidth limitations. Earlier studies have focused on developing machine learning (ML) modules to address latency issues and compress data before transmission. This study investigates the impact of communication latency and bandwidth limitations on collaborative perception. We examine the interrelationships between CP performance and communication limitations. We present the results of numerical experiments using the V2X-Sim dataset on state-of-the-art CP methods. Our findings demonstrate the need for designing a CP system in a connected environment that effectively mitigates the impact of latency and communication interruptions, thus improving the reliability and safety of CP systems.

Deep learning recognition approaches can potentially perform better if we can extract a discriminative representation that controllably separates nuisance factors. In this paper, we propose a novel approach to explicitly enforce the extracted discriminative representation d, extracted latent variation l (e,g., background, unlabeled nuisance attributes), and semantic variation label vector s (e.g., labeled expressions/pose) to be independent and complementary to each other. We can cast this problem as an adversarial game in the latent space of an auto-encoder. Specifically, with the to-be-disentangled s, we propose to equip an end-to-end conditional adversarial network with the ability to decompose an input sample into d and l. However, we argue that maximizing the cross-entropy loss of semantic variation prediction from d is not sufficient to remove the impact of s from d, and that the uniform-target and entropy regularization are necessary. A collaborative mutual information regularization framework is further proposed to avoid unstable adversarial training. It is able to minimize the differentiable mutual information between the variables to enforce independence. The proposed discriminative representation inherits the desired tolerance property guided by prior knowledge of the task. Our proposed framework achieves top performance on diverse recognition tasks, including digits classification, large-scale face recognition on LFW and IJB-A datasets, and face recognition tolerant to changes in lighting, makeup, disguise, etc.

Weakly Supervised Object Detection (WSOD) has emerged as an effective tool to train object detectors using only the image-level category labels. However, without object-level labels, WSOD detectors are prone to detect bounding boxes on salient objects, clustered objects and discriminative object parts. Moreover, the image-level category labels do not enforce consistent object detection across different transformations of the same images. To address the above issues, we propose a Comprehensive Attention Self-Distillation (CASD) training approach for WSOD. To balance feature learning among all object instances, CASD computes the comprehensive attention aggregated from multiple transformations and feature layers of the same images. To enforce consistent spatial supervision on objects, CASD conducts self-distillation on the WSOD networks, such that the comprehensive attention is approximated simultaneously by multiple transformations and feature layers of the same images. CASD produces new state-of-the-art WSOD results on standard benchmarks such as PASCAL VOC 2007/2012 and MS-COCO.

This paper considers the problem of image set-based face verification and identification. Unlike traditional single sample (an image or a video) setting, this situation assumes the availability of a set of heterogeneous collection of orderless images and videos. The samples can be taken at different check points, different identity documents $etc$ . The importance of each image is usually considered either equal or based on a quality assessment of that image independent of other images and/or videos in that image set. How to model the relationship of orderless images within a set remains a challenge. We address this problem by formulating it as a Markov Decision Process (MDP) in a latent space. Specifically, we first propose a dependency-aware attention control (DAC) network, which uses actor-critic reinforcement learning for attention decision of each image to exploit the correlations among the unordered images. An off-policy experience replay is introduced to speed up the learning process. Moreover, the DAC is combined with a temporal model for videos using divide and conquer strategies. We also introduce a pose-guided representation (PGR) scheme that can further boost the performance at extreme poses. We propose a parameter-free PGR without the need for training as well as a novel metric learning-based PGR for pose alignment without the need for pose detection in testing stage. Extensive evaluations on IJB-A/B/C, YTF, Celebrity-1000 datasets demonstrate that our method outperforms many state-of-art approaches on the set-based as well as video-based face recognition databases.

The human visual system uses numerous cues for depth perception, including disparity, accommodation, motion parallax and occlusion. It is incumbent upon virtual-reality displays to satisfy these cues to provide an immersive user experience. Multifocal displays, one of the classic approaches to satisfy the accommodation cue, place virtual content at multiple focal planes, each at a different depth. However, the content on focal planes close to the eye do not occlude those farther away; this deteriorates the occlusion cue as well as reduces contrast at depth discontinuities due to leakage of the defocus blur. This paper enables occlusion-aware multifocal displays using a novel ConeTilt operator that provides an additional degree of freedom --- tilting the light cone emitted at each pixel of the display panel. We show that, for scenes with relatively simple occlusion configurations, tilting the light cones provides the same effect as physical occlusion. We demonstrate that ConeTilt can be easily implemented by a phase-only spatial light modulator. Using a lab prototype, we show results that demonstrate the presence of occlusion cues and the increased contrast of the display at depth edges.

Recent successes of deep learning-based recognition rely on maintaining the content related to the main-task label. However, how to explicitly dispel the noisy signals for better generalization remains an open issue. We systematically summarize the detrimental factors as task-relevant/irrelevant semantic variations and unspecified latent variation. In this paper, we cast these problems as an adversarial minimax game in the latent space. Specifically, we propose equipping an end-to-end conditional adversarial network with the ability to decompose an input sample into three complementary parts. The discriminative representation inherits the desired invariance property guided by prior knowledge of the task, which is marginally independent to the task-relevant/irrelevant semantic and latent variations. Our proposed framework achieves top performance on a serial of tasks, including digits recognition, lighting, makeup, disguise-tolerant face recognition, and facial attributes recognition.

We consider the problem of comparing the similarity of image sets with variable-quantity, quality and un-ordered heterogeneous images. We use feature restructuring to exploit the correlations of both inner&inter-set images. Specifically, the residual self-attention can effectively restructure the features using the other features within a set to emphasize the discriminative images and eliminate the redundancy. Then, a sparse/collaborative learning-based dependency-guided representation scheme reconstructs the probe features conditional to the gallery features in order to adaptively align the two sets. This enables our framework to be compatible with both verification and open-set identification. We show that the parametric self-attention network and non-parametric dictionary learning can be trained end-to-end by a unified alternative optimization scheme, and that the full framework is permutation-invariant. In the numerical experiments we conducted, our method achieves top performance on competitive image set/video-based face recognition and person re-identification benchmarks.

This paper targets the task with discrete and periodic class labels (e.g., pose/orientation estimation) in the context of deep learning. The commonly used cross-entropy or regression loss is not well matched to this problem as they ignore the periodic nature of the labels and the class similarity, or assume labels are continuous value. We propose to incorporate inter-class correlations in a Wasserstein training framework by pre-defining (i.e., using arc length of a circle) or adaptively learning the ground metric. We extend the ground metric as a linear, convex or concave increasing function w.r.t. arc length from an optimization perspective. We also propose to construct the conservative target labels which model the inlier and outlier noises using a wrapped unimodal-uniform mixture distribution. Unlike the one-hot setting, the conservative label makes the computation of Wasserstein distance more challenging. We systematically conclude the practical closed-form solution of Wasserstein distance for pose data with either one-hot or conservative target label. We evaluate our method on head, body, vehicle and 3D object pose benchmarks with exhaustive ablation studies. The Wasserstein loss obtaining superior performance over the current methods, especially using convex mapping function for ground metric, conservative label, and closed-form solution.

Recent advances in domain adaptation show that deep self-training presents a powerful means for unsupervised domain adaptation. These methods often involve an iterative process of predicting on target domain and then taking the confident predictions as pseudo-labels for retraining. However, since pseudo-labels can be noisy, self-training can put overconfident label belief on wrong classes, leading to deviated solutions with propagated errors. To address the problem, we propose a confidence regularized self-training (CRST) framework, formulated as regularized self-training. Our method treats pseudo-labels as continuous latent variables jointly optimized via alternating optimization. We propose two types of confidence regularization: label regularization (LR) and model regularization (MR). CRST-LR generates soft pseudo-labels while CRST-MR encourages the smoothness on network output. Extensive experiments on image classification and semantic segmentation show that CRSTs outperform their non-regularized counterpart with state-of-the-art performance. The code and models of this work are available at https://github.com/yzou2/CRST.

Thanks to the increased popularity of Global Position System (GPS) devices (such as smartphones and GPS navigators), the amount of GPS data that can be collected is increasing tremendously. This paper aims to develop novel methods of inferring and updating road topology maps from a large amount of crowd-sourced GPS data. We explore map inference using a three-stage approach, which incorporates a novel Multi-Source Variable Rate (MSVR) signal reconstruction mechanism. Unlike conventional map inference methods based on map graph theory, our approach, to the best of our knowledge, is the first estimation theory method used for map inference. In particular, our approach explicitly leverages the nature of GPS error models, and addresses the unique challenges of vehicular GPS data (asynchronous, varying sampling rate, and under-sampled); as a result, our MSVR approach can better handle inherent GPS errors, reconstruct road shapes more accurately, and better deal with variable GPS data density in empirical environments. The maps inferred from this data are compared to Open Street Map (OSM) maps as ground truth. We evaluate our method using the Tsinghua University's Beijing Taxi Dataset and Shanghai Jiao Tong University's SUVnet Dataset.

In this paper, our goal is to find an optimized approach that can learn features from multichannel EEG time-series data to perform automatic seizure detection. In general, it is not easy to learn robust features from EEG signals due to the variations in both intra and inter-patient variability. However, to achieve good generalization, we use an algorithm that tries to capture spectral, temporal and spatial information, in contrast to standard EEG analysis techniques that ignore spatial aspects. The first stage of this algorithm is to transform EEG signals into a sequence of topology-preserving multi-spectral and temporal images. After that, these generated images are fed as inputs to a convolutional neural network. By overcoming the lack of data, especially the positive samples, and creating a process to deal with unbalanced datasets and optimizing the complexity of the network, our convolutional neural network learns a general spatially invariant representation of a seizure in a reasonable time and improves sensitivity, specificity and accuracy result comparable to the state-of-the-art results.

We present a virtual reality display that is capable of generating a dense collection of depth/focal planes. This is achieved by driving a focus-tunable lens to sweep a range of focal lengths at a high frequency and, subsequently, tracking the focal length precisely at microsecond time resolutions using an optical module. Precise tracking of the focal length, coupled with a high-speed display, enables our lab prototype to generate 1600 focal planes per second. This enables a novel first-of-its-kind virtual reality multifocal display that is capable of resolving the vergence-accommodation conflict endemic to today's displays.

While deep learning methods have achieved state-of-theart performance in many challenging inverse problems like image inpainting and super-resolution, they invariably involve problem-specific training of the networks. Under this approach, each inverse problem requires its own dedicated network. In scenarios where we need to solve a wide variety of problems, e.g., on a mobile camera, it is inefficient and expensive to use these problem-specific networks. On the other hand, traditional methods using analytic signal priors can be used to solve any linear inverse problem; this often comes with a performance that is worse than learning-based methods. In this work, we provide a middle ground between the two kinds of methods - we propose a general framework to train a single deep neural network that solves arbitrary linear inverse problems. We achieve this by training a network that acts as a quasi-projection operator for the set of natural images and show that any linear inverse problem involving natural images can be solved using iterative methods. We empirically show that the proposed framework demonstrates superior performance over traditional methods using wavelet sparsity prior while achieving performance comparable to specially-trained networks on tasks including compressive sensing and pixel-wise inpainting.

As smart vehicles have become more ubiquitous, the capability now exists to detect environmental road features (e.g., potholes, road incline angle, etc.) from their embedded sensor data. By aggregating data from multiple vehicles, crowdsourcing can be leveraged to detect environmental information with improved accuracy. We focus on using such data to detect and localize potholes on multi-lane roads. Extracting information from aggregated vehicle data is challenging due to undersampling sensors, sensor mobility, asynchronous sensor operation, sensor noise, vehicle and road heterogeneity, and GPS position error. GPS position error is particularly problematic in multi-lane environments since the position error is generally larger than standard lane widths. In this paper, we investigate these issues and develop a crowdsourced system to detect and localize potholes in multi-lane environments using accelerometer data from embedded vehicle sensors. Our crowdsourced system reduces the required network bandwidth by determining road incline and bank angle information in each vehicle to filter acceleration components that do not correspond to pothole conditions. We evaluate our system on simulated and real-world data, analyze tradeoffs in the number of vehicles and the amount of bandwidth required for accurate detection, and compare the results to the simpler single lane detection scenario.

We consider the problem of approximating order-constrained transitive distance (OCTD) and its clustering applications. Given any pairwise data, transitive distance (TD) is defined as the smallest possible "gap" on the set of paths connecting them. While such metric definition renders significant capability of addressing elongated clusters, it is sometimes also an over-simplified representation which loses necessary regularization on cluster structure and overfits to short links easily. As a result, conventional TD often suffers from degraded performance given clusters with "thick" structures. Our key intuition is that the maximum (path) order, which is the maximum number of nodes on a path, controls the level of flexibility. Reducing this order benefits the clustering performance by finding a trade-off between flexibility and regularization on cluster structure. Unlike TD, finding OCTD becomes an intractable problem even though the number of connecting paths is reduced. We therefore propose a fast approximation framework, using random samplings to generate multiple diversified TD matrices and a pooling to output the final approximated OCTD matrix. Comprehensive experiments on toy, image and speech datasets show the excellent performance of OCTD, surpassing TD with significant gains and giving state-of-the-art performance on several datasets.

Random features is an approach for kernel-based inference on large datasets. In this paper, we derive performance guarantees for random features on signals, like images, that enjoy sparse representations and show that the number of random features required to achieve a desired approximation of the kernel similarity matrix can be significantly smaller for sparse signals. Based on this, we propose a scheme termed compressive random features that first obtains low-dimensional projections of a dataset and, subsequently, derives random features on the low-dimensional projections. This scheme provides significant improvements in signal dimensionality, computational time, and storage costs over traditional random features while enjoying similar theoretical guarantees for achieving inference performance. We support our claims by providing empirical results across many datasets.

Face recognition is widely used for a variety of applications, such as identifying people for security purposes, as well as photo album organization. A challenge is to perform accurate face recognition when there exist partial occlusions of the face such as scarves or sunglasses. Correlation Filters (CFs) are an occlusion-tolerant object recognition method, potentially suited to deal with partial occlusions. In this paper, we introduce a new class of correlation filters called Masked Correlation Filters (MCFs), that are designed specifically to handle partial occlusions in face images. The benefits of using MCFs are illustrated using well-known face image data sets.

Visual tracking in unconstrained environments often involves following an object that exhibits a number of appearance changes from factors such as scale change, rotation, and illumination. Effective tracking requires adapting a tracker to the object's changing appearance over time. When a target becomes occluded by other objects in the scene, a naive tracker may end up learning the appearance of the occluding object. Our work introduces a method of detecting occlusion by considering the color profile of the target to prevent inappropriate tracker updates while the target is occluded. We show improved overlap and central location precision with three visual trackers when adding our hue-based occlusion detection to each tracking system.

Correlation filters (CFs) are a well-known pattern classification approach used in biometrics. A CF is a spatial-frequency array that is specifically synthesized from a set of training patterns to produce a sharp correlation output peak at the location of the best match for an authentic image comparison and no such peak for an impostor image comparison. The underlying premise when using CFs is that this correlation output peak behavior on training data ideally extends to testing data. Yet in 1:1 verification scenarios, where there is limited training data available to represent pattern distortions, the correlation output from an authentic comparison can be difficult to discern from the correlation output from an impostor. In this paper we introduce Stacked Correlation Filters (SCFs), a simple and powerful approach to address this problem by training an additional set of classifiers which learn to differentiate correlation outputs from authentic and impostor match pairs. This is done by training a series of stacked modular CFs with each layer refining the output of the previous layer. Our basic premise is that since correlation outputs have an expected shape, an additional CF can be trained to recognize such shape and refine the final output. As previous works with CFs have only focused on individual filter design or application, which assumes the CF to provide a sharp peak, this is a new CF paradigm that can benefit many existing CF designs and applications.

We propose locally rewritable codes (LWC) for resistive memories inspired by locally repairable codes (LRC) for distributed storage systems. Small values of repair locality of LRC enable fast repair of a single failed node since the lost data in the failed node can be recovered by accessing only a small fraction of other nodes. By using rewriting locality, LWC can improve endurance and power consumption which are major challenges for resistive memories. We point out the duality between LRC and LWC, which indicates that existing construction methods of LRC can be applied to construct LWC.

Crowdsourcing data from multiple sensing agents has become a fundamental mechanism for extracting information in large-scale cyber-physical systems (CPS). However, there has been little attention paid to the difficulties crowdsourcing presents when the desired information is represented by continuous signals. Crowdsourcing can result in unique types of noise in the aggregated data sample set. Errors from clock synchronization, GPS location determination, or sensor heterogeneity result in biased and correlated errors in the sampled data from each individual sensor. Furthermore, in CPS sensors are often mobile and operate asynchronously. This affects the spatial sampling rates among sensors and results in nonuniform sample spacing. We refer to these sampling issues as the noisy multi-source, variable-rate (MSVR) sampling problem. This work introduces and investigates signal reconstruction algorithms given MSVR sampling conditions. These signal reconstructions are investigated for vehicular applications, using a joint road inclination and bank angle signal estimation algorithm that has access to MSVR sampled vehicle GPS and accelerometer data. The signal reconstruction and angle determination algorithms are validated on simulated and real-world vehicle data, where we reconstruct a road elevation signal with 0.89 m root-mean-square error.

Resistive memories, such as phase change memories and resistive random access memories, have attracted significant research interest because of their scalability, non-volatility, fast speed, and rewritability. However, their write endurance needs to be improved substantially for large-scale deployment of resistive memories. In addition, their write power consumption is much higher than the power consumption of read operation. Inspired by locally repairable codes (LRCs) recently introduced for distributed storage systems, we propose locally rewritable codes (LWCs) for resistive memories. We define a novel parameter ofrewriting locality, which can be connected torepair localityof LRC. As small values of repair locality of LRC enable fast repair in distributed storage systems, small values of rewriting locality of LWC are able to reduce the problems of write endurance and write power consumption. We show how a small value of rewriting locality can improve write endurance and power consumption by deriving the upper bounds on writing cost. Also, we point out the dual relation of LRC and LWC, which indicates that the existing construction methods of LRC can be applied to construct LWC. Finally, we investigate the construction of LWC with error correcting capability for random errors.

Shingled bit-patterned media (BPM) recording is a promising candidate to improve the magnetic recording density above 1 Tb/in2while keeping the conventional writer design. However, shingled magnetic recording (SMR) leads to a narrower track pitch, and thus increased inter-track interference motivating the investigation of array readers and 2-D sectors. In this paper, we investigate SMR on BPM with 2-D sectors at the channel density of 3.71 Tb/in2. Use of 2-D sectors allows the possibility of using either hard or soft information from already detected tracks in that 2-D sector to improve the detection performance on the current track. An important question in this context is whether all tracks in a 2-D sector should be of the same nominal width or if there is a potential benefit to using different widths for adjacent tracks. Based on our investigation, we show that alternating track widths [i.e., one less-trimmed (fat) track followed by one more-trimmed (narrow) track] are beneficial to a bit error rate (BER) perspective. In particular, we propose and investigate the use of a bidirectional decision feedback modified Viterbi (BD-DFMV) detector to improve the BER performance. Our simulations indicate that for a hypothetical 2-D sector with 13 tracks and 5% media noise and at the target BER of 10-2, using alternating track widths and readback with a single reader, the BD-DFMV algorithm can provide about 1.6 dB signal-to-noise ratio (SNR) gain, compared with using the same detection algorithm but on a 2-D sector with uniform track widths. When using a reader array instead of a single reader, this SNR gain improves to about 2.0 dB. The alternating track width structure also appears to provide more robustness to track mis-registration compared with the 2-D uniform track width case.

As a promising bio-cryptographic technique, the fuzzy extractor seamlessly binds biometrics and cryptography for template protection and key generation. However, most existing methods hardly solve the following issues simultaneously: (1) Fingerprint registration, (2) Verification accuracy, (3) Security strength, and (4) Computational efficiency. In this article, we introduce a bio-crypto-oriented fingerprint verification scheme - Selective Vertex-indexed Triangulation (SViT) which maps minutia global topology to local triangulation with minimum information loss. Then, a SViT-based fuzzy extractor framework (FE-SViT) is proposed and high verification accuracy is achieved. The FE-SViT is highly parallelizable and efficient which makes it suitable for embedded devices.

In this paper, a novel subject-adaptable heartbeat classification model is presented, in order to address the significant interperson variations in ECG signals. A multiview learning approach is proposed to automate subject adaptation using a small amount of unlabeled personal data, without requiring manual labeling. The designed subject-customized models consist of two models, namely, general classification model and specific classification model. The general model is trained using similar subjects out of a population dataset, where a pattern matching based algorithm is developed to select the subjects that are "similar" to the particular test subject for model training. In contrast, the specific model is trained mainly on a small amount of high-confidence personal dataset, resulting from multiview-based learning. The learned general model represents the population knowledge, providing an interperson perspective for classification, while the specific model corresponds to the specific knowledge of the subject, offering an intraperson perspective for classification. The two models supplement each other and are combined to achieve improved personalized ECG analysis. The proposed methods have been validated on the MIT-BIH Arrhythmia Database, yielding an average classification accuracy of 99.4% for ventricular ectopic beat class and 98.3% for supraventricular ectopic beat class, which corresponds to a significant improvement over other published results.

In a long-haul sensor network, sensors are remotely deployed over a large geographical area to perform certain tasks, such as target tracking. In this paper, we study the scenario where sensors take measurements of one or more dynamic targets and send state estimates of the targets to a fusion center via satellite links. The severe loss and delay inherent over the satellite channels reduce the number of estimates successfully arriving at the fusion center, thereby limiting the potential fusion gain and resulting in suboptimal accuracy performance of the fused estimates. In addition, the errors in target-sensor data association can also degrade the estimation performance. To mitigate the effect of imperfect communications on state estimation and fusion, we consider retransmission and retrodiction. The system adopts certain retransmission-based transport protocols so that lost messages can be recovered over time. Moreover, retrodiction/smoothing techniques are applied so that the chances of incurring excess delay due to retransmission are greatly reduced. We analyze the extent to which retransmission and retrodiction can improve the performance of delay-sensitive target tracking tasks under variable communication loss and delay conditions. Simulation results of a ballistic target tracking application are shown in the end to demonstrate the validity of our analysis.

Recently introduced 3D vertical flash memory is expected to be a disruptive technology since it overcomes scaling challenges of conventional 2D planar flash memory by stacking up cells in the vertical direction. However, 3D vertical flash memory suffers from a new problem known as fast detrapping, which is a rapid charge loss problem. In this paper, we propose a scheme to compensate the effect of fast detrapping by intentional inter-cell interference (ICI). In order to properly control the intentional ICI, our scheme relies on a coding technique that incorporates the side information of fast detrapping during the encoding stage. This technique is closely connected to the well-known problem of coding in a memory with defective cells. Numerical results show that the proposed scheme can effectively address the problem of fast detrapping.

The use of array readers (i.e., multiple read elements positioned to read stored data on multiple adjacent tracks) is one of the recent concepts to improve the areal density of magnetic recording systems. Array readers provide diversity gain to deal with noise and enable the handling of inter-track interference (ITI) as well as multi-track detection. In this paper, an array-reader two-track (AR2T) detection system is studied. We present the design of 2-D equalizer that jointly processes array readback signal streams over two tracks. Such a 2-D equalizer can be designed to approximate a 1-D or a 2-D partial-response (PR) target. It is shown that 2-D PR target can result in less residual ITI. For the detection of signals corresponding to 2-D PR target, we propose a novel symbol-based detection algorithm that processes two input signal streams jointly. We also implement a pattern-dependent noise-predictive (PDNP) version of proposed detector. Simulation results show that more than 4 dB performance gain can be achieved by the AR2T detection system compared to the traditional single-reader single-track detection system under aggressive track density conditions.

The increased availability of embedded vehicle sensors allows for the detection of road features such as potholes. Despite being a promising approach, current vehicle embedded sensors operate at low frequencies and undersample sensor signals, thus degrading detection accuracy. One emerging solution is to crowdsource such undersampled sensor data from multiple vehicles to increase the detection accuracy. Aggregating sensor data from multiple vehicles, nonetheless, is a challenging task given the heterogeneity among vehicles, asynchronous sensor operation, GPS error, and sensor noise. Additionally, there may be bandwidth restrictions in vehicular networks which limit the amount of data available for aggregation. We investigate these issues by focusing on the problem of pothole detection. To quantify the detection accuracies and effects of real-world limitations, we design and evaluate three crowdsourcing pothole detection schemes involving vehicles and the Cloud. We also address the issue of lack of extensive model training data by demonstrating that a detection model applicable to real-world systems can be derived using simulated data. We validate our pothole detection methods using 38.1 km of real-world data collected from driving on roads in Warren, Michigan.

We propose a vision-based highway border detection algorithm using structured Hough voting. Our approach takes advantage of the geometric relationship between highway road borders and highway lane markings. It uses a strategy where a number of trained road border and lane marking detectors are triggered, followed by Hough voting to generate corresponding detection of the border and lane marking. Since the initially triggered detectors usually result in large number of positives, conventional frame-wise Hough voting is not able to always generate robust border and lane marking results. Therefore, we formulate this problem as a joint detection-and-tracking problem under the structured Hough voting model, where tracking refers to exploiting inter-frame structural information to stabilize the detection results. Both qualitative and quantitative evaluations show the superiority of the proposed structured Hough voting model over a number of baseline methods.

Correlation filters (CFs) are a class of classifiers that are attractive for object localization and tracking applications. Traditionally, CFs have been designed in the frequency domain using the discrete Fourier transform (DFT), where correlation is efficiently implemented. However, existing CF designs do not account for the fact that the multiplication of two DFTs in the frequency domain corresponds to a circular correlation in the time/spatial domain. Because this was previously unaccounted for, prior CF designs are not truly optimal, as their optimization criteria do not accurately quantify their optimization intention. In this paper, we introduce new zero-aliasing constraints that completely eliminate this aliasing problem by ensuring that the optimization criterion for a given CF corresponds to a linear correlation rather than a circular correlation. This means that previous CF designs can be significantly improved by this reformulation. We demonstrate the benefits of this new CF design approach with several important CFs. We present experimental results on diverse data sets and present solutions to the computational challenges associated with computing these CFs. Code for the CFs described in this paper and their respective zero-aliasing versions is available at http://vishnu.boddeti.net/projects/correlation-filters.html.

The periocular region as a biometric trait has recently gained considerable traction, especially under challenging scenarios where reliable iris information is not available for human authentication. In this paper, we consider the problem of one-to-one (1 : 1) matching of highly nonideal periocular images captured in-the-wild under unconstrained imaging conditions. Such images exhibit considerable appearance variations, including nonuniform illumination variations, motion and defocus blur, off-axis gaze, and nonstationary pattern deformations. To address these challenges, we propose periocular probabilistic deformation models (PPDMs) that: 1) reduce the image matching problem to matching local image regions and 2) approximate the periocular distortions by local patch level spatial translations whose relationships are modeled by a Gaussian Markov random field. Given a periocular image pair, we determine the distortion-tolerant similarity metric by regularizing local match scores by the maximum aposteriori probability estimate of the relative local deformations between them. Unlike the existing global periocular image matching techniques, by accounting for local image deformations in the periocular matching process, PPDM exhibits greater tolerance to pattern variations. We demonstrate the effectiveness of our model via extensive evaluation on a large number of in-the-wild periocular images. We find that PPDMs outperform many benchmark 1 : 1 image matching techniques (improving verification rates at 0.1% false accept rate by ~30% over previous work and ~40% when compared with the best baseline) in challenging scenarios leading to state-of-the-art verification performance on multiple real-world periocular data sets.

IEEE 802.11p-based Dedicated Short Range Communications (DSRC) is considered a promising wireless technology for enhancing transportation safety and improving highway efficiency. One major challenge of IEEE 802.11p technology development ensuring its communication reliability in highly dynamic Vehicle-to-Vehicle (V2V) environments. In this paper, by investigating the characteristics of V2V channels through empirical measurements, we show that the decreased reliability of V2V communication is because the IEEE 802.11p design does not have sufficient number of training symbols in time domain and pilot subcarriers in frequency domain to enable estimation of the fast-changing V2V channels accurately. To tackle this challenge, we propose a new approach called Tentpoles scheme, which utilizes a small subset of data symbols and subcarriers protected by a strong Error Correction Code (ECC) to track and estimate V2V channel variations. Since this subset of data symbols and subcarriers provide dual-use, our Tentpoles scheme not only improves the accuracy of channel estimation and communication reliability, but has only a small impact on effective throughput. Through extensive simulations in synthetic V2V channels and emulation experiments using empirical measurements, our proposed Tentpoles scheme is shown to significantly outperform other V2V channel estimation and equalization methods in terms of communication reliability, at the cost of only moderate loss of throughput.

We address the problem of detecting rear-view (obstacle free) ground surface using a vehicle production camera. This task is considerably more challenging than general front-view road detection, as the associated challenges widely range from low picture quality, fisheye distortion and large objects, to the absence of useful priors such as vanishing points and road structure. Regarding the challenges, we propose a feature that can simultaneously capture local appearance and context information. In addition, the task suffers from strong appearance variations such as shadows and ground markers. Therefore, we propose a novel conditional random field (CRF) model which includes hidden states indicating confident nodes and propagate their confidence to neighboring nodes. We show that our proposed feature and model can jointly achieve robustness against large objects and shadows/markers, showing excellent detection performance under low quality inputs.

A self-driving car, to be deployed in real-world driving environments, must be capable of reliably detecting and effectively tracking of nearby moving objects. This paper presents our new, moving object detection and tracking system that extends and improves our earlier system used for the 2007 DARPA Urban Challenge. We revised our earlier motion and observation models for active sensors (i.e., radars and LIDARs) and introduced a vision sensor. In the new system, the vision module detects pedestrians, bicyclists, and vehicles to generate corresponding vision targets. Our system utilizes this visual recognition information to improve a tracking model selection, data association, and movement classification of our earlier system. Through the test using the data log of actual driving, we demonstrate the improvement and performance gain of our new tracking system.

Alignment of 3D objects from 2D images is one of the most important and well studied problems in computer vision. A typical object alignment system consists of a landmark appearance model which is used to obtain an initial shape and a shape model which refines this initial shape by correcting the initialization errors. Since errors in landmark initialization from the appearance model propagate through the shape model, it is critical to have a robust landmark appearance model. While there has been much progress in designing sophisticated and robust shape models, there has been relatively less progress in designing robust landmark detection models. In this paper we present an efficient and robust landmark detection model which is designed specifically to minimize localization errors thereby leading to state-of-the-art object alignment performance. We demonstrate the efficacy and speed of the proposed approach on the challenging task of multi-view car alignment.

In this paper, we propose an encoding scheme for partitioned linear block codes (PLBC) which mask the stuck-at defects in memories. In addition, we derive an upper bound and the estimate of the probability that masking fails. Numerical results show that PLBC can efficiently mask the defects with the proposed encoding scheme. Also, we show that our upper bound is very tight by using numerical results.

When two signals or images to be correlated or convolved are of different sizes, overlap-add (OLA) and overlap-save (OLS) algorithms are often employed to reduce computational complexity. In this paper, we investigate the use of these algorithms for multidimensional correlations required in many image and video matching applications. In contrast to conventional wisdom, we show that OLA is more computationally efficient and OLS is more memory efficient. We present theoretical expressions for computational complexity for an arbitrary dimensionality and validate these expressions experimentally. We also discuss parameter selection from both memory and computational perspectives.

Most memories suffer from both permanent defects and intermittent random errors. The partitioned linear block codes (PLBC) were proposed by Heegard to efficiently mask stuck-at defects and correct random errors. The PLBC have two separate redundancy parts for defects and random errors. In this paper, we investigate the allocation of redundancy between these two parts. The optimal redundancy allocation will be investigated using simulations and the simulation results show that the PLBC can significantly reduce the probability of decoding failure in memory with defects. In addition, we will derive the upper bound on the probability of decoding failure of PLBC and estimate the optimal redundancy allocation using this upper bound. The estimated redundancy allocation matches the optimal redundancy allocation well.

This paper deals with the detection of potential collision for a reversing vehicle. The collision detection relies on monitoring the surroundings of reversing vehicle by means of on-board cameras mounted on the side window and rear side of the vehicle. We first detect the stationary vehicle(s) in the region of interest (ROI) using histogram of oriented gradients (HOG) features and support vector machine (SVM) based classification. This is followed by the use of a Gaussian Mixture Model (GMM) to detect dynamic vehicle(s) in the ROI. Trajectories and speeds of the dynamic vehicles are predicted through the particular position of the detected vehicles in each frame and through their observed trajectory and speed patterns. Finally a potential collision is predicted based on the trajectories and speeds of both vehicles. The experimental results indicate that the proposed framework performs well in different traffic scenarios and can contribute to the advancement for improving the safety of car driving.

In this paper, we investigate to achieve low Packet Error Rate (PER) for Vehicle-to-Vehicle (V2V) communications within the IEEE 802.11p standard. We first present the best achievable PER with the assumption of perfect channel knowledge. Then we propose to approach the best achievable PER with a relaxed assumption of partial channel knowledge, i.e., knowing a few separated channels in a packet. In order to obtain partial channel knowledge in an 802.11p standard-compliant way, we propose a sub-packet decoding processing in which the successful decoding of the previous sub-packet will provide partial channel knowledge for decoding the following sub-packet. Then we design a double decoding scheme to achieve low PER in a standard-compliant way. The double decoding scheme utilizes the error detection information provided by CRC check in medium-access (MAC) layer and combines two different decoding methods in physical (PHY) layer. The results show that comparing to previous work, the proposed double decoding can improve PER significantly.

We describe a template-based framework to bind class-specific information to a set of image patterns and retrieve that information by matching the template to a query pattern of the same class. This is done by mapping the class-specific information to a set of spatial translations which are applied to the set of image patterns from which a template is designed, taking advantage of the properties of correlation filters. The bound information is retrieved during matching with an authentic query by estimating the spatial translations applied to the images that were used to design the template. In this paper, we focus on the problem of binding information to biometric signatures as an application of our framework. Our framework is flexible enough to allow spreading the information to be bound over multiple pattern classes which, in the context of biometric key-binding, enables multiclass and multimodal biometric key-binding. We demonstrate the effectiveness of the proposed scheme via extensive numerical results on multiple biometric databases.

Support vector machine (SVM) classifiers are popular in many computer vision tasks. In most of them, the SVM classifier assumes that the object to be classified is centered in the query image, which might not always be valid, e.g., when locating and classifying a particular class of vehicles in a large scene. In this paper, we introduce a new classifier called Maximum Margin Correlation Filter (MMCF), which, while exhibiting the good generalization capabilities of SVM classifiers, is also capable of localizing objects of interest, thereby avoiding the need for image centering as is usually required in SVM classifiers. In other words, MMCF can simultaneously localize and classify objects of interest. We test the efficacy of the proposed classifier on three different tasks: vehicle recognition, eye localization, and face classification. We demonstrate that MMCF outperforms SVM classifiers as well as well known correlation filters.

The cardiosynchronous signal obtained through Radio Frequency Impedance Interrogation (RFII) is a non-invasive method for monitoring hemodynamics with potential applications in combat triage and biometric identification. The RFII signal is periodic in nature dominated by the heart beat cycle. The first step in both of these applications is to segment the signal by identifying a fiducial point in each heart beat cycle. A continuous wavelet transform was utilized to locate the fiducial points with high temporal resolution. Cepstral Analysis was used to estimate the average heart rate to focus on the appropriate portion of the time-frequency spectrum. Robust heartbeats from RFII signals collected from four subjects were segmented using this method.

Radio Frequency Impedance Interrogation (RFII) measures hemodynamic function via resonance frequency coupling to a hydrophilic protein molecule. The RFII device generates a cardiosynchronous waveform from the identification of blood movement in the time, frequency, and voltage domains. This paper examines RFII signals with the end goal of allowing confirmation of the identity of a subject in an operational setting. An Optimal Trade-off Synthetic Discriminant Function (OT-SDF) was applied to filter the data stream for subject identification. Preliminary results using the OT-SDF Filters demonstrate 63.3% successful single-heartbeat subject identification. However, each individual's correlation surfaces appear to have a unique waveform morphology that is visually distinct from the other individuals in the data set. Improved identification was seen with second-degree correlation suggesting that a second-degree correlation may hold great potential as a biometric feature extraction identifier. We show that using correlation plane outputs as features actually provide a robust biometric identifier and significant higher identification accuracy.

Raindrops on vehicles' windshields can degrade the performance of in-vehicle vision systems. In this paper, we present a novel approach that detects and removes raindrops in the captured image when using a single in-vehicle camera. When driving in light or moderate rainy conditions, raindrops appear as small circlets on the windshield in each image frame. Therefore, by analyzing the color, texture and shape characteristics of raindrops in images, we first identify possible raindrop candidates in the regions of interest (ROI), which are small locally salient droplets in a raindrop saliency map. Then, a learning-based verification algorithm is proposed to reduce the number of false alarms (i.e., clear regions mis-detected as raindrops). Finally, we fill in the regions occupied by the raindrops using image inpainting techniques. Numerical experiments indicate that the proposed method is capable of detecting and reducing raindrops in various rain and road scenarios. We also quantify the improvement offered by the proposed method over the state-of-the-art algorithms aimed at the same problem and the benefits to the in-vehicle vision applications like clear path detection.

Detecting objects in shadows is a challenging task in computer vision. For example, in clear path detection application, strong shadows on the road confound the detection of the boundary between clear path and obstacles, making clear path detection algorithms less robust. Shadow removal, relies on the classification of edges as shadow edges or non-shadow edges. We present an algorithm to detect strong shadow edges, which enables us to remove shadows. By analyzing the patch-based characteristics of shadow edges and non-shadow edges (e.g., object edges), the proposed detector can discriminate strong shadow edges from other edges in images by learning the distinguishing characteristics. In addition, spatial smoothing is used to further improve shadow edge detection. Numerical experiments show convincing results that shadows on the road are either removed or attenuated with few visual artifacts, which benefits the clear path detection. In addition, we show that the proposed method outperforms the state-of-art algorithms in different conditions.

In a long-haul sensor network, sensors are remotely deployed over a large geographical area to perform certain tasks. We consider a class of such networks where sensors take measurements of one or more dynamic targets and send state estimates of the target(s) to a fusion center via satellite links. The severe loss and delay inherent over the satellite channels render insufficient the number of estimates successfully arriving at the fusion center, thereby limiting the potential fusion gain and resulting in suboptimal accuracy performance of the fused estimates. The system can adopt certain retransmission-based transport protocols so that lost messages can be recovered over time. However, excess delay may be incurred that can potentially violate the deadline for reporting the estimate. For many applications, though, retrodiction/smoothing techniques can be applied so that the chances of incurring such excess delay are greatly reduced. In this work, we analyze the extent to which retrodiction, along with message retransmission, can improve the performance of delay-sensitive state estimation tasks. Results of numerical and simulation studies of an illustrative example and a ballistic target tracking application are shown in the end to demonstrate the validity of our analysis.

In this paper, a novel iterative encoding scheme is proposed for memory systems suffering from stuck-at errors. The stuck-at errors can be efficiently managed by using side information about stuck-at memory cells during encoding, while encoding for unconstrained number of stuck-at errors is intractable due to its exponential complexity. The proposed coding scheme employs an iterative encoding algorithm using cross-entropy method, which has a polynomial time complexity. In addition, any linear block code (LBC) can be concatenated with the proposed code, to correct for both residual stuck-at errors and random (soft) errors. The proposed coding schemes are evaluated by numerical simulations using a memory channel undergoing both stuck-at and random errors. Simulation results show that the cross-entropy based coding scheme provides an improved block error rate (BLER) performance, or alternatively, a higher overall storage capacity.

Iris recognition is believed to offer excellent recognition rates for iris images acquired under controlled conditions. However, recognition rates degrade considerably when images exhibit impairments such as off-axis gaze, partial occlusions, specular reflections and out-of-focus and motion-induced blur. In this paper, we use the recently-available face and ocular challenge set (FOCS) to investigate the comparative recognition performance gains of using ocular images (i.e., iris regions as well as the surrounding peri-ocular regions) instead of just the iris regions. A new method for ocular recognition is presented and it is shown that use of ocular regions leads to better recognition rates than iris recognition on FOCS dataset. Another advantage of using ocular images for recognition is that it avoids the need for segmenting the iris images from their surrounding regions.

With the goal of avoiding obstacles on the road using only a single camera during autonomous driving, we propose an example-based clear path detection method that also considers the additional perspective cue from the estimated vanishing point. First, because of the benefit that the vanishing point provides knowledge for clear path detection, we apply an example-based method to estimate initial vanishing point candidates. Then, instead of building a pre-trained clear path model over the limited training set, we propose an example-based global image matching method to get an approximate idea of clear path candidate regions, and use a Gaussian Mixture Model (GMM) for local image patch modeling to further improve the clear path detection. Finally, we develop an iterative probabilistic refinement to improve performance of both components. Experimental results of real road scenes are presented to illustrate the effectiveness of the proposed method.

This paper provides an editorial statement for the special issue on biometric security in mobile computing environment.

In space, radiation particles can introduce temporary or permanent errors in memory systems. To protect against potential memory faults, either thick shielding or error correcting codes (ECC) are used. Thick shielding translates into increased mass and conventional ECCs designed for memories are typically capable of only correcting a single error and detecting a double error. Decoding is usually performed through hard-decisions where bits are treated as either correct or flipped in polarity. In this work, we demonstrate that low-density parity-check (LDPC) codes that are already prevalent in many communication applications can also be used to protect memories in space. We develop a channel that models memory error events in a space radiation environment. We describe how to compute soft symbol reliabilities on our channel and compare the performance of soft-decision decoding LDPC codes against conventional hard-decision decoding of Reed-Solomon (RS) codes and Bose-Chaudhuri-Hoquenghem (BCH) codes for a specific memory structure.

In using image analysis to assist a driver to avoid obstacles on the road, traditional approaches rely on various detectors designed to detect different types of objects. We propose a framework that is different from traditional approaches in that it focuses on finding a clear path ahead. We assume that the video camera is calibrated offline (with known intrinsic and extrinsic parameters) and vehicle information (vehicle speed and yaw angle) is known. We first generate perspective patches for feature extraction in the image. Then, after extracting and selecting features of each patch, we estimate an initial probability that the patch corresponds to clear path using a support vector machine (SVM) based probability estimator on the selected features. We finally perform probabilistic patch smoothing based on spatial and temporal constraints to improve the initial estimate, thereby enhancing detection performance. We show that the proposed framework performs well even in some challenging examples with shadows and illumination changes.

In this paper, we describe a prior-based vanishing point estimation method through global perspective structure matching (GPSM). In contrast to the traditional approaches which require an undistorted image with straight roads for vanishing point estimation, our method first infers vanishing point candidates of an input image from an image database with pre-labeled vanishing points. An image-based retrieval method is used to identify the best candidate images in the database by matching image's perspective structure. The initial estimation of input image's vanishing point is calculated from the pre-labeled vanishing points of the best candidates. Probabilistic refinement (PR) is then used to optimize the vanishing point estimate. Experimental results show that the proposed method works well in a variety of on-road driving environments (e.g., in urban, highway and country-side areas), especially with traffic captured by a fish-eye back-aid camera.

Two-dimensional magnetic recording (TDMR) is a data storage paradigm that has recently been proposed to record more than one channel bit to one magnetic grain in order to achieve 10 Tbit/in² of user bit density on a magnetic recording medium with 20 Tgrains/in². In addition to conventional noise in magnetic recording, the TDMR channel also suffers from write errors due to the randomness in size and placement of the grains; however aggressive two-dimensional (2D) signal processing schemes may enable reliable data retrieval. In this study, the symmetric information rate (SIR) of a TDMR channel model based on random Voronoi grains, named the ideal readback TDMR channel model, is analyzed. Using numerical simulations, a lower bound of the SIR is evaluated for a half grain per channel bit ideal readback TDMR channel. Although approximately half of the channel bits may not be recorded in this channel, the resulting lower bound of the SIR is 0.46 ~ 0.51 information bits per grain. This result indicates the potential of TDMR to achieve 10 Tbit/in² of user bit density on a magnetic medium with 20 Tgrains/in².

Correlation filters (CFs) can detect multiple targets in one scene making them well-suited for automatic target recognition (ATR) applications. We present a method to efficiently compute the Quadratic CF (QCF) capable of detecting multiple targets from two classes. We use a Kalman filter framework to combine information from successive correlation outputs in a probabilistic way integrating the ATR tasks of detection, recognition, and tracking.

Towards the goal of achieving better error correction performance in data storage systems, iterative soft decoding of low density parity check (LDPC) codes and soft-decision decoding of Reed-Solomon (RS) codes have started receiving increasing research attention. However, even with increased computing power, complexities of soft-decision decoding algorithms are still too high for real products which require high throughput and small hardware area. Another problem is that the performance gains of those approaches are smaller for magnetic recording channels than they are for memoryless additive white Gaussian noise (AWGN) channels. We propose a new soft-decision decoding algorithm (based on the Chase algorithm), which takes advantage of pattern reliability instead of symbol reliability or bit reliability. We also present a modified Viterbi algorithm that provides probable error patterns with corresponding reliabilities. Simulation results of the proposed algorithms over the partial response (PR) channel show attractive performance gains. The proposed algorithm dramatically reduces the number of iterations compared to the conventional Chase2 algorithm over the PR channel.

In this work, a new analytical approach is used to evaluate the error performance of bit-patterned media (BPM) magnetic recording channels that employ one-dimensional (1D) and two-dimensional (2D) generalized partial response (GPR) equalizers to combat the significant inter-track interference (ITI) expected in BPM magnetic recording systems. The probability density function of ITI is obtained analytically and is used to estimate the bit error rate (BER) from the Viterbi detector. The proposed method takes into account most of the important factors affecting the BER such as ITI, un-equalized intersymbol interference (ISI), colored noise and the distance and the multiplicity of error events. In this work, it is shown that for 1D channels, modeling ITI and un-equalized ISI by Gaussian PDFs leads to inaccurate BERs and that the non-Gaussian distribution of the ITI and un-equalized ISI must be taken into account for more accurate BER estimates. This method provides fast and accurate estimates of BERs for moderate to high signal-to-noise ratios (SNRs). By using this analytical method, time-consuming numerical simulations for error performance evaluation can be avoided.

Iris recognition can offer high-accuracy person recognition, particularly when the acquired iris image is well focused. However, in some practical scenarios, user cooperation may not be sufficient to acquire iris images in focus; therefore, iris recognition using camera systems with a large depth of field is very desirable. One approach to achieve extended depth of field is to use a wavefront-coding system as proposed by Dowski and Cathey, which uses a phase modulation mask. The conventional approach when using a camera system with such a phase mask is to restore the raw images acquired from the camera before feeding them into the iris recognition module. In this paper, we investigate the feasibility of skipping the image restoration step with minimal degradation in recognition performance while still increasing the depth of field of the whole system compared to an imaging system without a phase mask. By using a simulated wavefront-coded imagery, we present the results of two different iris recognition algorithms, namely, Daugman's iriscode and correlation-filter-based iris recognition, using more than 1000 iris images taken from the iris challenge evaluation database. We carefully study the effect of an off-the-shelf phase mask on iris segmentation and iris matching, and finally, to better enable the use of unrestored wavefront-coded images, we design a custom phase mask by formulating an optimization problem. Our results suggest that, in exchange for some degradation in recognition performance at the best focus, we can increase the depth of field by a factor of about four (over a conventional camera system without a phase mask) by carefully designing the phase masks.

In this work, we elaborate on a rather intuitive hypothesis: face recognition of low-resolution faces can be improved if the processes of reconstruction and recognition are considered simultaneously, instead of sequentially, without feedback or any interaction. Given a high-resolution training set, matching low-resolution probe images with good accuracy is an open problem. We have recently introduced [Hennings-Yeomans, Baker, and Kumar, CVPR, June 2008] a new framework for low-resolution face recognition that uses models from an image formation process, super-resolution priors and face feature extraction methods. By measuring how well an intermediate super-resolution reconstruction of the probe image fits into the models used in the process, the proposed matching algorithm extracts new features for recognition. In this paper, we present results for an improved design of these new features. We show that the proposed algorithm improves performance in both, identification and verification tasks on a large database of 337 subjects that also captures illumination variations.

Face recognition degrades when faces are of very low resolution since many details about the difference between one person and another can only be captured in images of sufficient resolution. In this work, we propose a new procedure for recognition of low-resolution faces, when there is a high-resolution training set available. Most previous super-resolution approaches are aimed at reconstruction, with recognition only as an after-thought. In contrast, in the proposed method, face features, as they would be extracted for a face recognition algorithm (e.g., eigenfaces, Fisher-faces, etc.), are included in a super-resolution method as prior information. This approach simultaneously provides measures of fit of the super-resolution result, from both reconstruction and recognition perspectives. This is different from the conventional paradigms of matching in a low-resolution domain, or, alternatively, applying a super-resolution algorithm to a low-resolution face and then classifying the super-resolution result. We show, for example, that recognition of faces of as low as 6 times 6 pixel size is considerably improved compared to matching using a super-resolution reconstruction followed by classification, and to matching with a low-resolution training set.

Concatenations of inner low-density parity-check (LDPC) codes and outer Reed-Solomon (RS) codes were evaluated in magnetic recording channel models by simulations using field programmable gate arrays (FPGAs). Fixed inner LDPC codes were used and the optimal code rates for the outer RS code were obtained. It is observed that the outer RS codes do not always improve sector error rates for a fixed user bit density. The effect of various parameters on the sector error rates was evaluated and discussed.

We propose a soft-decision decoding algorithm of Reed-Solomon (RS) codes using successive error-and-erasure decoding. Extensive simulations are conducted to show the possible performance gain of the proposed method. We derive a formula for performance estimation based on ordered statistics of symbol reliability, which matches well with the results of the simulation. The proposed method with almost the same average complexity as a conventional hard-decision decoder outperforms Koetter-Vardy (KV) algorithm and Chase2-GMD algorithm (CGA).

In synchronous timing recovery approaches often employed in magnetic recording readback channels, the timing error detector (TED) output is assumed to be linearly related to the underlying phase offset. However, the noisy readback signal samples in high-density magnetic recording systems are nonlinearly related to the underlying timing disturbances. Thus, a nonlinear state space models may describe such systems more accurately and an improved timing recovery solution may be obtained via an extended Kalman filter (EKF) approach. In this paper, EKF based timing recovery is investigated for the acquisition stage. Numerical simulations indicate that the EKF based timing recovery scheme yields smaller root mean square (RMS) error in timing estimates, which enables it to handle larger timing errors more reliably. In addition, the proposed method converges faster than the conventional timing recovery schemes, which may allow us to reduce the length of the preamble pattern used for the acquisition.

In bit-patterned media (BPM) aimed at magnetic recording densities of 1 Tbit/in2and higher, adjacent tracks may become very close leading to significant inter-track interference (ITI). Read-head offset or track mis-registration (TMR) can further degrade the performance of the channel. To investigate the effects of ITI and TMR on the performance of the channel, we have developed a two-dimensional (2D) pulse response simulator for BPM. Simulation results using these pulse responses suggest that equalizers and detectors optimized for zero TMR may not perform well in the presence of TMR. To mitigate the effects of TMR, we propose a modified trellis for the Viterbi algorithm (VA). The modified VA (MVA) takes into account the ITI while computing branch metrics. Simulation results show that the MVA can improve the bit error rate in the presence of TMR.

Template matching of iris images for biometric recognition typically suffers from both local deformations between the template and query images and large occlusions from the eyelid. In this work, we model deformation and occlusion as a set of hidden variables for each iris comparison. We use afield of directional vectors to represent deformation and a field of binary variables to represent occlusion. We impose a probability distribution on these fields using a lattice-type undirected graphical model, in which the graph edges represent interdependencies between neighboring iris regions. Gabor wavelet-based similarity scores and intensity statistics are used as observations in the model. Loopy belief propagation is applied to estimate the conditional distributions on the hidden variables, which are in turn used to compute final match scores. We present underlying theory as well as experimental results from both the CASIA iris database and the database provided for the iris challenge evaluation (ICE). We show that our proposed method significantly improves recognition accuracy on these datasets over existing methods.

The presence of error floor in low density parity check (LDPC) codes is of great concern for potential applications of LDPC codes to data storage channels, which require the error correcting code (ECC) to maintain the near-capacity error correcting performance at frame error rate as low as 10-12. In order to investigate the error floor of LDPC codes under partial response channels used in data storage systems, we propose a new estimation method combining analytical tools and simulation, based on the concept of trapping sets. The definition of trapping sets is based on the dominant error patterns observed in the decoding process. The goal is to accurately estimate the error rate in the error floor region for certain types of LDPC codes under the partial response channel and further extend the frame error rate down to 10-14or lower. Towards this goal, we first use field programmable gate array (FPGA) hardware simulation to find the trapping sets that cause the decoding failure in the error floor region. For each trapping set, we extract the parameters which are key to the decoding failure rate caused by this trapping set. Then we use a much simpler in situ hardware simulation with these parameters to obtain the conditional decoding failure rate. By considering all the trapping sets we find, we obtain the overall frame error rate in the error floor region. The estimation results for a length -4623 QC-LDPC code under the EPR4 channel are within 0.3 dB of the direct simulation results. In addition, this method allows us to estimate the frame error rate of a LDPC code down to 10-14or lower.

An analytical method to identify lists of the most probable error events for arbitrary partial response (PR) targets is described. The dominant error events are extracted via a computer search considering not only the minimum distance but also additive white noise amplification and data-dependent transition jitter noise. The method uses an efficient scheme to take into account the effect of data-dependent transition jitter noise. Combining the error event analysis and a multinomial model, approximate sector error rates can be calculated without the need for time-consuming simulations. The results from the analytical method are compared with simulation results.

In this paper we analyze and demonstrate the subspace generalization power of the kernel correlation feature analysis (KCFA) method for producing compact low dimensional subspace that has good representation ability to work on unseen, untrained datasets. Examining the portability of an algorithm across different datasets is an important practical aspect of face recognition applications where the technology cannot be dataset-dependant in real-world practical applications. In most face recognition literature, algorithms are demonstrated on datasets by training on some part of the dataset and testing on the remainder. In general, the training and testing data have the same people but different capture sessions so essentially, some of the expected variation and people are modeled in the training set. In this paper we describe how we efficiently build a compact feature space using kernel correlation filter analysis on the generic training set of the FRGC dataset, and test the built subspace on other well-known face datasets. We show that the feature subspace produced by KCFA has good representation and discrimination to unseen datasets and produces good verification and identification rates compared to other subspace methods such as PCA. Its efficiency, lower dimensionality (the KCFA is only a 222 dimensional subspace) and discriminative power make it more practical and powerful than PCA as a powerful lower dimensionality reduction method for modeling faces and facial variations.

Reed-Solomon (RS) codes are defined in several different ways which are not exactly equivalent. Distinct algorithms for encoding and decoding of RS codes have been developed based on the different definitions. This paper clarifies the relationships among the several definitions of RS codes, especially focusing on the connection between non-binary BCH codes, which are not necessarily narrow-sense, and Generalized Reed- Solomon (GRS) codes. With the relationships, we introduce code space conversion which converts objects of a GRS code into corresponding objects of another GRS code with the same dimension. This transformation can be used for the flexible application of the decoder designed for a specific RS code. We show that the interpolation-based soft-decision RS decoders can be applied to non narrow-sense RS codes using code space conversion.

The use of bit-patterned media is one of the approaches being investigated to extend magnetic recording densities to 1 Tbit/in2and beyond. In patterned media, track pitch may be small causing adjacent tracks to have significant interference on the replay waveform from the main data track. To mitigate the effect of such inter-track interference (ITI), we propose the use of a two-dimensional (2D) generalized partial response (GPR) equalizer. We select both the equalizer and the partial response target using the minimum mean squared error (MMSE) criterion. However, we avoid the need for a 2D Viterbi algorithm by imposing a constraint on the 2D target that forces the adjacent track contributions (in the ideal case) to zero. Simulation results show that this 2D equalizer significantly improves the bit error rate (BER). In this work, the effect of a 2D GPR equalizer on the performance of a patterned media system in the presence of track misregistration (TMR) is also investigated. Based on the simulation results, the 2D equalization method appears to be more tolerant to TMR than the conventional GPR The main drawback of the proposed method is the need for simultaneously acquiring the signals from three adjacent tracks.

We describe a general probabilistic framework for matching patterns that experience in-plane nonlinear deformations, such as iris patterns. Given a pair of images, we derive a maximum a posteriori probability (MAP) estimate of the parameters of the relative deformation between them. Our estimation process accomplishes two things simultaneously: It normalizes for pattern warping and it returns a distortion-tolerant similarity metric which can be used for matching two nonlinearly deformed image patterns. The prior probability of the deformation parameters is specific to the pattern-type and, therefore, should result in more accurate matching than an arbitrary general distribution. We show that the proposed method is very well suited for handling iris biometrics, applying it to two databases of iris images which contain real instances of warped patterns. We demonstrate a significant improvement in matching accuracy using the proposed deformed Bayesian matching methodology. We also show that the additional computation required to estimate the deformation is relatively inexpensive, making it suitable for real-time applications.

The 12 regular papers and three correspondences in this special issue focus on human detection and recognition. The papers represent gait, face (3-D, 2-D, video), iris, palmprint, cardiac sounds, and vulnerability of biometrics and protection against the spoof attacks.

We propose a palmprint classification algorithm with the use of multiple correlation filters per class. Correlation filters are two-class classifiers that produce a sharp peak when filtering a sample of their class and a noisy output otherwise. For every class, we train the filters for a palm at different locations, where the palmprint region has a high degree of line content. With the use of a line detection procedure and a simple line energy measure, any region of the palm can be scored and the top-ranked regions are used to train the filters for each class. Using an enhanced palmprint segmentation algorithm, our proposed classifier achieves an average equal error rate of 1.12 times10-4% on a large database of 385 classes using multiple filters of size 64 times 64 pixels. The average false acceptance rate when the false rejection rate is zero is 2.25 times10-4%.

In high-density magnetic recording systems, a runlength constraint (typically (0, k)) is imposed to aid then timing recovery system. However, the traditional encoding structure where the runlength limited (RLL) encoder follows the ECC encoder is not a good fit when using LDPC codes. The main difficulty is that the LDPC decoder requires soft information, which is incompatible the typical RLL decoders which output hard decoding decisions. In this paper, we propose a structure so that soft information can be delivered directly to the LDPC decoder while the runlength constraint is satisfied. By integrating the bit-flipping scheme proposed in [Vasic and Pedagani, IEEE Trans. Magn, vol. 40, no. 5, pp. 1738-1743, May 2004] and an interleaver into the system, we are able to take advantage of the near-capacity performance of the LDPC code given the runlength limit. The simulations show that the BER performance loss of the LDPC code (rate 8/9, length 4932) due to the RLL constraint is less than 0.1 dB. The error floor introduced by bit-flipping scheme is limited to bit error rate (BER) levels lower than 10-7.

Recently, Direct Linear Discriminant Analysis (D-LDA) and Gram-Schmidt LDA methods have been proposed for face recognition. By also utilizing some of the null-space of the within-class scatter matrix, they exhibit better performance compared to Fisherfaces and Eigenfaces. However, these linear subspace methods may not discriminate faces well due to large nonlinear distortions in the face images. Redundant class dependence feature analysis (CFA) method exhibits superior performance compared to other methods by representing nonlinear features well. We show that with a proper choice of kernel parameters used with the proposed Kernel Correlation Filters within the CFA framework, the overall face recognition performance is significantly improved. We present results of this proposed approach on a large scale database from the Face Recognition Grand Challenge (FRGC) which contains over 36,000 images.

Current state-of-the-art approach for performing face sketch recognition transforms all the test face images into sketches, and then performs recognition on sketch domain using the sketch composite. In our approach we propose the opposite; which has advantages in a real-time sysrtem; we propose to generate a realistic face image from the composite sketch using a Hybrid subspace method and then build an illumination tolerant correlation filter which can recognize the person under different illumination variations from a surveillance video footage. We show how effective proposed algorithm works on the CMU PIE (Pose Illumination and Expression) database.

Face recognition is one of the least intrusive biometric modalities that can be used to identify individuals from surveillance video. In such scenarios the users are under the least co-operative conditions and thus the ability to perform robust face recognition in such scenarios is very challenging. In this paper we focus on improving the face recognition performance on a large database with over 36,000 facial images from the Face Recognition Grand Challenge Phase-II data collected by University of Notre Dame. We particularly focus on Experiment 4 which is the most challenging and captured in uncontrolled conditions where the baseline PCA algorithm yields 12% verification rate at 0.1% FAR. We propose a novel approach using class-dependent kernel discrete cosine transform features which improves the performance significantly yielding a 91.33% verification rate at 0.1% FAR, and we also show that by working in the DCT transform domain for obtaining non-linear features is more optimal than working in the original spatial-pixel domain which only yields a verification rate of 85% at 0.1% FAR. Thus our proposed method outperforms the baseline by 79.33% in verification rate @0.1% False Acceptance Rate.

Future data storage systems must cope with higher densities and lower signal-to-noise ratios. Decision-aided timing recovery loops will likely need a longer delay in the sequence detector to make reliable bit decisions. Loop delay degrades the dynamics of the phase-locked loop (PLL). In this paper we present a process to determine optimal gains for PLL with delay assuming that the timing drift is well modeled by a second order random walk. Simulation shows that the loss of lock rate from the proposed method is similar to the delay compensation method in [1, 2], with both approaches offering much lower loss of lock rates than conventional PLL without delay compensation. The simulation also shows there is an optimal delay. Method in [1, 2] adds components to the PLL, whereas no extra hardware is needed for the proposed method.

Two-dimensional (2-D) face recognition (FR) is of interest in many verification (1:1 matching) and identification ($1:N$matching) applications because of its nonintrusive nature and because digital cameras are becoming ubiquitous. However, the performance of 2-D FR systems can be degraded by natural factors such as expressions, illuminations, pose, and aging. Several FR algorithms have been proposed to deal with the resulting appearance variability. However, most of these methods employ features derived in the image or the space domain whereas there are benefits to working in the spatial frequency domain (i.e., the 2-D Fourier transforms of the images). These benefits include shift-invariance, graceful degradation, and closed-form solutions. We discuss the use of spatial frequency domain methods (also known as correlation filters or correlation pattern recognition) for FR and illustrate the advantages. However, correlation filters can be computationally demanding due to the need for computing 2-D Fourier transforms and may not match well for large-scale FR problems such as in the Face Recognition Grand Challenge (FRGC) phase-II experiments that require the computation of millions of similarity metrics. We will discuss a new method [called the class-dependence feature analysis (CFA)] that reduces the computational complexity of correlation pattern recognition and show the results of applying CFA to the FRGC phase-II data.

Face recognition is a challenging task. This paper introduces a novel set of biometrics, defined in the frequency domain and representing a form of "facial asymmetry." A comparison with existing spatial asymmetry measures suggests that the frequency-domain representation provides an efficient approach for performing human identification in the presence of severe expressions and for expression classification. Error rates of less than 5% are observed for human identification and around 25% for expression classification on a database of 55 individuals. Feature analysis indicates that asymmetry of the different face parts helps in these two apparently conflicting classification problems. An interesting connection between asymmetry and the Fourier domain phase spectra is then established. Finally, a compact one-bit frequency-domain representation of asymmetry is introduced, and a simplistic Hamming distance classifier is shown to be more efficient than traditional classifiers from storage and the computation point of view, while producing equivalent human identification results. In addition, the application of these compact measures to verification and a statistical analysis are presented

Correlation filtering methods are becoming increasingly popular for image recognition and location. The recent introduction of optimal tradeoff circular harmonic function filters allowed the user to specify the response of a correlation filter to in-plane rotation distortion. In this paper we introduce a new correlation filter design that can provide a user-specified response to in-plane scale distortion. The design is based on the Mellin radial harmonic (MRH) transform and incorporates multiple harmonics into the correlation filter for improved discrimination capability. Additionally, the filter design minimizes the average correlation energy in order to achieve sharp correlation peaks, and thus we refer to these filters as minimum average correlation energy Mellin radial harmonic (MACE-MRH) filters. We present underlying theory, a MACE-MRH filter design method, and numerical simulation results.

Subspace methods such as PCA, LDA, ICA have become a standard tool to perform visual learning and recognition. In this paper we propose representational oriented component analysis (ROCA), an extension of OCA, to perform face recognition when just one sample per training class is available. Several novelties are introduced in order to improve generalization and efficiency: (1) combining several OCA classifiers based on different image representations of the unique training sample is shown to greatly improve the recognition performance. (2) To improve generalization and to account for small misregistration effect, a learned subspace is added to constrain the OCA solution, (3) a stable/efficient generalized eigenvector algorithm that solves the small size sample problem and avoids overfitting. Preliminary experiments in the FRGC Ver 1.0 dataset show that ROCA outperforms existing linear techniques (PCA, OCA) and some commercial systems.

In data storage systems, the phase-locked loop (PLL) in timing recovery can be modeled as a Kalman filter. Goal of the acquisition stage is to estimate the initial phase and frequency disturbances. Kalman filter achieves linear minimum mean square error (LMMSE) estimation, but it requires knowing the exact prior information. In practice, the prior information is not precisely known. In this paper we develop closed form expressions for the mean squared error (MSE) of phase and frequency estimation by Kalman filter with imprecise prior information. With them, we can quantitatively tell how much prior information mismatch can be tolerated.

The nonlinear nature of holographic data storage systems (HDSS) suggests that nonlinear equalization and detection techniques may be beneficial. The complexity involved in nonlinear methods does not often make them practical solutions. Support vector machines (SVMs) are recently being studied for pattern recognition applications. We investigated linear SVM detection and observed that the bit error rate (BER) using SVM for data detection on linear minimum mean squared error (LMMSE) equalized holographically recorded and retrieved 2D data pages is about 17% better than the simple threshold detection on unequalized pages.

Low-density parity check (LDPC) codes with their near-Shannon capacity limit error correcting performance and iterative decoding algorithm are being evaluated for digital communications applications. For LDPC codes to be used in real systems, their error floors need to be investigated. In this paper, we evaluate the performance of disjoint difference set (DDS)-based LDPC codes (with column weights 3, 4, 5) and array code-based LDPC codes (with column weights 3, 4, 5) in the additive white Gaussian noise (AWGN) channel using a high-speed field programmable gate array (FPGA) simulation platform. The error floor regions (bit error rates down to 10/sup -12/) of those codes are presented. For better performance of array codes, a girth optimization method is proposed and the FPGA evaluation results are presented.

We introduce wavelet packet correlation filter classifiers. Correlation filters are traditionally designed in the image domain by minimizing some criterion function of the image training set. Instead, we perform classification in wavelet spaces that have training set representations which provide better solutions to the optimization problem in the filter design. We propose a pruning algorithm to find these wavelet spaces using a correlation energy cost function, and we describe a match score fusion algorithm for applying the filters trained across the packet tree. The proposed classification algorithm is suitable for any object recognition task. We present results by implementing a biometric recognition system using the NIST 24 fingerprint database, and show that applying correlation filters in the wavelet domain results in considerable improvement of the standard correlation filter algorithm.

A new frequency domain face recognition method using wavelet decomposition and quaternion correlation filters is proposed. The wavelet decomposition of the face image leads to a wavelet subband representation, which contains four subband images corresponding to four orthogonal channels. These four subbands can be encoded into a 2D quaternion number array. The quaternion correlation filter method is developed to perform pattern recognition jointly on multichannel 2D signals. The proposed method has been shown to achieve significant improvement in face recognition results compared to the traditional advanced correlation filter method for handling illumination variations of face images. We experimented with the CMU PIE database consisting of 65 people with 21 illumination variations per person, showing that our method can achieve close to 100% recognition accuracy using just a single training image of a person under neutral frontal lighting and testing on all other unseen harsh illumination conditions.

In disk forensics application, the goal is to determine whether a particular floppy disk was recorded in a particular drive. This paper discusses how image processing is used to determine whether a disk was recorded in drive A or drive B. Microscope images of floppy disk surfaces are used to extract features. Each image exhibits tracks and gaps as parallel shallow arcs, and the track and gap widths are extracted as features. We apply an image processing method to the shallow arcs to extract their widths. The Hough transform (PVC Hough, 1962) and the concept of entropy are used to segment each arc, and a simple and unbiased parabola fitting is used to fit a circle to the arc segment. Results show this method extracts track and gap widths of real images successfully resulting in the two floppy drives being distinguishable.

We investigated the performance of three face verification algorithms (correlation filters, Individual PCA, and FisherFaces) on an image database collected by a cell phone camera. Cell phone camera images tend to be of poorer quality along with experiencing scale and dynamic illumination changes due to cell phone portability. While Individual PCA and FisherFaces work in the image domain, correlation filters work in the frequency domain and offer advantages such as shift-invariance, the ability to accommodate in-class image variability, and closed-form expressions. Verification results suggest that, with this database, correlation filters can offer a better performance than Individual PCA and comparable performance with FisherFaces with fewer filters.

This work describes a generalized decoder implementation for structured low-density parity check (LDPC) codes. The decoder features low logic consumption, efficient memory management, and full parameterization for reconfiguration. The goal is to provide a unified solution for fast evaluation of a broad class of structured LDPC codes utilizing the properties of field programmable gate arrays (FPGA): high speed and configurability. As a fully reconfigurable core, it is ready to be used in different applications to lower the design to market time. The throughput and resource consumptions are evaluated.

Digital zero phase start (DZPS) (Buch B., US patent 6,191,906, 2001) is a method used in timing recovery for data storage systems to jump-start the acquisition process. The DZPS sums readback signal samples and uses the arctan function. We propose a simplified DZPS, which only averages the readback signal samples and the arctan function operation can be omitted. We investigate the statistics of the two methods and find that the simplified DZPS degrades the phase estimate mean square error negligibly.

Correlation filters are an attractive image processing technique for object recognition. They can provide the necessary recognition accuracy for many applications, but it would be desirable to reduce the complexity of the correlation filter algorithm (in terms of computation and storage space). This is especially true for biometric identification tasks, where multiple correlation filters must be tested against a single image. We propose an algorithm for match metric computation that trades a (usually minor) degradation in accuracy for an orders-of-magnitude complexity reduction. This algorithm analyzes ID cross-sections of the frequency domain in which the filter is applied. We compare our proposed technique to the standard technique using a dataset of face images.

The paper summarizes new research in performing face recognition using advanced correlation filters. We examine the performance of such filters in the area of biometrics for face authentication. We also compare results when the filters are applied to face identification. Our results are based on the illumination subsets of the CMU PIE database. We also present methods that reduce the memory requirements of these filters to run on limited computational resources, including computationally efficient methods of synthesizing these filters. Finally, we describe an online training algorithm implemented on a face verification system for synthesizing correlation filters from a video stream to handle pose/scale variations. The system also uses an efficient scheme to perform face localization within the current framework during the authentication stage.

We present a new tracking algorithm to perform timing recovery for low-SNR partial response data storage channels. This approach, designated "dual segmented Kalman filters based symbol timing recovery (DSK-STR)", is based on the conventional phase-locked loop (PLL) based timing recovery scheme. However, DSK-STR uses two intertwined PLLs and the gains of these PLLs are updated at each symbol instant according to a modified Kalman filter algorithm. The performance of DSK-STR is investigated via Monte-Carlo simulations and it is seen that DSK-STR offers a 3.5 dB signal-to-noise ratio (SNR) gain over the conventional PLL-based timing recovery scheme for a target loss of lock rate of 1/spl times/10/sup -5/ in longitudinal magnetic recording channels.

In this paper we show mathematical formulation of incrementally building advanced correlation filters used in authentication systems that are based on face and fingerprint images as biometrics for verification. This method is crucial for incorporating such algorithms on small devices with limited memory and computational resources. We also present results that show that these correlation filters perform well for face and fingerprint images. We used the PIE (pose, illumination and expression) database from CMY to test the verification performance using face images. Similarly, for fingerprint images we used the NIST special database 24 to evaluate verification performance.

Future very high density data storage systems will exhibit significantly more intersymbol interference (ISI) and significantly lower signal-to-noise ratio (SNR). Advanced signal detection algorithms, such as noise predictive maximum likelihood (NPML) and iterative soft decoding, are aimed at coping with such lower SNRs and higher ISI. However, at such low SNRs, because of their large residual timing jitter, current timing recovery schemes suffer frequently from loss of lock (the event where the estimated phase drift differs significantly from the actual phase drift for a significantly long duration leading to misindexing of the detected bits and thus error bursts) potentially offsetting the SNR gains provided by advanced detection methods. The main contribution of this paper is that by approximating the phase drifts present in recording systems as a piecewise linear phase drift model, we propose a novel timing recovery scheme, which is named frequency offset feedforward symbol timing recovery (FOSTR). For such a piecewise linear phase drift model, the problem of estimating the time-changing phase drift can be transformed to the problem of estimating the slopes (i.e., frequency offsets) and initial phase offsets of several linear ramps. The performance of the timing recovery based on this approach will be better than current methods in low SNR because the number of parameters to be estimated is significantly smaller. Bit-by-bit simulations with iterative soft decoding (low density parity check (LDPC) code of rate 16/17 and codeword size 4352 bits is used) show that FOSTR results in a significantly smaller residual timing jitter than that of the conventional decision-directed PLL-based feedback timing recovery schemes, although the adjusted ("adjusted" means the sectors suffering from loss of lock are not taken into account in bit error rate calculation) bit error rate (BER) performance of FOSTR is only about 0.6 dB better for a target adjusted BER of 1/spl times/10/sup -5/ than that of the conventional timing recovery scheme because iterative soft decoding is robust to residual timing jitter. However, the loss of lock rate (i.e., the fraction of sectors suffering from loss of lock) is significantly reduced by FOSTR.

This paper constructs and analyzes a class of regular LDPC codes with column weight of j=2, in contrast to the often-used j/spl ges/3 setting. These codes possess several significant features. First, they are free of 6-cycle, and can be easily constructed for a large range of code rates. Secondly, the parity check matrix of the code can be represented by a simple set, thus lending itself to a low complexity implementation. Thirdly, the proposed codes concatenated with proper precoder outperform j/spl ges/3 LDPC codes for partial response (PR) channels. Finally, they exhibit block error statistics significantly different from LDPC codes with j/spl ges/3, making them more compatible with Reed-Solomon error correction codes. The LDPC coded partial response (PR) channel is formulated as a dynamical model and analyzed using density evolution technique, which is used to explain the behavior of the concatenated system. A high rate (8/9) code with block size 4608 is constructed as an example, and its bit error rate (BER), block error statistics, and decoding convergence over ideal PR channel are investigated using simulation. The simulation results are consistent with the density evolution analysis, both indicating that LDPC codes with j=2 are attractive for partial response channels. PR targets for magnetic recording channel are used as examples to illustrate the performance of the proposed codes.

Biometric recognition refers to the process of matching an input biometric to stored biometric information. In particular, biometric verification refers to matching the live biometric input from an individual to the stored biometric template about that individual. Examples of biometrics include face images, fingerprint images, iris images, retinal scans, etc. Thus, image processing techniques prove useful in the biometric recognition. We discuss spatial frequency domain image processing methods useful for biometric recognition.

Data pages retrieved from a volume holographic data storage channel (VHSC) suffer mainly from intersymbol interference (ISI) and noise. ISI can be combatted using equalizers that are designed based on careful modeling of the VHSC. In this paper, we present a simple and accurate ISI model for the VHSC and utilize this efficient model to suggest a new iterative magnitude-squared decision feedback equalization (IMSDFE) method. Numerical simulations show that the outputs from the proposed model agree well with expected results and that the IMSDFE is a promising equalization/detection scheme to reduce the bit error rate (BER) of VHSC.

Dropouts are intermittent losses of signal commonly seen in magnetic tape recording readout. The main reason for such losses is the increased spacing between the head and the medium due to media defects or debris particles. The resulting signal is not only degraded by the apparent amplitude loss, but the characteristics of the pulses due to transitions are also changed. Moreover, in many cases, the locations of the pulses are altered, causing excessive amounts of peakshift. In this paper, a model, linking the liftoff to these effects, is presented. Experimental verification of the model using actual signals from a test tape drive is also given. Artificial dropout waveforms generated using this model are used to test two different read channel strategies. The first approach is an equalization-based correction scheme that attempts to undo the dropout. Results indicate that the dropout effects can be almost completely eliminated if an appropriate equalization procedure is applied. As an alternative approach, it is shown that the use of turbo coding in the presence of dropouts appears to be promising.

Turbo codes and low-density parity check (LDPC) codes with iterative decoding have received significant research attention because of their remarkable near-capacity performance for additive white Gaussian noise (AWGN) channels. Previously, turbo code and LDPC code variants are being investigated as potential candidates for high-density magnetic recording channels suffering from low signal-to-noise ratios (SNR). We address the application of turbo codes and LDPC codes to magneto-optical (MO) recording channels. Our results focus on a variety of practical MO storage channel aspects, including storage density, partial response targets, the type of precoder used, and mark edge jitter. Instead of focusing just on bit error rates (BER), we also study the block error statistics. Our results for MO storage channels indicate that turbo codes of rate 16/17 can achieve coding gains of 3-5 dB over partial response maximum likelihood (PRML) methods for a 10/sup -4/ target BER. Simulations also show that the performance of LDPC codes for MO channels is comparable to that of turbo codes, while requiring less computational complexity. Both LDPC codes and turbo codes with iterative decoding are seen to be robust to mark edge jitter.

We consider the issue of combating signal dropouts in magnetic tape recording. Dropouts are a major limiting factor in the performance of the channel as they cause long bursts of bit errors. To attack this problem, a model of the dropout event has been developed. Based on that model, a strategy has been identified in which an envelope detector followed by a feedforward equalizer is used. The equalization process is done in such a way as to address the three major effects caused by a dropout event: amplitude loss, pulse widening and peakshift. The crucial part of this scheme is the spacing profile estimator, which makes use of an envelope detector. With this scheme, the effects of the dropout are corrected at the expense of a delay in the channel. The model of the dropout and the correction scheme are all presented.

Turbo codes have been investigated for magnetic recording channels as potential candidates to succeed the current Viterbi-based partial response maximum likelihood (PRML) approaches. In contrast, relatively less attention has been devoted to the investigation of turbo decoding methods for optical data storage channels. In this paper, we present the simulation results of applying turbo decoding to an optical data storage channel. Our results focus on a variety of practical optical storage channel aspects including storage density, partial response targets, and the type of precoder used.

Adaptive filters have been traditionally used in data storage channels mainly for equalization purposes. In this paper, an adaptive filters is investigated for timing recovery so that equalization and timing recovery can be carried out simultaneously. Simulation results show that the bit error rate (BER) performance of the adaptive filter timing recovery (AFTR) method is comparable to other existing schemes. The main advantages of AFTR are that it is potentially simpler and can lead to a fully digital implementation.

Correlation methods are becoming increasingly attractive tools for image recognition and location. This renewed interest in correlation methods is spurred by the availability of high-speed image processors and the emergence of correlation filter designs that can optimize relevant figures of merit. In this paper, a new correlation filter design method is presented that allows one to optimally tradeoff among potentially conflicting correlation output performance criteria while achieving desired correlation peak value behavior in response to in-plane rotation of input images. Such controlled in-plane rotation response is useful in image analysis and pattern recognition applications where the sensor follows a pre-arranged trajectory while imaging an object. Since this new correlation filter design is based on circular harmonic function (CHF) theory, we refer to the resulting filters as optimal tradeoff circular harmonic function (OTCHF) filters. Underlying theory, OTCHF filter design method, and illustrative numerical results are presented.

A holographic data storage system (HDSS), just like other storage systems, suffers from interference between adjacent symbols and from noise. In this paper, we present a method for applying the linear minimum mean squared error (LMMSE) equalization to a HDSS. Numerical results with five defocused pages of data indicate that a significant improvement in the bit error rate (BER) is possible using LMMSE equalization.

Competitive learning paradigms are usually defined with winner-take-all (WTA) activation rules. The paper develops a mathematical model for competitive learning paradigms using a generalization of the WTA activation rule (g-WTA). The model is a partial differential equation (PDE) relating the time rate of change in the ;density' of weight vectors to the divergence of a vector field called the neural flux. Characteristic trajectories are used to study solutions of the PDE model over scalar weight spaces. These solutions show how the model can be used to design competitive learning algorithms which estimate the modes of unknown probability density functions.

The competitively inhibited neural network (CINN) is a competitive learning paradigm which is modeled by a collection of ordinary differential equations. A sliding threshold condition has been derived for determining the activity of a CINN neuron. This condition allows the development of a mathematical model for CINN learning. The model is a nonlinear diffusion equation whose solution quantitatively characterizes the learning process. Simulation experiments have validated this model

The competitively inhibited neural net (CINN) is a special class of competitive learning paradigm which is amenable to formal analysis. The authors present evidence that the CINN is capable of globally optimizing certain problems. They suggest that the CINN is capable of locating the primary mode of a source density function. In the event that this density represents a performance functional (such as in maximum-likelihood estimation), the CINN can be used to locate the global optimum of the performance functional. The mechanisms behind this global search property have been explained using a nonlinear diffusion model of CINN learning, and simulation experiments have corroborated this capability of the CINN for a minimally deceptive problem

t-random-error correcting and all-unidirectional-error detecting (t-EC/AUED) codes obtained by appending a suffix to each codeword of a linear block code are presented. Although this technique has been previously used to construct such codes, a different approach is proposed that involves starting with a nonoptimal t-EC linear code to obtain more efficient t-EC/AUED codes. Specifically, t-EC/AUED codes constructed from an even-weight t-EC linear code are shown to require a shorter suffix than when a general t-EC linear code is used. The t-EC/AUED codes obtained in this manner are thoroughly compared to the best known codes, and are shown to require fewer bits in almost all cases.>

The probabilities of the bits produced by an optimal binary source encoder for a memoryless source are analyzed. The class of sources for which the probability of zero must equal 1/2 for any such encoder is described. Procedures that minimize (maximize) the absolute difference of the bit probabilities are discussed. Corresponding upper bounds are derived and compared for the class of uniform sources.>

Very efficient suboptimal algorithms for the design of phase-only filters (POFs) and binary-phase-only filters (BPOFs) are presented. A reduction of 2 to 3 orders of magnitude in computer time is obtained over previous algorithms. The loses in signal-to-noise ratio is negligible (less than 0.001 dB). The authors have implemented and tested the new algorithm with excellent results. In their computer simulations they used the 32 by 32 gray level tank image. To compute the OPOF, the tank image is zero-padded and a 64 by 64 FFT is performed.>

Decoding rules to minimize the probability of information bit errors in binary linear block codes are investigated. A novel decoding rule is proposed and compared to a recently proposed rule (see M. Elia and G. Prati, 1985) and to the standard rule, with respect to both the information bit error rate and the information word error rate. It is concluded that the proposed rule not only can be implemented much more efficiently that the standard rule, but also is superior with respect to either error rate. Some particular codes have been presented to show that the authors' rule can coincide or can differ from Elia and Prati's rule when applied to a given code.>

Although optimal binary source coding using symbol blocks of increasing length must eventually yield a code having an average codeword length arbitrarily close to the source entropy, it is known that the sequence of average codeword lengths need not be nonincreasing. The sequence is, however, bounded above by the average codeword length of the source, and certain subsequences must be nondecreasing. Sufficient conditions are obtained describing sources for which a decrease in average codeword length is achieved when coding pairs of symbols. A sufficient condition specifying sources for which no such decrease is possible is also obtained.

A set of images is modeled as a stochastic process and Karhunen-Loeve expansion is applied to extract the feature images. Although the size of the correlation matrix for such a stochastic process is very large, we show the way to calculate the eigenvectors when the rank of the correlation matrix is not large. We also propose an iterative algorithm to calculate the eigenvectors which save computation time andc omputer storage requirements. This iterative algorithm gains its efficiency from the fact that only a significant set of eigenvectors are retained at any stage of iteration. Simulation results are also presented to verify these methods.