Intra prediction is a crucial component of hybrid video coding framework due to its remarkable ability to reduce spatial redundancy in video signals. Unlike the single-mode based intra prediction in HEVC and VVC, intra fusion prediction methods, that combine the results of multiple angular prediction modes, were newly adopted by Enhanced Compression Model (ECM). However, intra fusion prediction over-relies on local spatial correlations and neglects potential texture similarities in non-adjacent regions. To overcome these limitations and elevate the accuracy of luma intra prediction, a Recurrent Intra Prediction Mode (RIPM) is proposed in this paper, which is composed of two sub-modules, i.e., Recurrent Intra Merge Mode (RIMM) and Recurrent Block Vector Substitution Module (RBVSM). RIMM utilizes the recurrent spatial texture information of the adjacent and non-adjacent spaces for adaptive mode derivation and prediction within the intra fusion prediction framework. RBVSM is a sophisticated mechanism for adaptive prediction mode selection and weight assignment during intra fusion prediction, resulting in enhanced coding performance with minimal impact on computational complexity. The proposed method, implemented on top of ECM-12.0, demonstrates a 0.095% BD-rate gain for the luma component under All Intra configuration, with negligible complexity increase. Currently, RIPM is under study in Exploration Experiments (EE) for ECM in JVET.

Intra-picture prediction as in HEVC exploits the nearest reference line adjacent to the current coding unit (CU) for prediction of samples. If this reference line represents a discontinuity, the reference samples in this reference line can differ to a large extent from the original samples and may lead to a large prediction error. We propose a multiple reference lines (MRLs) coding to allow not only the nearest reference line 0 but also reference lines 1 and 3 to be candidates for angular intra prediction as shown in Fig. 1. To reduce the complexity arising from additional lines to be checked at encoder side, we further propose to restrict the MRL to angular most probable modes (MPMs) only. The MRL coding signals the reference line index before the intra prediction mode. This allows to not signal the MPM flag of the current CU and implicitly derive it as true when a non-zero reference line index is signaled. Experimental results are provided to evaluate the performance of the proposed MRL coding on top of the VVC test model VTM-2.0.1. 26 test sequences in different categories, including 4k, 1080p, 720p, WVGA, WQVGA resolutions and screen contents are tested. Two coding structures are evaluated, all intra (AI) and random access (RA). The objective coding efficiency is measured in terms of Bjøntegaard Delta (BD) rate (%) computed using four rate/PSNR points that were generated by using quantization parameters 22, 27, 32 and 37. Lower (negative) BD-rate implies better compression rate. Table 1 shows that the presented MRL provides 0.46% bitrate savings for an all-intra and 0.2% for a random-access configuration on average. Furthermore, it provides 1.45% bitrate reduction for screen content test sequences, which are representing an increasingly important video application. Because of a fairly good trade-off between coding efficiency and complexity, the proposed MRL coding mode with MPM restriction was adopted into the current VVC draft standard.

To realize the medical video applications, this paper proposes several lossless compression methods over high efficiency video coding (HEVC). A generalized intra block copy (GIBC) is first proposed to predict the coding unit by a reference block, whose samples could be fully or partially reconstructed. A cyclic block padding technique is also proposed to predict the unreconstructed samples in the reference block by geometrically co-located blocks. Based on the feature distribution analyses for palette coding, we further propose an HEVC-based medical video coder (HMC), which combines the GIBC, line-coded palette coding and intra palette predictor without mutual conflicts. Experimental results show that, compared to the lossless HEVC, the proposed GIBC and HMC respectively save up to 13.9% and 22.3% bits over medical videos.

Children with intellectual disabilities (ID) may need to undergo long-term cognitive training to enhance the possibility of social inclusion. Pedestrian safety is a critical issue when it comes to participating in community activities for children with ID. Currently, no commercial pedestrian safety training products are available for public special education schools. Consequently, teachers normally gave lectures with the support of slides and pictures to enhance children's ability of getting around in the community. This study employs the Microsoft Kinect technology and image recognition technology to create a pedestrian safety training system which may be applicable for people with ID. To motivate people with ID to engage in the training, we gamified the training of pedestrian safety for children with ID in special education school settings. By leveraging the Scratch language and Kinect2Scratch tool, the teacher who will use the system may be able to do the customization without technical support. Preliminary results of 15 children with ID, aged 9-10, testing the game are presented.

This paper proposes a novel temporal distortion cost (TDC) for sample adaptive offset (SAO) filter in High Efficiency Video Coding. The TDC addresses the images with temporal distortion by evaluating the errors between pre/post-SAO samples and raw samples in continuous frames. Due to high computational complexity of TDC, we further propose a low-complexity TDC (L-TDC) to estimate the value of TDC by a difference classification scheme and a rate-distortion decision scheme. We also contribute to deriving a spatio-temporal distortion cost (STDC) and a low-complexity STDC (L-STDC) by incorporating a spatial distortion cost into the proposed TDCs. Simulation results show that, compared to the classical SAO filter, the proposed SAO filters can save up to around 10% ~ 15% bitrates under the same flicker measure, and achieve better flicker measure with over 10% improvement in average under the same bitrate. Some valuable threshold analyses are also provided.

A Hadamard transformed sum of absolute difference (SATD)-based joint decision algorithm (SJDA) is proposed for intra prediction encoder in High Efficiency Video Coding. The SJDA performs the SATD-only distortion costs in 4×4 prediction units (PUs) to select rough modes of larger PUs. An adaptive block selection scheme is proposed in SJDA to adaptively adjust the PU sizes every 16×16 block by an approximate SATD-based rate distortion cost. To smartly select intra prediction modes that will be enabled for rate-distortion optimization (RDO) calculations, the SJDA proposes a novel evaluation signaled scheme to signal the evaluation results to a decision module by deriving a decision rule. Compared to HM, the SJDA averagely saves 73% encoding time with +3.3% BD-rate, while reducing 60% RDO modules in a parallelized encoder, 80% SATD operations, and 77% RDO operations. Simulations also show the proposed algorithms are superior to classical ones.

This paper proposes arbitrary reference tier coding (ARTC) schemes to allow the intra directional modes to exploit farther reference tiers for intra prediction. The proposed ARTCs with four reference tiers are applied to a developing future video codec, Joint Exploration Test Model (JEM), under quad-tree (QT) and QT plus binary tree (QTBT) split structures respectively. A joint intra mode and reference tier decision (JMTD) encoding algorithm exclusive to the QTBT is further proposed to efficiently select the intra mode and reference tier. The simulation results show that the ARTC can improve up to 1.2% coding gain compared to the JEM, and the JMTD-based ARTC can reduce 40% encoding time without any coding gain drop compared to the ARTC under QTBT structure. Several analyses are also provided to show that 40-60% encoding time of the ARTC can be further saved with 0.28-0.31% BD-rate increase by disabling some new technologies.

This study improved the ARCoach system that provided vocational training for individuals with cognitive impairments using an Augmented Reality (AR) based task prompting technology. Using cyclic codes to enhance AR tag recognition rates, the proposed system not only provided picture cues but also identified incorrect task steps on the fly and helped make corrections. Experimental data showed that the participant considerably increased the target response, thus improving vocational job skills during the intervention phases. In addition to assistive technology, our approach can apply to other AR applications where tag recognition rates need to be improved.

Children with cerebral palsy (CP) may need to undergo long-term physical rehabilitation to enhance neural development. Currently, commercial rehabilitation products cannot be customized easily and are expensive; consequently, public special-education schools generally cannot afford purchasing these products. This study employs the Microsoft Kinect technology and image recognition technology to create a rehabilitation system which may be applicable for people with CP. To motivate people with CP to engage in exercise training, we gamified the movement training of self-feeding and self-dressing for young adults with CP in special education school settings. By leveraging the Scratch language and Kinect2Scratch tool, the physical therapists (PTs) who will use the system may be able to do the customization without technical support. Preliminary results of 5 healthy college students testing the exercise games are presented. The precision and reliability of the Kinect2Scratch Games are 100% and 97.5%.

We study anomaly detection in a context that considers user trajectories as input and tries to identify anomalies for users following normal routes such as taking public transportation from the workplace to home or vice versa. Trajectories are modeled as a discrete-time series of axis-parallel constraints ("boxes") in the 2-D space. The anomaly can be estimated by considering two trajectories, where one trajectory is the current movement pattern and the other is a weighted trajectory collected from N norms. The proposed system was implemented and evaluated with eight individuals with cognitive impairments. The experimental results showed that recall was 95.0% and precision was 90.9% on average without false alarm suppression. False alarms and false negatives dropped when axis rotation was applied. The precision with axis rotation was 97.6% and the recall was 98.8%. The average time used for sending locations, running anomaly detection, and issuing warnings was in the range of 15.1-22.7 s. Our findings suggest that the ability to adapt anomaly detection devices for appropriate timing of self-alerts will be particularly important.

This paper proposes a powerful symmetry- constrained concurrent blind filter (SC-CBF) for adaptive multi- antenna beamforming systems. The proposed SC-CBF beamformer is created according to the concurrently activated symmetry-constrained constant modulus algorithm (SC-CMA) and symmetry-constrained decision-directed filter (SC-DD). Both SC-CMA and SC-DD are obtained from a generalized ideal weight property, which is derived based on an ideal blind beamforming weight expression. Following the generalized ideal weight property, the performance of the SC-CBF beamformer is quite robust to the error propagation occurring in low signal-to-noise ratios compared to the traditional CBF one. The simulation results have shown that the proposed SC-CMA and SC-CBF beamformers provide much faster convergence rates than the existing adaptive blind beamformers.

A capacity maximization problem subject to a total power constraint in OFDM-based wireless multicast systems is investigated in this paper. A subcarrier and power allocation method is proposed based on a property observed from the simulation in the multicast system. We design a criterion to efficiently assign subcarriers to multicast groups. Simulation results reveal that the proposed method outperforms the existing suboptimal methods. With the help of the improved mechanism and the designed criterion, the performance of the proposed method is almost the same as that of the optimal solution.

Depth ordering is instrumental for understanding the 3D geometry of an image. Humans are surprisingly good at depth ordering even with abstract 2D line drawings. In this paper we propose a learning-based framework for depth ordering inference. Boundary and junction characteristics are important clues for this task, and we have developed new features based on these attributes. Although each feature individually can produce reasonable depth ordering results, each still has limitations, and we can achieve better performance by combining them. In practice, local depth ordering inferences can be contradictory. Therefore, we propose a Markov Random Field model with terms that are more global than previous work, and use graph optimization to encourage a globally consistent ordering. In addition, to produce better object segmentation for the task of depth ordering, we propose to explicitly enforce closed loops and long edges for the occlusion boundary detection. We collect a new depth-order dataset for this problem, including more than a thousand human-labeled images with various daily objects and configurations. The proposed algorithm shows promising performance over conventional methods on both synthetic and real scenes.

This paper presents a novel allocation scheme to solve a subcarrier and power allocation problem for multiuser SC-FDMA uplink systems with a localized subcarrier mapping mode. In the localized mode, subcarriers assigned to a user are required to be consecutive. The limitation makes the problem considered in this paper different from those of conventional multiuser OFDM systems. The proposed scheme contains the chunk and the iterative subcarrier assignments. The chunk assignment utilizes an interior term to allocate a set of subcarriers to a user. The iterative assignment re-assigns the adjacent subcarriers between users to other users to improve the achievable rates. Furthermore, a constant power allocation approach is employed in conjunction with the proposed scheme. Simulation results show that the proposed scheme outperforms the existing schemes.

An autonomous task-prompting system is presented to increase workplace and life independence for people with cognitive impairments such as traumatic brain injury, intellectual disability, schizophrenia, and Down syndrome. This paper describes an approach to providing distributed cognition support of work engagement for persons with cognitive disabilities. In the pilot study, a prototype was built and tested in a community-based rehabilitation program involving pre-service food preparation training of 8 participants with cognitive impairments. The results show improvement in helping with task engagement is statistically significant compared to the oral-instruction method. A follow-up comparative study with 2 participants evaluated the shadow-team approach against the proposed system. Although the number of participants was few, the participants were studied in depth and the findings were very promising. The results in the autonomous task prompting without staff intervention indicate that the performance is statistically as good as the shadow-team approach. Our findings suggest that acquisition of job skills may be facilitated by the proposed system in conjunction with operant conditioning strategies.

This paper presents an active-learning algorithm for piecewise planar 3D reconstruction of a scene. While previous interactive algorithms require the user to provide tedious interactions to identify all the planes in the scene, we build on successful ideas from the automatic algorithms and introduce the idea of active learning, thereby improving the reconstructions while considerably reducing the effort. Our algorithm first attempts to obtain a piecewise planar reconstruction of the scene automatically through an energy minimization framework. The proposed active-learning algorithm then uses intuitive cues to quantify the uncertainty of the algorithm and suggest regions, querying the user to provide support for the uncertain regions via simple scribbles. These interactions are used to suitably update the algorithm, leading to better reconstructions. We show through machine experiments and a user study that the proposed approach can intelligently query users for interactions on informative regions, and users can achieve better reconstructions of the scene faster, especially for scenes with texture-less surfaces lacking cues like lines which automatic algorithms rely on.

Images taken from scenes under water suffer distortion due to refraction. While refraction causes magnification with mild distortion on the observed images, severe distortions in geometry reconstruction would be resulted if the refractive distortion is not properly handled. Different from the radial distortion model, the refractive distortion depends on the scene depth seen from each light ray as well as the camera pose relative to the refractive surface. Therefore, it's crucial to obtain a good estimate of scene depth, camera pose and optical center to alleviate the impact of refractive distortion. In this work, we formulate the forward and back projections of light rays involving a refractive plane for the perspective camera model by explicitly modeling refractive distortion as a function of depth. Furthermore, for cameras with an inertial measurement unit (IMU), we show that a linear solution to the relative pose and a closed-form solution to the absolute pose can be derived with known camera vertical directions. We incorporate our formulations with the general structure from motion framework followed by the patch-based multiview stereo algorithm to obtain a 3D reconstruction of the scene. We show through experiments that the explicit modeling of depth-dependent refractive distortion physically leads to more accurate scene reconstructions.

Semi-supervised learning uses both labeled and unlabeled data for machine learning tasks. It's especially useful in the scenarios where labeled data is very scarce or expensive to obtain. In this work, we present kernel LLC, the kernel locality-constrained linear coding within a data-dependent kernel space, for data representation. The data-dependent kernel captures the underlying data geometry on the ambient feature space. The kernel LLC further exploits the locality association among the data on its manifold. Promising results on both image classification and content-based image retrieval scenarios suggest kernel LLC to be a good candidate for data representation in semi-supervised learning.

A laser scan is useful in building the 3D model, and in one ran it can capture thousands of 3D points. However these 3D points are sparse compared to a normal image, which can easily have millions of pixels. To achieve a denser 3D map, 3D-interpolation is applied to each pixel in the image. In this work we propose an algorithm to combine the 3D geometry and the color for 3D-interpolation. We segment the 3D points based on their latent surfaces, and combine the surfaces with color through Markov Random Field. We find that the 3D geometry provides rich information for interpolation: 3D points with similar colors can be robustly clustered where not possible in the color space, and the interpolation can be performed on a better fitting surface rather than on the locally linear ones. Our experiments show that the proposed algorithm outperforms the baselines.

In this paper, we present a novel interactive 3D reconstruction algorithm which renders a planar reconstruction of the scene. We consider a scenario where the user has taken a few images of a scene from multiple poses. The goal is to obtain a dense and visually pleasing reconstruction of the scene, including non-planar objects. Using simple user interactions in the form of scribbles indicating the surfaces in the scene, we develop an idea of 3D scribbles to propagate scene geometry across multiple views and perform co-segmentation of all the images into the different surfaces and non-planar objects in the scene. We show that this allows us to render a complete and pleasing reconstruction of the scene along with a volumetric rendering of the non-planar objects. We demonstrate the effectiveness of our algorithm on both outdoor and indoor scenes including the ability to handle featureless surfaces.

We propose a self-growing strategy to improve the learning speed of the symmetric radial basis function (SRBF) beamformer for multi-antenna systems. This novel beamforming scheme is designed based on a geometric relationship between the hidden nodes of SRBF and the current array outputs. The center vectors of this novel scheme can be quickly and properly initialized, so the fast learning can be achieved. We have shown that, under a short training sequence, bit-error rates are greatly improved, even when the number of center vectors is really huge.

As Internet technologies evolve, their applications have changed various aspects of human life. Here, we attempt to examine their potential impact on services for families with developmentally delayed children. Our research is thus designed to utilize wireless mobile communication technologies, location services, and search technology in an effort to match families of specific needs with potential care providers. Based on the investigation conducted by our counselors, this paper describes a platform for smooth communication between professional communities and families with children with developmental disabilities (CDD). This research also looks into the impact of management of mobile social network services and training on the operation of these services. Interaction opportunities, care, and support to families with CDD are introduced.

10.5244/C.24.46

Segmentation-based approach has shown significant success in stereo matching. By assuming pixels within one image segment belong to the same 3D surface, robust depth estimation can be achieved by taking the whole segment into consideration. However, segmentation has been mostly used for stereo matching at integer disparities rather than subpixel disparities. One major reason is that small segments may be insufficient for estimating surfaces like slanted planes, while large segments may contain segmentation errors impacting the accuracy of depth estimation. In this work, we propose a segmentation-based scheme for subpixel stereo matching. Instead of using a fixed segmentation, segments are evolved to find a better support for reliable surface estimation. Given an initial estimation of segmentation and depth, the proposed algorithm jointly optimizes the segmentation and depth by evolving the segmentation at the pixel level and updating the plane parameters at the segment level. Justified with experiments performed on the Middlebury benchmark, we show that the proposed method achieves significant improvements for subpixel stereo matching.

In this paper, we propose the use of 3D information to augment the Markov random field (MRF) model for object recognition. Conventional MRF for image-based object recognition usually uses appearance and 2D location as features in the model. We estimate rough 3D information from stereo image pairs, and incorporate this information into node and edge potential models in the conventional MRF. Introducing 3D information into the node potential allows to leverage the distribution statistics of 3D location for different classes. We solve the object recognition problem by finding the globally optimal class assignment that minimizes an energy function defined in the augmented MRF. We show that the introduction of 3D distance in the edge potential can help distinguish “true” neighbors from “fake” neighbors in 2D. We demonstrate improved recognition results by using the proposed technique.

Active models are widely used in applications like image segmentation and tracking. Region-based active models are known for robustness to weak edges and high computational complexity. We found previous region-based models can easily get stuck in local minimums if initialization is far from the true object boundary. This is caused by an inherent ambiguity in evolution direction of the level set function when minimizing the energy. To solve this problem, we propose an intensity re-weighting (IR) model to bias the evolution process in certain direction. IR model can effectively avoid local minimums and enable much faster convergence of the evolution process. The proposed method is applied to both real and synthetic images with promising results.

Direct implementations of bilateral filtering show O(r2) computational complexity per pixel, where r is the filter window radius. Several lower complexity methods have been developed. State-of-the-art low complexity algorithm is an O(1) bilateral filtering, in which computational cost per pixel is nearly constant for large image size. Although the overall computational complexity does not go up with the window radius, it is linearly proportional to the number of quantization levels of bilateral filtering computed per pixel in the algorithm. In this paper, we show that overall runtime depends on two factors, computing time per pixel per level and average number of levels per pixel. We explain a fundamental trade-off between these two factors, which can be controlled by adjusting block size. We establish a model to estimate run time and search for the optimal block size. Using this model, we demonstrate an average speedup of 1.2-26.0x over the previous method for typical bilateral filtering parameters.

The widespread use of cellular telephones and the availability of user location information are facilitating personalized location-based applications and, in particular, mobile location-based social networking (MLSN). The research was started with an aim to use MLSN as a measure to increase social connectedness and improve the quality of life for the majority of otherwise-employable persons who remain unemployed, rarely access appropriate community services, and are socially isolated. An interactive, multimedia, location-based application approach has been proposed to address the needs that frequently arise in the field work of supported employment. An MLSN prototype has been implemented and the results are presented as a case study. Throughout the years of 2007–2008, we had eight test subjects with cognitive disabilities who participated in four independent sessions each, resulting in a total of 32 sessions as a whole. Precision of the MLSN system in the worst case was 89% and the recall in the worst case was 94% when users taking public transportation requested assistance from their care providers within 1 km. When the search range was extended to 3 km, both the precision and recall achieved 100%. To the best of our knowledge, the use of MLSN to assist people with various disabilities has not been previously reported in the literature except in the work by Chang et al. (2007 Chang , Y.-J. , Liu , H.-H. , Chou , L.-D. , Chen , Y.-W. , and Shin , H.-Y. 2007 . A general architecture of mobile social network services. Paper read at the 2nd International Conference on Convergence Information Technology, November 21–23, Gyeongju, Korea. [Google Scholar], 2008 Chang , Y.-J. , Liu , H.-H. , Peng , S.-M. , and Wang , T.-Y. 2008 . Potential of mobile social networks as assistive technology: A case study in supported employment for people with severe mental illness. Paper read at the 10th ACM International Conference on Computers and Accessibility (ASSETS 2008), October 13–15, Halifax, Canada. [Google Scholar], 2009 Chang , Y.-J. , Liu , H. H. , and Wang , T.-Y. 2009 . Mobile social networks as quality of life technology for people with severe mental illness . IEEE Wireless Communications 16 : 34 – 40 . [Google Scholar]) and Chou et al. (2008 Chou , L.-D. , Lai , N.-H. , Chen , Y.-W. , Chang , Y.-J. , Huang , L.-F. , Chiang , W.-L. , Chiu , H.-Y. , and Yang , J.-Y. 2008. Management of mobile social network services for families with developmental delay children. Paper read at IEEE Healthcom, July 7–9, Singapore. [Google Scholar]).

Wayfinding systems are an assistive technology targeting disabled patients who are mobile and need to travel in both indoor and outdoor environments, thus increasing workplace and life independence. Such a personal guidance system can improve the quality of life and minimize risks such as wandering behavior, which has been previously observed in patients with acquired brain injury. The proposed system called AssistMote is a full-ledged navigation solution using wireless sensor networks (WSN) for ambient condition detection, routing engines for navigation, and PDAs as the user interface, to assist with indoor wayfinding for individuals with multiple special needs. Ambient intelligence such as inadequate illumination and wet floors can be critical to wayfinding for the targeted population. The routing algorithm is based on the innovative use of Dijkstra's shortest path with the special requirements of navigation applications taken into account. The study reports the prototype design, experiments, and detailed performance evaluation based on computer simulations and field trials of eight human subjects with multiple disabilities who participated in pedestrian navigation in a five-floor building with 91 sensor nodes. AssistMote achieved 95% success compared with 37.5% in a controlled study that used verbal instructions and paper maps.

A challenge to individuals with cognitive impairments in workplaces is how to remain engaged, recall task routines, and transition autonomously across tasks in a way relying on limited cognitive capacity. A novel task prompting system is presented with an aim to increase workplace and life independence for people with traumatic brain injury, cerebral palsy, intellectual disability, schizophrenia, and Down syndromes. This paper describes an approach to providing distributed cognition support of work engagement for persons with cognitive disabilities. The unique strength of the system is the ability to provide unique-to-the-user prompts that are triggered by context. As this population is very sensitive to issues of abstraction (e.g. icons) and presents the designer with the need to tailor prompts to a 'universe-of-one' the use of picture or verbal cues specific to each user and context is implemented. The key to the approach is to spread the context awareness across the system, with the context being flagged by beacon sources and the appropriate response being evoked by displaying the appropriate task prompting cues indexed by the intersection of specific end-user and context ID embedded in the beacons. By separating the context trigger from the pictorial or verbal response, responses can be updated independently of the rest of the installed system, and a single beacon source can trigger multiple responses in the PDA depending on the end-user and their specific tasks. A prototype is built and tested in field experiments involving eight individuals with cognitive impairments. The experimental results show the task load of the human-device interface is low or very low and the capabilities of helping with task engagement are high and reliable.

Relevance feedback has been taken as an essential tool to enhance content-based information retrieval systems by keeping the user in the retrieval loop. Among the fundamental relevance feedback approaches, feature space warping has been proposed as an effective approach for bridging the gap between high-level semantics and the low-level features. By examining the fundamental behavior of the feature space warping, we propose a new approach to harness its strength and resolve its weakness under various data distributions. Experiments on both synthetic data and real data reveal significant improvement from the proposed method.

In this paper, we propose a general architecture of mobile social networks where individuals of the networks form a virtual community, conversing and connecting with one another based on mobile communication. Second, mobile social networks are implemented in collaboration with a supported employment program set up for people with severe mental illness.

A challenge to individuals with cognitive impairments in wayfinding is how to remain oriented, recall routines, and travel in unfamiliar areas in a way relying on limited cognitive capacity. According to psychological model of spatial navigation and the requirements of rehabilitation professionals, a novel wayfinding system is presented with an aim to increase workplace and life independence for people suffering from diseases such as traumatic brain injury, cerebral palsy, mental retardation, schizophrenia, Down syndromes, and Alzheimer's disease. This paper describes an approach to providing distributed cognition support of travel guidance for persons with cognitive disabilities. The unique strength of the system is the ability to provide unique-to-the-user prompts that are triggered by context. As this population is very sensitive to issues of abstraction (e.g. icons) and presents the designer with the need to tailor prompts to a 'universe-of-one' the use of images specific to each user and context is implemented. The key to the approach is to spread the context awareness across the system, with the context being flagged by the QR-code tags and the appropriate response being evoked by displaying the appropriate path guidance images indexed by the intersection of specific end-user and context ID embedded in the tags. By separating the context trigger from the pictorial response, responses can be updated independently of the rest of the installed system, and a single QR-code tag can trigger multiple responses in the PDA depending on the end-user and their specific path. A prototype is built and tested in field experiments with real patients. The experimental results show the human-computer interface is friendly and the capabilities of wayfinding are reliable.

In this paper, a 3-layer Bayesian hierarchical detection framework (BHDF) is proposed for robust parking space detection. In practice, the challenges of the parking space detection problem come from luminance variations, inter- occlusions among cars, and occlusions caused by environmental obstacles. Instead of determining the status of parking spaces one by one, the proposed BHDF framework models the inter-occluded patterns as semantic knowledge and couple local classifiers with adjacency constraints to determine the status of parking spaces in a row-by-row manner. By applying the BHDF to the parking space detection problem, the available parking spaces and the labeling of parked cars can be achieved in a robust and efficient manner. Furthermore, this BHDF framework is generic enough to be used for various kinds of detection and segmentation applications.

The study follows action science approach to the problem of nonprofit housing services. After 4 months of action sciencebased activities, such as organized participant observations, indepth interviews, field work, and focus group studies, the main findings are (1) Web 2.0 suits nonprofit organizations better than traditional webs in terms of maintenance cost and usability, (2) mapping tools make better GUI with respect to web-based housing services, and (3) context-aware personalization can translate to better user experiences as an RDFS-based working prototype has been built and tested. A user survey shows high level user satisfaction. Although the case study was carried in a nonprofit housing organization, the practices in action research approach can be applied to other NPOs as well.

In this paper, we present a wayfinding prototype system based geo-coded QR codes for individuals with cognitive impairments. The design draws upon the cognitive models of spatial navigation and consists of wayfinding devices and a tracking system. Compared to the sensor network approach, it is easy to deploy because of low cost and short time frame. The prototype is tested with routes on a campus where a rehabilitation trained job coach oversees the process. The results show the prototype is user friendly and promising with high reliability.

One of the teaching challenges in computer-based laboratory is tracking student progress and giving assistance on an individual basis in the lab classroom. When finishing a lab, students are asked to submit their results at their own blogs and make a notice by placing a link at the instructor's blog. Pursuing the links and checking out students' blogs, the instructor can grasp immediate status of student progress and make informed judgments on which students to walk towards and initiate further interactions with. We also use blogs as tools of social learning. By browsing the selected peer works and the comments made by instructors on students' blogs, students are empowered to stimulate new ideas different from each other. Survey results shows that students who took the lab courses are satisfied with turning in lab results through blog posting, getting instructor's in-class interactions, piracy reduction, and social learning.

A method to detect moving objects on non-stationary background is proposed. The concurrent motions of foreground and background pixels make it extremely difficult to maintain a plausible background model for background subtraction. In our method, motion fields of aligned neighboring frames are fused to reduce parallax effects in moving blob detection. A fused color background model is further developed to refine shapes of detected objects. Finally, moving blob information is incorporated into the adaptation process of background model. Only confidently marked background pixels are adapted into background models with each incoming frame. Experimental results shown robust, well-shaped moving object detection can be obtained under unconstrained scenes.

In this paper, we propose a novel face authentication scheme using the active appearance model (AAM) and the hidden Markov model (HMM). The proposed face authentication system can be divided into two parts. First, the AAM is used to extract the low-dimensional feature vectors including combined texture and shape information of individual face images. The extracted feature vectors are further classified into several clusters using vector quantization. The clustered feature vectors are then characterized using HMMs to make full use of the temporal information across the face images. After all parameters in the HMMs are calculated, we can dynamically determine the thresholds for face authentication. An iterative algorithm is also proposed to automatically determine a suitable number of HMM states and a suitable number of observation classes to achieve good authentication accuracy. The experimental results show the efficacy of the proposed method

In this paper, we introduce a novel method for key generation based on biometrics. Given the biometrics, a set of reliable features are extracted. Each feature is compared with multiple thresholds to generate a multi-bit key. By cascading the multi-bit keys, we obtain one biokey that can be used for security applications. In order to generate a reliable bio-key, an optimal thresholding method is proposed to minimize the authentication error rate, in terms of the false accept rate (FAR) and the false reject rate (FRR). The experimental results show that the proposed approach of key generation is user-friendly and reliable.

Instead of using PINs and passwords as cryptographic keys that are either easy to forget or vulnerable to dictionary attacks, easy-to-carry and difficult-to-transfer keys can be generated based on user-specific biometric information. A framework is proposed to generate stable cryptographic keys from biometric data that is unstable in nature. The proposed framework differs from prior work in that user-dependent transforms are utilized to generate more compact and distinguishable features. Thereby, a longer and more stable bitstream can be generated as the cryptographic key. Experiments are performed on a face database to verify the feasibility of the proposed framework. The preliminary result is very encouraging.

Fast, reliable, and marker-free facial expression analysis still remains to be a difficult task in computer vision research. In this paper, the concept of speech-assisted facial expression analysis and synthesis is proposed, which shows that the speech-driven facial animation technique not only can be used for expression analysis. From the input speech, the mouth shape can is estimated from the audio-visual model. Thus, the large search space of mouth appearance is reduced for mouth tracking. Similarly, the modeling technique is extended from modeling speech and mouth shape to facial movements and detail facial texture changes. In this way, a virtual conferencing system with video realistic avatars is realized to meet real-time requirement.

Facial animation standardized by MPEG-4 provides a common form of description and transmission for talking head related applications. With animated talking heads, a virtual conferencing system can be created by providing a 3-D virtual environment for face-to-face communication and casual navigation. Not only bit-rate consumption is reduced, 3-D visualization is also provided which greatly improves the sense of presence when compared to conventional video conferencing system. An integrated architecture is presented by taking advantage of 2-D video coding and 3-D model-based coding to create video realistic avatars for virtual face-to-face conferencing system. Preliminary experimental results indicate more than 4 dB improvement in PSNR can be achieved at the same bit rate when compared to conventional video coding. The incorporation of 3-D facial models also enables rendering from arbitrary viewpoints for use in a virtual conferencing system.

A 3D textured polygonal facial model estimation algorithm is presented. The algorithm takes facial model estimation, texture mapping, and head pose estimation as complementary processes, which cooperate to adapt the facial model from a generic facial model to a user-accustomed one through image sequences. The proposed scheme is performed with a single camera without calibration and requires only a little manual adjustment, which proves to be a feasible approach for facial model estimation.

In model-based virtual conferencing systems, the changes of facial expressions on human faces are major focus of all users. In order to represent the detail changes of facial expressions, two algorithms for facial feature point tracking are developed to track motion of facial features. The first algorithm achieves medium to high accuracy with low computational complexity. And the second algorithm adopts hierarchical mesh models to algorithm is utilized to translate the tracking result to facial animation parameters, which can be used for driving MPEG-4 compliant talking heads.

This paper presents an H.323 standard compliant video conferencing system implementation. The proposed system not only serves as an MCU (Multipoint Control Unit) for multipoint connection but also provides a gateway function between the H.323 LAN (Local Area Network) and the H.324 WAN (Wide Area Network) users. The proposed video conferencing system provides user-friendly object compositing and manipulation features including 2-D video object scaling, re-positioning, rotating, and dynamic bit-allocation in a 3-D virtual environment. A segmentation scheme based on pre- stored background information is proposed for real-time segmentation of the foreground video objects at the client side. Chroma-key insertion is used to facilitate video objects extraction and manipulation. We have implemented the virtual conference system prototype with an integrated graphic user interface to demonstrate the feasibility of the proposed methods.

This paper presents a novel face-assisted video coding scheme to enhance the visual quality of the face regions in video telephony applications. A skin-color based face detection and tracking scheme is proposed to locate the face regions in real-time. After classifying the macroblocks into the face and non-face regions, we present a dynamic distortion weighting adjustment (DDWA) scheme to drop the static non-face macroblocks, and the saved bits are used to compensate the face region by adjusting the distortion weighting of the face macroblocks. The quality of face regions will thus be enhanced. Moreover, the computation originally required for the skipped macroblocks can also be saved. The experimental results show that the proposed method can significantly improve the PSNR and the subjective quality of face regions, while the degradation introduced on the non-face areas is relatively insensitive to human perception. The proposed algorithm is fully compatible with the H.263 standard, and the low complexity feature makes it well suited to implement for real-time applications.

In this paper, an implementation of a virtual chat room system for multimedia communications is presented. In addition to the conventional ability of chat room applications to transmit talk messages between users, the system provides a 3-D virtual environment and adopts 3-D facial models for character representation. Each participant can communicate with the others with keyboard, mouse, and his facial expressions. Essential implementation issues including the facial expression analysis process are detailed in this article.

We consider a class of zero-one integer programming feasibility problems (0-1 ILPF problems) in which the coefficients of variables can be integers, and the objective is to find an assignment of binary variables so that all constraints are satisfied. We propose a Lagrangian formulation in the continuous space and develop a gradient search in this space. By using two counteracting forces, one performing gradient search in the primal space (of the original variables) and the other in the dual space (of the Lagrangian variables), we show that our search algorithm does not get trapped in local minima and reaches equilibrium only when a feasible assignment to the original problem is found. We present experimental results comparing our method with backtracking and local search (based on random restarts). Our results show that 0-1 ILPF problems of reasonable sizes can be solved by an order of magnitude faster than existing methods.

Finding the global optimal solution for a general nonlinear program is a difficult task except for very small problems. We identify a class of nonlinear programming problems called polynomial programming problems (PP). A polynomial program is an optimization problem with a scalar polynomial objective function and a set of polynomial constraints. By using Groebner Bases, we can determine the global minimum of a polynomial program in a reasonable amount of time and memory.>

In a distributed computing environment, a testbed is built to investigate topology learning behaviors of bridges for LAN interconnection. It is flexible to configure on any networked computing facility and thus serves as an environment for rapid prototyping of bridge systems. Unlike a pure simulator, the testbed employs distributed computing in itself so that the parallelism embedded in protocols under test can be exploited as well. With modular design and unified interprocess interfacing, it can be employed to explore a variety of topology algorithms. In addition, as part of testbed functions, it can simulate network topology of interest. It also provides the challenges of distributed environments, performance reports, dynamical topology change scenarios, and a graphical user interface. By direction linking the actual code of the protocols to the testbed, the phases of design, development, and testing of new systems are fully integrated.>

This paper concerns scheduling parallel programs in multiprocessing systems. The goal is to improve the job turnaround time and speedup. As most programs are not fully parallelizable, parallel programs consisting of serial, partially parallel, and (perfectly) parallel stages, are modeled by task graphs which depict the parallelism profiles. In another perspective, the study is intended to be a queueing counterpart of Amdahl's Law. Various two-stage cases are investigated with FCFS or priority scheduling policies and the optimal strategy is determined. Results indicate that, with nonhomogeneous parallelism profiles, it would be a better processor allocation policy to give a preferential treatment to serial stages. The study might hopefully improve the design of parallel operating systems as well as parallel algorithms, databases, and parallelizing compilers.