When humans interact with each other, eye gaze movements have to support motor control as well as communication. On the one hand, we need to fixate the task goal to retrieve visual information required for safe and precise action-execution. On the other hand, gaze movements fulfil the purpose of communication, both for reading the intention of our interaction partners, as well as to signal our action intentions to others. We study this Gaze Dialogue between two participants working on a collaborative task involving two types of actions: 1) individual action and 2) action-in-interaction. We recorded the eye-gaze data of both participants during the interaction sessions in order to build a computational model, the Gaze Dialogue, encoding the interplay of the eye movements during the dyadic interaction. The model also captures the correlation between the different gaze fixation points and the nature of the action. This knowledge is used to infer the type of action performed by an individual. We validated the model against the recorded eye-gaze behavior of one subject, taking the eye-gaze behavior of the other subject as the input. Finally, we used the model to design a humanoid robot controller that provides interpersonal gaze coordination in human-robot interaction scenarios. During the interaction, the robot is able to: 1) adequately infer the human action from gaze cues; 2) adjust its gaze fixation according to the human eye-gaze behavior; and 3) signal nonverbal cues that correlate with the robot's own action intentions.

Pedestrian trajectory prediction is a relevant task for many kinds of intelligent systems. However, it can be quite challenging, since humans can be influenced by a plethora of factors. Two types of factors have been getting more relevance: the presence of obstacles and social interactions. Existing trajectory prediction methods that incorporate both criteria require information like video images, which may not be readily available. We propose a new model, named Arc-LSTM-SMF, which considers the existence of obstacles and social interactions. To our knowledge, it is one of the first methods to include both factors while requiring only pedestrian trajectories as input to accurately perform prediction. This model integrates Sparse Motion Fields with LSTM networks, and introduces a new pooling layer that simulates a field of view for each pedestrian. We evaluate our model using standard geometric metrics, as well as metrics related to obstacle avoidance and pedestrian collision avoidance. The proposed Arc-LSTM-SMF is able to outperform several state-of-the-art models on popular pedestrian datasets.

Video analysis often requires locating and tracking target objects. In some applications, the localization system has access to the full video, which allows fine-grain motion information to be estimated. This paper proposes capturing this information through motion fields and using it to improve the localization results. The learned motion fields act as a model-agnostic temporal regularizer that can be used with any localization system based on keypoints. Unlike optical flow-based strategies, our motion fields are estimated from the model domain, based on the trajectories described by the object keypoints. Therefore, they are not affected by poor imaging conditions. The benefits of the proposed strategy are shown on three applications: 1) segmentation of cardiac magnetic resonance; 2) facial model alignment; and 3) vehicle tracking. In each case, combining popular localization methods with the proposed regularizer leads to improvement in overall accuracies and reduces gross errors.

Providing reliable descriptions of the agents in a video scene is an essential task in many applications, such as surveillance. However, most works focus solely on the characterization of pedestrians, which is not sufficient to describe complex scenes, where a variety of vehicles (e.g., bikes and cars) are also present. In this work we address this limitation and propose a framework based on switching motion fields to efficiently characterize the different agents in a scene. Our method achieves a balanced accuracy of 91.9% on the identification of bikers and pedestrian classes on three challenging scenarios, and provides comprehensive information about their behaviors.

Complex surveillance scenarios comprise different types of agents (e.g., bikers, cars, and pedestrians) that must be efficiently characterized in order to facilitate tasks such as tracking or abnormality detection. This paper proposes an unsupervised hierarchical multiple motion fields model to represent different types of agents, which relies in the combination of hierarchical Markov model and velocity fields. Model parameters are estimated using the expectation-maximization algorithm. The proposed framework was applied to synthetic and real datasets (Stanford Drone Dataset), showing the ability to characterize and classify different agents in an unsupervised way.

Dermoscopy image analysis (DIA) is a growing field, with works being published every week. This makes it difficult not only to keep track of all the contributions, but also for new researchers to identify relevant information and new directions to be explored. Several surveys have been written in the past decade, but these tend to cover all of the steps of a CAD system, which can be overwhelming. Moreover, in these works, each of the steps is briefly discussed due to lack of space. Among the different blocks of the CAD system, the most relevant is the one devoted to feature extraction. This is also the block where existing works exhibit the most variability. Therefore, we believe that it is important to review the state-of-the-art on this matter. This work thoroughly explores the several types of features that have been used in DIA. A discussion on their relevance and limitations, as well as suggestions for future research are provided.

This paper considers the representation of human trajectories in video signals. These trajectories are modeled by switched dynamical models, based on motion fields that drive the pedestrian during consecutive time intervals. This paper addresses the estimation of uncertainty in trajectory generation by using space-varying covariance matrices estimated from the video data. Experimental results show that the proposed model outperforms previous methods, based on static and isotropic covariance matrices.

Obtaining the segmentation of an object in a sequence of images is usually achieved using a tracking methodology. However, in some applications, the whole sequence is available beforehand. This means that the segmentations can be determined simultaneously for all the frames in the sequence and taking into account the motion of the object. This paper proposes a new framework to incorporate motion information in the segmentation of image sequences using an active shape model (ASM). The motion of the object is modeled using a vector field, which is learned and refined online as the segmentation algorithm proceeds. The vector field is determined from the trajectories described by ASM points throughout the sequence. The vector field, in turn, influences the estimation of the ASM parameters by acting as a regularizer, ensuring that the segmentations are in agreement with the expected motion. The results show that coupling these models during the segmentation leads to an increase in performance, in particular by guarantee more consistent segmentations and by avoid gross errors in more challenging frames.

This paper proposes a novel methodology to describe the trajectories performed by pedestrians in long-range surveillance scenarios. The proposed approach describes the trajectories by using sparse motion/vector fields together with a space-varying switching mechanism embedded in a Hidden Markov Model framework. Despite the diversity of motion patterns that may occur in a given scenario, the observed trajectories do not lie in the entire surveilled area. Instead, they are constrained to patterns corresponding to typical motions. To achieve a compact representation, we propose a sparse model estimated using the ℓ1norm applied to the log prior distribution of the vector fields. Experimental evaluation is conducted in real scenarios, and testify the usefulness of the proposed approach in modeling typical trajectories that occur in a far-field surveillance setup.

This paper proposes a new approach for the segmentation of the endocardium of the left ventricle using short axis magnetic resonance (MR) images. The proposed method comprises two main stages. First, each image is converted to polar coordinates, and an edge map is computed from the transformed image. Then, the contour of the left ventricle (LV) is estimated by computing the optimal path along the edge map, using a dynamic programming approach. The system is evaluated on a public database comprising 660 magnetic resonance volumes and the results testify its usefulness both in terms of running time and accuracy. The proposed methodology is able to segment a whole volume in 1.5 seconds achieving an average Dice similarity coefficient of 85.9% (8.3%), which compares favorably with related state-of-the-art methods.

A learning system that is able to predict the degradation state of impact craters on optical images is presented in this paper. It is based on the extraction of visual features along the crater rim together with the decision with a SVM classifier. The algorithm achieved a sensitivity of 89% and a specificity of 96% (preserved vs non-preserved) in a dataset of annotated craters from Mars.

The 3D segmentation of the left ventricle (LV) in cardiac MRI is a challenging problem, due to the presence of other anatomical structures and artifacts (outliers) around the LV. In this paper, a new formulation of a Robust Active Shape Model (RASM) is presented that is able to deal with those outliers. Instead of using the traditional one-to-one mapping of edge points and model points to compute the shape model parameters, the proposed approach uses a one-to-many mapping strategy and groups these edge points into edge segments (strokes). Then, a probabilistic framework provides a robust estimation of the model parameters, in which the influence in the segmentation of the unreliable outliers is reduced. The proposed method was tested on a public dataset comprising 660 volumes. The results indicate that this methodology provides accurate segmentations that are competitive with other state-of-the art methods.

This book is part of the CRC Press Digital Imaging and Computer Vision Series of books with a focus on the growing field of digital analysis of dermoscopy images. The dermatoscope has become a ubiquitous tool for dermatologists to evaluate pigmented lesions, but it can have a steep learning curve, and the current reporting/grading schemes are highly subjective, making it difficult to achieve interobserver consistency. Having more algorithmic approaches towards processing dermoscopy images could lead to more consistent interpretation and aid in training residents, fellows and practicing dermatologists. Additionally, the existence of this digital imaging data will allow more advanced computer-aided diagnosis (CAD) methods to be applied in this field, allow data to be shared among clinicians through common databases and electronic medical records, and advance research in this field by providing expanded access to imaging data for a larger variety of skin conditions. This book provides a comprehensive summary of the current state of the art in digital dermoscopic image analysis across a wide variety of topics.

An algorithm to automatically delineate the contour of impact craters detected on remotely sensed images from the surface of Mars is presented in this letter. It processes the crater images in polar coordinates and is constituted by two main steps: edge enhancement, which is constructing an edge map based upon the intensity transitions along radial lines intersecting the center of the crater, and crater delineation, which is determining an optimal path from the minimization of an energy functional by Dynamic Programming. We obtained the performance of 96% of correct crater delineations evaluated in a data set of 1045 craters depicted from High Resolution Imaging Science Experiment (HiRISE) and Thermal Emission Imaging System (THEMIS) images (resolutions of 0.25-0.50 and 100 m/pixel, respectively) exhibiting a large diversity of terrains, crater dimensions, and degradations.

Statistical shape models have been extensively used in a wide range of applications due to their effectiveness in providing prior shape information for object segmentation problems. The most popular method is the Active Shape Model (ASM). However, accurately fitting the shape model to an object boundary under a cluttered environment is a challenging task. Under such assumptions, the model is often attracted towards invalid observations (outliers), leading to meaningless estimates of the object boundary. In this paper, we propose a novel algorithm that improves the robustness of ASM in the presence of outliers. The proposed framework assumes that both type of observations (valid observations and outliers) are detected in the image. A new strategy is devised for treating the data in different ways, depending on the observations being considered as valid or invalid. The proposed algorithm assigns a different weight to each observation. The shape parameters are recursively updated using the Expectation-Maximization method, allowing a correct and robust fit of the shape model to the object boundary in the image. Two estimation criteria are considered: 1) the maximum likelihood criterion; and 2) the maximum a posteriori criterion that uses priors for the unknown parameters. The methods are tested with synthetic and real images, comprising medical images of the heart and image sequences of the lips. The results are promising and show that this approach is robust in the presence of outliers, leading to a significant improvement over the standard ASM and other state of the art methods.

Robustness is one of the most important characteristics of computer-aided diagnosis systems designed for dermoscopy images. However, it is difficult to ensure this characteristic if the systems operate with multisource images acquired under different setups. Changes in the illumination and acquisition devices alter the color of images and often reduce the performance of the systems. Thus, it is important to normalize the colors of dermoscopy images before training and testing any system. In this paper, we investigate four color constancy algorithms: Gray World, max-RGB, Shades of Gray, and General Gray World. Our results show that color constancy improves the classification of multisource images, increasing the sensitivity of a bag-of-features system from 71.0% to 79.7% and the specificity from 55.2% to 76% using only 1-D RGB histograms as features.

The segmentation of the left ventricle (LV) is an important task to assess the cardiac function in ultrasound images of the heart. This paper presents a novel methodology for the segmentation of the LV in three-dimensional (3-D) echocardiographic images based on the probabilistic data association filter (PDAF). The proposed methodology begins by initializing a 3-D deformable model either semiautomatically, with user input, or automatically, and it comprises the following feature hierarchical approach: 1) edge detection in the vicinity of the surface (low-level features); 2) edge grouping to obtain potential LV surface patches (mid-level features); and 3) patch filtering using a shape-PDAF framework (high-level features). This method provides good performance accuracy in 20 echocardiographic volumes, and compares favorably with the state-of-the-art segmentation methodologies proposed in the recent literature.

In this paper we present a sea vessel detection algorithm in aerial image sequences acquired by an unmanned aerial vehicle. The proposed method is robust to variable background lighting, highlights due to sun reflections, vehicle self motion and scale changes. By relying in simple blob analysis rules, based on both spatial and temporal constraints, the algorithm is capable of real-time operation onboard the vehicle, even with non optimized code. We evaluate our method on three sequences labeled with ground truth vessel position, with more that 2900 frames. Overall we are able to achieve very low false positive rates even in heavy sun reflection conditions.

Development of Computer Aided Diagnosis systems that mimic the performance of dermatologists when diagnosing dermoscopy images is a challenging task. Despite the relevance of color in the diagnosis of melanomas, few of the proposed systems exploit this characteristic directly. In this paper we propose a new methodology for color identification in dermoscopy images. Our approach is to learn a statistical model for each color using Gaussian mixtures. The results show that the proposed method performs well, with an average Spearman correlation of 0.7981, with respect to a human expert.

Several methods have been proposed to detect melanomas in dermoscopy images. However, most of these methods are tuned to specific acquisition conditions. That is, their performance is affected when the acquisition setup changes or when data comes from multiple sources. This is what happens with EDRA database that contains images acquired in three different hospitals. In this paper, we discuss the use of color compensation techniques that try to reduce the influence of the acquisition setup on the color features extracted from the images. We show that color compensation provides a significant improvement on the performance of two different systems.

Active shape models (ASM) have been extensively used in object segmentation problems because they constrain the solution, using shape statistics. However, accurately fitting an ASM to an image prone to outliers is difficult and poor results are often obtained. To overcome this difficulty we propose a robust algorithm based on the Expectation-Maximization framework that assigns different weights (confidence degrees) to the observations extracted from the image. This reduces the influence of outliers since they often receive low weights. We tested the proposed algorithm with synthetic and real images (e.g., lip images and cardiac ultrasound images) achieving promising results. The proposed algorithm performs significantly better than the standard ASM implementation.

This paper proposes an iterative natural gradient algorithm to perform the optimization of switching probabilities in a space-varying hidden Markov model, in the context of human activity recognition in long-range surveillance. The proposed method is a version of the gradient method, developed under an information geometric viewpoint, where the usual Euclidean metric is replaced by a Riemannian metric on the space of transition probabilities. It is shown that the change in metric provides advantages over more traditional approaches, namely: 1) it turns the original constrained optimization into an unconstrained optimization problem; 2) the optimization behaves asymptotically as a Newton method and yields faster convergence than other methods for the same computational complexity; and 3) the natural gradient vector is an actual contravariant vector on the space of probability distributions for which an interpretation as the steepest descent direction is formally correct. Experiments on synthetic and real-world problems, focused on human activity recognition in long-range surveillance settings, show that the proposed methodology compares favorably with the state-of-the-art algorithms developed for the same purpose.

This paper presents a novel manifold learning algorithm for high-dimensional data sets. The scope of the application focuses on the problem of motion tracking in video sequences. The framework presented is twofold. First, it is assumed that the samples are time ordered, providing valuable information that is not presented in the current methodologies. Second, the manifold topology comprises multiple charts, which contrasts to the most current methods that assume one single chart, being overly restrictive. The proposed algorithm, Gaussian process multiple local models (GP-MLM), can deal with arbitrary manifold topology by decomposing the manifold into multiple local models that are probabilistic combined using Gaussian process regression. In addition, the paper presents a multiple filter architecture where standard filtering techniques are integrated within the GP-MLM. The proposed approach exhibits comparable performance of state-of-the-art trackers, namely multiple model data association and deep belief networks, and compares favorably with Gaussian process latent variable models. Extensive experiments are presented using real video data, including a publicly available database of lip sequences and left ventricle ultrasound images, in which the GP-MLM achieves state of the art results.

Estimation of velocity fields from a video sequence is an important step towards activity classification in a surveillance system. It has been recently shown that multiple motion fields estimated from trajectories are an efficient tool to describe the movement of objects, allowing an automatic classification of activities in the scene. However, the trajectory detection in noisy environments is difficult, usually requiring some sort manual editing to complete or correct them. This paper proposes two novel contributions. First, an automatic method for building pedestrian trajectories in far-field surveillance scenarios is presented not requiring user intervention. This basically comprises the detection of multiple moving objects in a video sequence through the detection of the active regions, followed by the estimation of the velocity fields that is accomplished by performing region matching of the above regions at consecutive time instants. This leads to a sequence of centroids and corresponding velocity vectors, describing the local motions presented in the image. A motion correspondence algorithm is then applied to group the centroids in a contiguous sequence of frames into trajectories corresponding to each moving object. The second contribution is a method for automatically finding the trajectories from a library of previously computed ones. Experiments on extensive video sequences from university campuses show that motion fields can be reliably estimated from these automatically detected trajectories, leading to a fully automatic procedure for the estimation of multiple motion fields.

The segmentation of the left ventricle (LV) is an important tool to assess the cardiac function in ultrasound images of the heart. This paper presents a methodology for the segmentation of the LV in 3D echocardiography that is based on the probabilistic data association filter (PDAF). The proposed methodology comprises the following feature hierarchical approach: (i) edge detection in the vicinity of the surface (low level features); (ii) edge grouping to obtain potential LV surface patches (mid-level features); and (iii) patch filtering using a shape-PDAF framework (high level features). Also, we propose an automatic procedure to initialize the 3D deformable model. We show that the proposed methodology achieves remarkable accuracy between the obtained contour and expertise medical ground truth and compares favorably with the state-of-the-art segmentation methodologies.

The analysis of moving objects in image sequences (video) has been one of the major themes in computer vision. In this paper, we focus on video-surveillance tasks; more specifically, we consider pedestrian trajectories and propose modeling them through a small set of motion/vector fields together with a space-varying switching mechanism. Despite the diversity of motion patterns that can occur in a given scene, we show that it is often possible to find a relatively small number of typical behaviors, and model each of these behaviors by a "simple" motion field. We increase the expressiveness of the formulation by allowing the trajectories to switch from one motion field to another, in a space-dependent manner. We present an expectation-maximization algorithm to learn all the parameters of the model, and apply it to trajectory classification tasks. Experiments with both synthetic and real data support the claims about the performance of the proposed approach.

Many approaches to trajectory analysis, such as clustering or classification, use probabilistic generative models, thus not requiring trajectory alignment/registration. Switched linear dynamical models (e.g., HMMs) have been used in this context, due to their ability to describe different motion regimes. However, these models are not suitable for handling space-dependent dynamics that are more naturally captured by nonlinear models. As is well known, these are more difficult to identify. In this paper, we propose a new way of modeling trajectories, based on a mixture of parametric motion vector fields that depend on a small number of parameters. Switching among these fields follows a probabilistic mechanism, characterized by a field of stochastic matrices. This approach allows representing a wide variety of trajectories and modeling space-dependent behaviors without using global nonlinear dynamical models. Experimental evaluation is conducted in both synthetic and real scenarios. The latter concerning with human trajectory modeling for activity classification, a central task in video surveillance.

Positron Emission Tomography plays an important role as an Alzheimer's Disease (AD) early diagnosis tool, and also identifying Mild Cognitive Impairment (MCI) patients. The vast majority of 3D brain image-based computer aided diagnosis methods implemented so far relied simply on voxel intensity, as feature. In this article, we consider two alternative methods of feature extraction: 3D Haar-like features and histograms of gradient magnitude and orientation; their performance in the classification of AD vs. Cognitively Normal (CN), MCI vs. CN and AD vs. MCI patients is evaluated and compared to the one obtained when using voxel intensity only. Classification is accomplished through Support Vector Machines, after an automatic feature selection step. The features based on histograms of the gradient attained the best results in AD vs. CN discrimination, and 3D Haar-like features improved performance in all three classification tasks. These improvements encourage further investigation on these extraction strategies.

Video cameras monitoring human activities in public spaces are commonplace in cities worldwide. Such monitoring task is important for safety and security purposes but is also extremely challenging. In this paper, we propose a class of algorithms for far-field human activity recognition, a central task in video surveillance. More specifically, we explore a class of parametric motion vector fields learned from the trajectories described by people in real-world scenarios. The work proposed herein is a space dependent framework, in sense that the vector fields depend on the pedestrian position. Thus, the model is flexible leading to an expressive description of complex trajectories. Also, a model selection strategy is addressed to automatically choose the appropriate number of underlying motion fields presented in the trajectories. Experimental evaluation is conducted in real settings testifying the usefulness of the proposed approach for human activity recognition.

This paper addresses an image labeling problem, in which it is assumed that there are multiple sensors available at each pixel with some of them possibly inactive. In addition to not being known which sensors are active or inactive, the sensor measurements are also obtained in random unknown order. Given these incomplete observations, we wish to identify which sensors are active at each site and which observations were produced by each sensor. This labeling problem extends classic image segmentation, since it allows multiple labels (i.e., region overlapping). The paper provides methods to solve this problem in two scenarios: known and unknown sensor models. A new minimization algorithm, inspired by hierarchical clustering, is introduced to minimize the energy function resulting from the proposed inference criterion.

In this paper we introduce a novel method to determine the model order of a stochastic model for moving objects. The main assumption is that we make use of the knowledge that the obtained model is going to be used for some task, specifically, for trajectory classification. Particularly, the object motion is described by trajectories performed by the objects (e.g., pedestrians), during their motion, by representing them by a small and meaningful mixtures of vector fields. We present a discriminative method for model selection without resort to computationally expensive cross-validation procedures. The idea is, thus, to select the generative model achieving the best classification performance. Although the topic of application is video surveillance, the proposed method can easily be extended to other practical situations. Experiments with both synthetic and real data concerning pedestrian activities illustrate the performance of the proposed approach.

Many approaches to trajectory analysis tasks (such as clustering or classification) use probabilistic generative models, thus not requiring trajectory alignment/registration. Switched linear dynamical models (e.g., HMMs) have been used in this context, due to their ability to describe different motion regimes. However, this type of models is not suitable for handling space-dependent dynamics, that are more naturally captured by non-linear models. As is well known, these are more difficult to identify. We propose a new way of modeling trajectories, based on a mixture of parametric motion vector fields that depend on a small number of parameters. Switching among these fields follows a probabilistic mechanism, characterized by a field of stochastic matrices. This approach allows representing a wide variety of trajectories and modeling space-dependent behaviors without using global non-linear dynamical models. The proposed model is applied to human trajectory modeling, a central task in video surveillance.

An approach for the automated detection of dune fields on remotely sensed images of the surface of Mars is presented in this letter. It is based on the extraction of local information from images (i.e., gradient features), which, in turn, is tested with boosting and support vector machine classifiers. A detection rate of about 95% is obtained for fivefold cross validation on a set of 78 panchromatic images captured by the Mars Orbiter Camera of the Mars Global Surveyor probe on different locations of the planet.

The gradient vector flow (GVF) algorithm has been extensively used as a tool to extract the boundary of objects in images. However, its performance significantly degrades when the images have a cluttered background i.e., other structures in the vicinity of the object of interest. In this case, the elastic contour is often attracted towards invalid features. This paper discusses the robustness of GVF and proposes a way to improve its performance. The main difference concerns the use of more robust and informative features (middle level features) which significantly reduce the influence of noise. Experiments are presented to illustrate the performance of the improved GVF algorithm.

We propose a method to classify human trajectories, modeled by a set of motion vector fields, each tailored to describe a specific motion regime. Trajectories are modeled as being composed of segments corresponding to different motion regimes, each generated by one of the underlying motion fields. Switching among the motion fields follows a probabilistic mechanism, described by a field of stochastic matrices. This yields a space-dependent motion model which can be estimated using an expectation-maximization (EM) algorithm. To address the model selection question (how many fields to use?), we adopt a discriminative criterion based on classification accuracy on a held out set. Experiments with real data (human trajectories in a shopping mall) illustrate the ability of the proposed approach to classify complex trajectories into high level classes (client versus non-client).

A central issue in mixture-type models is the determination of a suitable number of components that best suits the observed data. In this paper, we address this issue in the context of trajectory classification based on mixtures of motion vector fields. We adopt a discriminative criterion for choosing among alternative models for each class, based on the classification accuracy on a held out dataset. The key idea is that we make use of the knowledge that the obtained model is going to be used for a specific task: classification. Experiments with both synthetic and real data concerning pedestrian activity classification illustrate the performance of the adopted criterion.

Alzheimer's disease (AD) is one of the most frequent type of dementia. Currently there is no cure for AD and early diagnosis is crucial to the development of treatments that can delay the disease progression. Brain imaging can be a biomarker for Alzheimer's disease. This has been shown in several works with MR Images, but in the case of functional imaging such as PET, further investigation is still needed to determine their ability to diagnose AD, especially at the early stage of Mild Cognitive Impairment (MCI). In this paper we study the use of PET images of the ADNI database for the diagnosis of AD and MCI. We adopt a Boosting classification method, a technique based on a mixture of simple classifiers, which performs feature selection concurrently with the segmentation thus is well suited to high dimensional problems. The Boosting classifier achieved an accuracy of 90.97% in the detection of AD and 79.63% in the detection of MCI.

This paper proposes an approach for recognizing human activities (more specifically, pedestrian trajectories) in video sequences, in a surveillance context. A system for automatic processing of video information for surveillance purposes should be capable of detecting, recognizing, and collecting statistics of human activity, reducing human intervention as much as possible. In the method described in this paper, human trajectories are modeled as a concatenation of segments produced by a set of low level dynamical models. These low level models are estimated in an unsupervised fashion, based on a finite mixture formulation, using the expectation-maximization (EM) algorithm; the number of models is automatically obtained using a minimum message length (MML) criterion. This leads to a parsimonious set of models tuned to the complexity of the scene. We describe the switching among the low-level dynamic models by a hidden Markov chain; thus, the complete model is termed a switched dynamical hidden Markov model (SD-HMM). The performance of the proposed method is illustrated with real data from two different scenarios: a shopping center and a university campus. A set of human activities in both scenarios is successfully recognized by the proposed system. These experiments show the ability of our approach to properly describe trajectories with sudden changes.

This work introduces a new approach to modeling object trajectories in image sequences. Trajectories performed by natural objects (e.g., people, animals) typically depend on the position of each object in the scene and can change in an unpredictable way. Despite this diversity, there is often a small number of typical motion patterns based on which it is possible to explain all the observed trajectories. To achieve this goal, we model each of these motion patterns using a motion field and allow objects to switch between fields in a space-varying, possible probabilistic, way. Our approach provides a space-dependent motion model which can be estimated using an expectation-maximization (EM) algorithm. Experiments with both synthetic and real data are presented to illustrate the ability of the proposed approach in modeling different motion patterns.

An approach to automatically detect impact craters on planetary surfaces is presented in this letter. It is built up from a boosting algorithm proposed by Viola and Jones (2004) whose simplicity combined with an original learning strategy leads to a fast and robust process with consistent results. The approach is validated with image data sets from Mars surface captured by the Mars Orbiter Camera onboard Mars Global Surveyor probe.

This work introduces a non-supervised algorithm for learning generative models for classification/recognition of human activities (specifically, pedestrian trajectories) with application to video surveillance. The proposed algorithm comprises two main features: (?) a set of low level dynamical models of the trajectories, estimated in unsupervised manner using the expectation-maximization (EM) algorithm and automatic model selection using the minimum message length (MML) criterion; (ii) a switching dynamical model described by an hidden Markov model (HMM) used to characterize the higher level activities. The hierarchical model with these two levels is herein denoted as switched dynamical hidden Markov model (SD-HMM). We illustrate the performance of the proposed technique for human activity recognition in a university campus.

Geometric active contours based on edges perform poorly in the presence of noise or clutter. When the edges have gaps or are indistinct, the contour leaks through the boundary. Furthermore, when spurious edge points that do not belong to the object are present in the image, the contour is stopped by them and either does not converge to the object boundary or there is oversegmentation. This paper addresses the second difficulty. We propose a novel technique which classifies image features as valid or invalid making the curve stop only at valid features and allowing it to bridge the invalid ones. This is incorporated in the stopping force of boundary based level sets, achieving a robust contour estimation. Our algorithm organizes edge points into connected segments (denoted herein as strokes) and classifies each segment as valid or invalid. A confidence degree (weight) is assigned to each stroke and updated during the evolution process. Thus, the proposed stopping force is adaptive. Experimental results with real data will be provided to illustrate the performance of the proposed algorithm.

This paper addresses object tracking in ultrasound images using a robust multiple model tracker. The proposed tracker has the following features: 1) it uses multiple dynamic models to track the evolution of the object boundary, and 2) it models invalid observations (outliers), reducing their influence on the shape estimates. The problem considered in this paper is the tracking of the left ventricle which is known to be a challenging problem. The heart motion presents two phases (diastole and systole) with different dynamics, the multiple models used in this tracker try to solve this difficulty. In addition, ultrasound images are corrupted by strong multiplicative noise which prevents the use of standard deformable models. Robust estimation techniques are used to address this difficulty. The multiple model data association (MMDA) tracker proposed in this paper is based on a bank of nonlinear filters, organized in a tree structure. The algorithm determines which model is active at each instant of time and updates its state by propagating the probability distribution, using robust estimation techniques.

Multiplicative noise is often present in medical and biological imaging, such as magnetic resonance imaging (MRI), Ultrasound, positron emission tomography (PET), single photon emission computed tomography (SPECT), and fluorescence microscopy. Noise reduction in medical images is a difficult task in which linear filtering algorithms usually fail. Bayesian algorithms have been used with success but they are time consuming and computationally demanding. In addition, the increasing importance of the 3-D and 4-D medical image analysis in medical diagnosis procedures increases the amount of data that must be efficiently processed. This paper presents a Bayesian denoising algorithm which copes with additive white Gaussian and multiplicative noise described by Poisson and Rayleigh distributions. The algorithm is based on the maximum a posteriori (MAP) criterion, and edge preserving priors which avoid the distortion of relevant anatomical details. The main contribution of the paper is the unification of a set of Bayesian denoising algorithms for additive and multiplicative noise using a well-known mathematical framework, the Sylvester-Lyapunov equation, developed in the context of the Control theory.

This work introduces a semi-supervised approach for learning generative models for classification/recognition of human trajectories, with application to surveillance. The classifier is based on switched dynamical models, with each model describing a specific motion regime. We present a semi-supervised modified version of the classical Baum-Welch algorithm, which is able to take into account a subset of known model labels. The experimental results reported, using both synthetic and real data, show that the classifier learned with semi-supervision leads to a higher classification accuracy than the fully unsupervised version, thus validating the proposed approach.

It has been claimed that the first digit of real signals follows a logarithmic distribution, called Benford law. This paper shows that this distribution is a natural prior for the gradient of several types of medical images (MRI, CT, ultrasound) and proposes a reconstruction algorithm based on the Benford law which does not require any parameter tuning. Experimental results are presented to illustrate the performance of the reconstruction algorithm.

In this paper, we propose novel methods to evaluate the performance of object detection algorithms in video sequences. This procedure allows us to highlight characteristics (e.g., region splitting or merging) which are specific of the method being used. The proposed framework compares the output of the algorithm with the ground truth and measures the differences according to objective metrics. In this way it is possible to perform a fair comparison among different methods, evaluating their strengths and weaknesses and allowing the user to perform a reliable choice of the best method for a specific application. We apply this methodology to segmentation algorithms recently proposed and describe their performance. These methods were evaluated in order to assess how well they can detect moving regions in an outdoor scene in fixed-camera situations

The MAP method has the ability to deal with ill-posed problems, involving noisy data, by reducing the number of admissible solutions. However, this method is usually iterative and time consuming and not appropriate to work in a real time basis. In the context of image reconstruction/restoration, Gibbs priors with quadratic potential functions have been used because they allow simple mathematical formulations. However, these priors lead to smooth solutions with a poor representation of transitions. To overcome this problem, several authors have proposed other potential functions, non quadratic, that deal better with the transitions, but lead to an increase of complexity. In this paper, we come back to the Gibbs priors with quadratic potential functions to show that MAP algorithms can be formulated as a linear filtering problem. The recursive nature of the filtering methods allows us to obtain very fast and efficient restoration/reconstruction MAP algorithms. Furthermore, we show that by using quadratic potential functions involving high order differences between neighbors it is possible to significantly improve the solution at the transitions. In fact, with this strategy we are increasing the order of the filter that models the original MAP problem. The approach presented in this paper can be used as a method to design Gibbs priors based on the filter design theory.

The problem of non-linear dimension reduction is relevant to many different areas of knowledge. While the linear case can be solved by variations of PCA, the non-linear case is more complex. Recent advances incorporate geometrical information by estimating a manifold that approximates the data. The paper follows that trend and tackles some limitations of existing approaches -arbitrary topology and curvature of the manifold, unknown intrinsic dimension and, for mixture models, unknown number of mixture components. An algorithm, designated TBA (tangent bundle approximation), is presented that addresses the enumerated difficulties and is faster than existing methods, in datasets of a few thousand points. The motivation behind TBA is to perform motion tracking in video sequences, but the algorithm can be applied in a wide class of problems. The paper starts with a brief review of related work and then describes the TBA approach in detail. The algorithm is then subjected to comparative evaluation.

It was recently proposed an object tracking method, which is able to deal with object occlusions and group tracking, using Bayesian networks. The Bayesian network (BN) tracker has shown promising results in difficult situations but its architecture is limited to a maximum of 2 parents/2 children per node, in order to avoid the combinatorial explosion and difficult network generation procedures from the video signal. This paper addresses the major limitation of the BN tracker and presents a method to generalize the tracker to cope with arbitrary topologies, allowing the tracker to operate in more complex scenes.

This paper describes a new algorithm for the recognition of human activities. These activities are modelled using banks of switched dynamical models, each of which is tailored to a specific motion regime. Furthermore, it is assumed that model switching happens according to a space-dependent Markov chain, i.e., some transitions are more probable in specific regions of the image. Space dependence allows the model to represent the interaction between the person and static elements of the scene. The paper describes learning algorithms for space-dependent switched dynamical models and presents experimental results with synthetic and real data.

The paper proposes a Bayesian 3D ultrasound reconstruction/estimation from non-uniform ultrasound observations, non-Gaussian data fidelity term, and total variation (TV) based prior. To compute the maximum a posteriori (MAP) solution, we introduce a generalized expectation maximization (GEM) algorithm, which converges to the exact MAP solution in the case of convex data fidelity term. A set of experiments illustrates the effectiveness of the method.

This paper describes an algorithm for the extraction of multiple regions using multiple active contour models (ACMs). The algorithm organizes edge points into strokes and assigns a set of weights summing to one to each stroke. These weights represent the soft assignment of the stroke to each of the ACMs and depend on the distance between the stroke points and the ACM units. Both the weights and the ACMs energy minimization are computed using the expectation-maximization (EM) algorithm. The algorithm described in this paper is an extension of the adaptive snakes recently proposed in Nascimento, JC and Marques, JS (2003). Experimental results are provided to illustrate the performance of the proposed algorithm.

There has been a surge of interest in learning non-linear manifold models to approximate high-dimensional data. Both for computational complexity reasons and for generalization capability, sparsity is a desired feature in such models. This usually means dimensionality reduction, which naturally implies estimating the intrinsic dimension, but it can also mean selecting a subset of the data to use as landmarks, which is especially important because many existing algorithms have quadratic complexity in the number of observations. This paper presents an algorithm for selecting landmarks, based on LASSO regression, which is well known to favor sparse approximations because it uses regularization with an l1 norm. As an added benefit, a continuous manifold parameterization, based on the landmarks, is also found. Experimental results with synthetic and real data illustrate the algorithm.

Deformable models (e.g., snakes) perform poorly in many image analysis problems. The contour model is attracted by edge points detected in the image. However, many edge points do not belong to the object contour, preventing the active contour from converging toward the object boundary. A new algorithm is proposed in this paper to overcome this difficulty. The algorithm is based on two key ideas. First, edge points are associated in strokes. Second, each stroke is classified as valid (inlier) or invalid (outlier) and a confidence degree is associated to each stroke. The expectation maximization algorithm is used to update the confidence degrees and to estimate the object contour. It is shown that this is equivalent to the use of an adaptive potential function which varies during the optimization process. Valid strokes receive high confidence degrees while confidence degrees of invalid strokes tend to zero during the optimization process. Experimental results are presented to illustrate the performance of the proposed algorithm in the presence of clutter, showing a remarkable robustness.

Many object-tracking algorithms are based on low-level features detected in the image. Typically, the object shape and position are estimated to fit the observed features. Unfortunately, image analysis methods often produce invalid features (outliers) which do not belong to the object boundary. These features have a strong influence on the shape estimates, leading to meaningless tracking results. This paper proposes a robust tracking algorithm which is able to deal with outliers, inspired in the probabilistic data association filter proposed in the context of point tracking. The algorithm is based on two key concepts. First, middle level features (strokes) are used instead of low-level ones (edge points). Second, two labels (valid/invalid) are considered for each stroke. Since the stroke labels are unknown all labeling sequences are considered and a probability (confidence degree) is assigned to each of them. In this way, all the strokes contribute to track the moving object but with different weights. This allows a robust performance of the tracker in the presence of outliers. Experimental tests are provided to assess the performance of the proposed algorithm in lip and gesture tracking and surveillance applications.

This paper presents an algorithm for tracking the left ventricle in echocardiographic sequences, using multiple models. The use of multiple dynamic models is appropriate since the heart motion presents two phases (diastole and systole) with different dynamics. The main difficulty concerns the low contrast and speckle noise present in ultrasound images. To overcome this problem a robust multiple model tracker is used, based on a bank of nonlinear filters, organized in a tree structure. This algorithm determines which model is active at each instant of time and updates its state by propagating the probability distribution, using robust estimation techniques. It is shown in the paper that the proposed algorithm simultaneously copes with several dynamic models and with outliers. Furthermore the proposed algorithm provides high level information that is not available when a single model is used.

This paper addresses long term tracking of multiple objects with occlusions. Bayesian networks are used to model the interaction among the detected tracks and for conflict management, allowing the tracker to update the labelling decisions as soon as new information is available. If several objects overlap in the image domain and then become separated in the next frames, the proposed algorithm is able to accumulate the evidence extracted from the images and to disambiguate the competing labels. The system also provides a confidence degree for each labelling decision. Experimental results are provided to illustrate the performance of the proposed method for long term tracking of multiple pedestrians.

Bayesian reconstruction methods are slow and have been avoided in 3D ultrasound, since fast algorithms are required. The paper tries to overcome this difficulty by presenting a fast reconstruction algorithm designed in a Bayesian framework, based on data dependent IIR filtering. Thus the benefits of Bayesian volume reconstruction and the computational efficiency of IIR recursive filtering may be obtained. The proposed filter has space varying coefficients which depend on the observed data in each region. The proposed method is evaluated using synthetic and real data. The results are compared with the MAP estimates obtained using a nonlinear optimization method. The proposed IIR recursive filtering algorithm achieves comparable results with the MAP algorithm and is twenty five times faster.

Abstract This paper describes an algorithm for shape estimation in cluttered scenes. A new image potential is defined based on strokes detected in the image. The motivation is simple. Feature detectors (e.g. edge points detectors) produce many outliers, which hamper the performance of boundary extraction algorithms. To overcome this difficulty we organize edges in strokes and assign a confidence degree (weight) to each stroke. The confidence degrees depend on the distance of the stroke points to the boundary estimates and they are updated during the estimation process. A deformable model is used to estimate the object boundary, based on the minimization of an adaptive potential function which depends on the confidence degree assigned to each stroke. Therefore, the image potential changes during the estimation process. Both steps (weight update, energy minimization) are derived as the solution of a maximum likelihood estimation problem using the EM algorithm. Experimental tests are provided to illustrate the performance of the proposed algorithm.

This paper addresses 3D reconstruction in the presence of multiplicative noise and non-linear compression of the ultrasound data. Ultrasound images are often considered as being corrupted by multiplicative noise with Rayleigh distribution. However, commercial ultrasound equipment also performs a non-linear image compression which reduces the dynamic range of the ultrasound signal, for visualization purposes. In this paper the non-linear compression is explicitly modelled and considered for 3D reconstruction of medical data. The proposed algorithm is tested with synthetic and real data and the results are discussed.

Kalman filtering has been extensively used in object tracking. However, the tracker performance is severely affected in the presence of multiple objects and cluttered background. The reason is simple. Feature detection produces many outliers and the Kalman filter is not able to discriminate valid data from the clutter. This paper overcome this difficulty and describes a robust algorithm for object tracking denoted as S-PDAF (shape-probabilistic data association filter). Experimental tests show that significant robustness improvement is achieved by the S-PDAF algorithm.

This paper addresses 3D reconstruction of human organs from a set of ultrasound images, assuming that the probe position and orientation is available. Unfortunately, pose measurement errors produce significant misalignments which degrade the performance of the reconstruction algorithms. This paper proposes a method for correcting these distortions by estimating the true position and orientation of the ultrasound probe. This is achieved by a joint computation of the 3D data and probe parameters using a MAP criterion. An evaluation of the proposed algorithm, by Monte Carlo method is included in the paper.

This paper addresses the problem of tracking of objects with complex shape or motion dynamics. The approach followed relies on multiple models based on Gaussian mixtures and hidden Markov models. A tracking algorithm derived from nonlinear filtering is presented and illustrated in two situations. In the first, two points moving independently along a line are tracked, only one being observed at each time. In the second, two-dimensional objects are tracked, under severe shape deformations. Unlike other multi-model approaches, the proposed method relies on parametric techniques providing an efficient tool to update shape and motion estimates.

This article addresses the problem of tracking moving objects using deformable models. A Kalman-based algorithm is presented, inspired by a new class of constrained clustering methods, proposed by Abrantes and Marques (1996) in the context of static shape estimation. A set of data centroids is tracked using intra-frame and inter-frame recursions. Centroids are computed as weighted sums of the edge points belonging to the object boundary. The use of centroids introduces competitive learning mechanisms in the tracking algorithm leading to improved robustness with respect to occlusion and contour sliding. Experimental results with traffic sequences are provided.

Object tracking based on multipole models has been previously advocated as a way to tackle sudden changes of shape or motion parameters. The paper addresses the estimation of time-varying parameters described by a bank of stochastic models switched according to a probabilistic mechanism (Markov chain). A state estimation algorithm is proposed, based on the propagation of Gaussian mixtures in a multi-model framework.

The parameters to be estimated in dynamical image analysis are usually restricted to a subset of the Euclidean space. This constraint is often neglected in the design of image analysis algorithms although it allows one to achieve significant improvements. This paper describes an algorithm for the estimation of parameter evolution on manifolds exploiting three sources of information: the manifold geometry, the motion model and the sensor model. Examples are provided to show that the proposed algorithm outperforms classic estimation methods in these problems.

This paper describes a method for the segmentation of dynamic data. It extends well known algorithms developed in the context of static clustering (e.g., the c-means algorithm, Kohonen maps, elastic nets and fuzzy c-means). The work is based on an unified framework for constrained clustering recently proposed by the authors in [1]. This framework is extended by using a motion model for the clusters which includes global and local evolution of the data centroids. A noise model is also proposed to increase the robustness of the dynamic clustering algorithm with respect to outliers. 1 Introduction Many clustering algorithms try to approximate a set of data vectors # ### # # #### ## # by a smaller set of prototypes # ### # # #### ## #. The main problem consists of computing the number of prototypes and their locations in order to obtain the best fit, according to a specific criterion. This paper addresses dynamic clustering problems in which data is gathered during a time interva...

Boundary extraction is a data representation problem: image features are segmented and approximated by a parametric curve or a sequence of model points. However, the use of classic Pattern Recognition methods in bound-ary detection is unusual when compared with more popular approaches, e.g., active contours. This can be partially explained by their inability to separate boundary edges from other image strokes. This paper presents modified ver-sions of several clustering and neural networks algorithms (c-means, fuzzy c-means, Kohonen maps, elastic nets), enhanced with dynamic data segmen-tation capabilities. This is achieved by using a noise model. The noise model consists of a virtual unit equidistant of all data points, which can be geomet-rically interpreted as a noise plane parallel to the image plane. The proposed technique extends the unified framework recently proposed by Abrantes and Marques [1] in the context of edge linking with constrained clustering tech-niques. Results are provided in the paper to illustrate the segmentation capa-bility of the novel methods in the analysis of images with undesired strokes (e.g., inside the object). It is concluded that the noise model proposed in this paper allows a widespread use of classic constrained clustering algorithms in the context of shape analysis. 1

Deformable models are related to other data representation methods. It was recently proposed a class of models based on a fuzzy energy function which includes many well known algorithms (snakes, elastic nets, fuzzy and hard c-means and Kohonen maps). This paper describes a probabilistic extension of these algorithms in a Bayesian framework, using Gibbs-Boltzman distributions. It is shown that the new class of models minimizes an energy function with an additional term: the log partition function. The role of the log partition function in probabilistic versions of snakes, c-means and elastic nets is studied and analytic expressions are derived in the case of probabilistic snakes. The log partition function produces an additional force field which improves the performance of these algorithms in some applications.

Deformable models have been intensively investigated during the last decade. Several well known algorithms, proposed in other contexts can also be included in this class (e.g., Kohonen maps, elastic nets and fuzzy c-means). In all these methods the model parameters are obtained in a deterministic framework by the minimization of an energy function. This paper proposes a novel class of probabilistic shape models related to the unified framework presented by Abrantes and Marques (see IEEE Trans. Image Processing, p.1507-21, 1996). Shape modelling is addressed as a MAP estimation problem, by assuming that the image features are random variables with Gibbs-Boltzmann distribution, and provides extensions for several well known algorithms. The main difference between the proposed algorithms and the original ones lies on the partition function which depends on the model parameters and influences the shape estimates. For example, it is shown that in snakes the partition function generates short-range repulsive forces between the model units which prevent their collapse when they are attracted by common data.

This paper addresses the problem of object recognition based on contour descriptions. Two approaches, namely hidden Markov models (HMM) and syntactic modeling based on stochastic finite-state grammars (SFSG), are analyzed and applied to the classification of hardware tools. It is shown that both approaches are able to capture the data variability, leading to high classification performance. While the syntactic paradigm is flexible, the structure of the grammars being automatically inferred from the data, the HMMs are more robust in terms of training data sets requirements.

Boundary extraction is a key task in many image analysis operations. This paper describes a class of constrained clustering algorithms for object boundary extraction that includes several well-known algorithms proposed in different fields (deformable models, constrained clustering, data ordering, and traveling salesman problems). The algorithms belonging to this class are obtained by the minimization of a cost function with two terms: a quadratic regularization term and an image-dependent term defined by a set of weighting functions. The minimization of the cost function is achieved by lowpass filtering the previous model shape and by attracting the model units toward the centroids of their attraction regions. To define a new algorithm belonging to this class, the user has to specify a regularization matrix and a set of weighting functions that control the attraction of the model units toward the data. The usefulness of this approach is twofold: it provides a unified framework for many existing algorithms in pattern recognition and deformable models, and allows the design of new recursive schemes.

Snakes, elastic nets and Kohonen networks are well known algorithms which were developed in different contexts. However, these algorithms share common features allowing one to ask what is the relationship among them and suggesting their use in problems which have traditionally been tackled by only one of them. The paper addresses the problem of edge linking and proposes a new class of nonlinear recursive algorithms, based on a general cost function, which includes snakes, Kohonen maps, and elastic nets as special cases. This class provides an unified framework for several existing algorithms in pattern recognition and active contours and allows the design of new recursive schemes.

A class of edge linking algorithms with a common cost function was proposed by the authors. The cost function contains a quadratic regularization term and an image dependent term defined by a set of weighting functions. To define a new algorithm belonging to this class, the user has to specify a regularization matrix and a set of weighting functions which control the attraction of the model units towards the data. This paper compares the structure of the weighting functions associated with several well known algorithms which can be included in this framework (snakes, Kohonen (1982) maps, elastic nets, fuzzy c-means). This comparison allows a better understanding of their performance in edge linking operations. The paper also illustrates the design of new recursive schemes and describes techniques used to improve the convergence rate by more than one order of magnitude.

In the analysis of quasi-periodic textured images, good estimates of the periodicity parameters art instrumental to obtain an accurate evaluation of the texel characteristics. This paper presents a maximum likelihood algorithm for the estimation of periodicity parameters from the peaks of the localized Fourier spectrum. To derive the ML estimator, a stochastic model of the spectral peaks is required. Therefore, a probability model which incorporates false alarms, detection errors as well as random jitter in frequency estimates is described in this paper. The relationship between the ML algorithm proposed in this paper and the Hough transform is addressed. Finally, experimental tests with synthetic and natural images are presented to assess the performance of the ML estimator.>

The authors present an algorithm for the generation of frequency-modulation sinusoids (or complex exponentials), which avoids the use of trigonometric or exponential functions. It is shown that a sinusoid (exponential) with a polynomial phase of an arbitrary order N can be recursively generated with N complex multiplications per sample. This result extends a previous algorithm for the generation of complex Gaussian functions and it is derived in a more concise manner. The application of this novel technique in the context of harmonic coding of speech is also briefly described. The proposed technique allows significant computational savings and is therefore important for real-time implementation of harmonic coding algorithms.>

A harmonic coder designed for operation at 4.8 kb/s is presented. Sinusoidal modeling of speech is reviewed and the use of harmonically related frequencies is extended to unvoiced and transition regions. The estimation of the fundamental frequency with emphasis on the fine tuning of the pitch estimates is discussed. The structure of the 4.8-kbit/s harmonic coder is described, and a detailed discussion of phase quantization issues is presented. Experimental results are presented which reveal a reverberant character in many speech utterances and difficulties in some classes of sounds (plosives, liquids, and voiced fricatives). These distortions are mostly produced by the quantization of phase, but they are also explained by the difficulties of the sinusoidal analysis/synthesis in transition regions when the frame length is large.>

A scheme is discussed for long-term prediction in CELP (code-excited linear predictive) coding using fractional delay prediction. This technique permits a more accurate representation of voiced speech and achieves an improvement of synthetic quality for female speakers. The higher complexity of this type of predictor relative to the classical one is its major disadvantage. Suboptimal schemes in which the search for the functional pitch delay is restricted to a neighborhood of an integer pitch estimate can be envisaged to decrease the computational load.>

A systematic study of code excited linear predictive (CELP) coders is presented. The design of an optimized version complying with the specifications for a GSM half-rate coder is described. This study is divided into three parts: election of an unquantized configuration, assuming unquantized scale factors and filter coefficients, fine tuning of the parameter quantization, and adaptation to the GSM specifications. A strong emphasis is placed on efficient codebook search procedures, and two approaches are briefly described: the truncated autocorrelation approach and the unity-magnitude approach. The final coder version achieves a very good combination of speech quality and implementation complexity.>

The introduction of vector quantization in the harmonic coder framework can affect several sets of parameters, namely the amplitudes and the phases of the harmonic coefficients. A systematic experimental study is reported in which this type of quantization is compared with traditional scalar techniques for quantizing the harmonic parameters. The experiments showed that it is possible to quantize voice speech below 8 kb/s and still obtain high-quality synthetic signals.>

Sinusoid based models have been used in the last few years for high quality representation of voiced speech. In this paper we first review a theoretical basis for this kind of representation, and then introduce a spectral model for varying-frequency sinusoids. This model is used for estimating the sinusoid parameters end it is also intended for accurate representation of the spectrum of voiced speech. A time warping procedure is presented, which permits to reduce the overlap between high frequency spectral lines, with the purpose of improving their estimation.