Search

This paper evaluates the effectiveness of keypoint methods for content-based protection of digital images. These methods identify a set of "distinctive" regions (termed keypoints) in an image and encode them using descriptors that are robust to expected image transformations. To determine whether particular images were derived from a protected image, the keypoints for both images are generated and their descriptors matched. We describe a comprehensive set of experiments to examine how keypoint methods cope with three real-world challenges: (1) loss of keypoints due to cropping; (2) matching failures caused by approximate nearest-neighbor indexing schemes; (3) degraded descriptors due to significant image distortions. While keypoint methods perform very well in general, this paper identifies cases where the accuracy of such methods degrades.

The existing halftone image watermarking methods were proposed to embed a watermark bit in a halftone dot, which corresponds to a pixel, to generate stego halftone image. This one-to-one mapping, however, is not consistent with the one-to-many strategy that is used by current high-resolution devices, such as computer printers and screens, where one pixel is first expanded into many dots and then a halftoning processing is employed to generate a halftone image. Furthermore, electronic paper or smart paper that produces high-resolution digital files cannot be protected by the traditional halftone watermarking methods. In view of these facts, we present a high-resolution halftone watermarking scheme to deal with the aforementioned problems. The characteristics of our scheme include: (i) a high-resolution halftoning process that employs a one-to-many mapping strategy is proposed; (ii) a many-to-one inverse halftoning process is proposed to generate gray-scale images of good quality; and (iii) halftone image watermarking can be directly conducted on gray-scale instead of halftone images to achieve better robustness

The recognition of events in video data is a subject of much current interest. In this paper, we address several issues related to this topic. The first one is overfitting when very large feature spaces are used and relatively small amounts of training data are available. The second is the use of a framework that can recognise events at different time scales, as standard hidden Markov model (HMM) do not model well long-term term temporal dependencies in the data. In this paper we propose a method combining layered HMMs and an unsupervised low level clustering of the features to address these issues. Experiments conducted on the recognition task of different events in 7 rugby games demonstrates the potential of our approach with respect to standard HMM techniques coupled with a feature size reduction technique. While the current focus of this work is on events in sports videos, we believe the techniques shown here are general enough to be applied to other sources of data

VoIP applications require the ability to identify speakers in real time. This paper presents compressed speaker recognition (CSR), an innovative approach to perform speaker recognition directly from the compressed voice packets. CSR performs online speaker recognition from live packet streams of compressed voice packets by performing fast clustering over a defined subset of the features available in each compressed voice packet. Our experimental results show that CSR is highly scalable and accurate across a broad range of speakers

Modeling visual concepts using supervised or unsupervised machine learning approaches are becoming increasing important for video semantic indexing, retrieval, and filtering applications. Naturally, videos include multimodality data such as audio, speech, visual and text, which are combined to infer therein the overall semantic concepts. However, in the literature, most researches were conducted within only one single domain. In this paper we propose an unsupervised technique that builds context-independent keyword lists for desired visual concept modeling using WordNet. Furthermore, we propose an extended speech-based visual concept (ESVC) model to reorder and extend the above keyword lists by supervised learning based on multimodality annotation. Experimental results show that the context-independent models can achieve comparable performance compared to conventional supervised learning algorithms, and the ESVC model achieves about 53% and 28.4% improvement in two testing subsets of the TRECVID 2003 corpus over a state-of-the-art speech-based video concept detection algorithm

In this paper, we present a scalable (i.e. lossy-to-lossless) watermark scheme based on a recently standardized scalable audio coder-AAZ (R.S. Yu, et al., 2004). The proposed framework enables the recovery of the original lossless audio after watermark embedding, and in the meanwhile, is able to make the watermark adaptive such that the watermark distortion to the lossy host audio is minimized. An encryption mechanism is further employed for restricting unauthorized access to lossless audio and watermark removal. Based on this framework, we elaborate its possible applications on high-quality audio archiving and streaming. Experimental results demonstrate the validity of our proposal.

This paper proposes a practical content-dependent low-power DCT design with tolerable quality drop. Low-power issue has become more and more important, especially for portable devices. Unfortunately, low-power design always brings in significant quality drop at the same time. This work not only achieves ultra low power dissipation but also remains tolerable quality drop (about 0.1 dB) in most cases. The proposed architecture is based on distributed arithmetic architecture [A. Peled et. al., (1974)] and combined with more reliable PPA classification algorithm [T. Xanthopoulos et. al., (2000)]. It can accurately control bit-level calculation and avoid unnecessary calculation to save power. This characteristic is powerful and very useful in video encoder systems, since coefficients after DCT and Q become zero with high probability. This part of power can be saved without causing undesired quality drop

Bitstream binding language (BBL) provides an abstraction layer between XML multimedia containers and the way their resources and metadata are published in a bitstream. It allows multiple bindings from a single source document to facilitate interoperability and applicability of the multimedia content to a wide range of terminals and users. BBL introduces a number of features not found in other XML fragmentation techniques-such as support for random-access-and provides a highly general framework for combining metadata with multimedia content in a transport format. This paper presents an overview of BBL and provides details of a Java implementation using the MPEG-21 multimedia framework as a sample application

Continuing strides in processing, storage, sensing, and networking technologies are enabling people to capture their activities and experiences as greater volumes of ever-richer media. A big emerging challenge today is the organization, retrieval, and exploitation of such multimedia data surrounding the activities of individuals or enterprises. The field of multimedia electronic chronicles deals with the unified contextual organization, presentation, and analysis of temporal streams of multimedia data captured by individuals, groups, or organizations. The value of electronic chronicles is in converting activity and experience from the past into actionable intelligence in the present. Such multimedia electronic chronicles, with their associated techniques for search and navigation, analysis and reasoning, and prediction and alerting, will have enormous impact on various spheres of life spanning enhancement of personal life, business productivity, entertainment, and government operations. This paper, which serves as a companion to a keynote talk by the first author at ICME 2005, explains the notion of electronic chronicles and outlines the research challenges in this area

Bilateral filtering is an edge-preserving filtering technique that employs both geometric closeness and photometric similarity of neighboring pixels to construct its filter kernel. Multi-dimensional bilateral filtering is computationally expensive because the adaptive kernel has to be recomputed at every pixel. In this paper, we present a separable implementation of the bilateral filter. The separable implementation offers equivalent adaptive filtering capability at a fraction of execution time compared to the traditional filter. Because of this efficiency, the separable bilateral filter can be used for fast preprocessing of images and videos. Experiments show that better image quality and higher compression efficiency is achievable if the original video is preprocessed with the separable bilateral filter.

Automatic semantic classification of image databases is very useful for users' searching and browsing, but it is at the same time a very challenging research problem as well. In this paper, we develop a hidden semantic concept discovery methodology to address effective semantics-intensive image database classification. Each image is segmented into regions and then a uniform and sparse region-based representation is obtained. With this representation a probabilistic model based on statistical-hidden-class assumptions of the image database is proposed, to which the Expectation-Maximization (EM) technique is applied to analyze semantic concepts hidden in the database. Two methods are proposed to make use of the semantic concepts discovered from the probabilistic model for unsupervised and supervised image database classifications, respectively, based on the automatically learned concept vectors. It is shown that the concept vectors are more reliable and robust and thus promising than the low level features through the theoretic analysis and the experimental evaluations on a database of 10,000 general-purpose images.

This paper presents techniques in clustering the same-topic news stories according to event themes. We model the relationship of stories with textual and visual concepts under the representation of bipartite graph. The textual and visual concepts are extracted respectively from speech transcripts and keyframes. Co-clustering algorithm is employed to exploit the duality of stories and textual-visual concepts based on spectral graph partitioning. Experimental results on TRECVID-2004 corpus show that the co-clustering of news stories with textual-visual concepts is significantly better than the co-clustering with either textual or visual concept alone

Surviving geometric attacks in image authentication is considered to be of great importance. This is because of the vulnerability of classical watermarking and digital signature based schemes to geometric image manipulations, particularly local geometric attacks. In this paper, we present a general framework for image content authentication using salient feature points. We first develop an iterative feature detector based on an explicit modeling of the human visual system. Then, we compare features from two images by developing a generalized Hausdorff distance measure. The use of such a distance measure is crucial to the robustness of the scheme, and accounts for feature detector failure or occlusion, which previously proposed methods do not address. The proposed algorithm withstands standard benchmark (e.g. Stirmark) attacks including compression, common signal processing operations, global as well as local geometric transformations, and even hard to model distortions such as print and scan. Content changing (malicious) manipulations of image data are also accurately detected

As the VLSI technology advances continuously, ASIC can easily achieve the required performance and most of them are actually over-designed. Thus, architecture shrinking is inevitable in optimal designs especially when supply voltages are getting lower. However, conventional designs starting from minimization of algorithmic operations (e.g. multiply) may not always lead to optimal architectures, for the wires and the interconnection complexity significantly grow and have become predominant. This paper explores algorithms and architectures for the 2-D transform in H.264/AVC, of which the operations are very simple (i.e. only shift and add). We have shown that fewer operations do not always result in more compact designs. In our experiments with the UMC 0.18 /spl mu/m CMOS technology, the most straightforward matrix multiplication without separable 2-D operation or any fast algorithm has the best area efficiency for DI-size (720/spl times/480) video at 30 fps. It saves 48%, 34%, and 16% silicon area of the previous works respectively.

The development of mid-level concepts helps to bridge the gap between low-level feature and high-level semantics in video analysis. Most existing work combines the customized mid-level concepts and statistical models to detect particular events. Based on broadcast sports video production knowledge, we extend our previous work to present a unified framework for mid-level concept generation in this paper. A video segment is characterized via three essential aspects: camera shot size, an object appearing in a scene, and video production technology. These three aspects clearly summarize the primary concerns in terms of a generic concept generation. Within this framework, we can flexibly and clearly define meaningful mid-level concepts towards comprehensive video content analysis, such as replay classification and the detection of events (e.g. goal, shoot, attack, foul, offside, and out of bound, etc.)

In the last decade, the research topic of automatic analysis of facial expressions has become a central topic in machine vision research. Nonetheless, there is a glaring lack of a comprehensive, readily accessible reference set of face images that could be used as a basis for benchmarks for efforts in the field. This lack of easily accessible, suitable, common testing resource forms the major impediment to comparing and extending the issues concerned with automatic facial expression analysis. In this paper, we discuss a number of issues that make the problem of creating a benchmark facial expression database difficult. We then present the MMI facial expression database, which includes more than 1500 samples of both static images and image sequences of faces in frontal and in profile view displaying various expressions of emotion, single and multiple facial muscle activation. It has been built as a Web-based direct-manipulation application, allowing easy access and easy search of the available images. This database represents the most comprehensive reference set of images for studies on facial expression analysis to date.

This paper addresses the problem of optimizing the playback delay experienced by a population of heterogeneous clients, in video streaming applications. We consider a typical broadcast scenario, where clients subscribe to different portions of a scalable video stream, depending on their capabilities. Clients share a common channel, whose limited rate directly drives the playback delays imposed to the different groups of receivers. We derive an optimization problem, that targets a fair distribution of the playback delays among heterogeneous clients. A server-based scheduling strategy is then proposed, that takes into account the properties of the targeted clients, the channel status, and the structure of the media encoding. It is shown to offer significantly reduced playback delays per client population, as compared to traditional scheduling strategies. In the same time, PSNR performance is not affected, which altogether leads to an overall improvement of the quality of service

The transmission of multimedia streams with a particular and guaranteed quality of service (QoS) is a major challenge in multimedia communication over IP networks. In this paper, we analyze the distributed architecture of a video on demand (VoD) system to design global request handling and admission control strategies. We describe a highly available distributed VoD service using existing telephone network, which is inherently scalable and fault tolerant. The VoD service is provided by media server farms (MSF), which may reside in different local telephone exchange of a telephone network. We present the design and simulation results for a new distributed admission control and QoS adaptation controller architecture, called semi-distributed admission control (SD-AC) architecture for a set of MSFs

Periodic broadcasting (PB) schemes are the most promising solution for building large-scale video streaming services. Existing PB schemes are all built around the traditional client-server model. This paper proposes a radically different architecture where only end-user hosts are used to build the PB system, thereby eliminating the need for a central server or even a service provider. Two unique problems pertaining to this decentralized, peer-to-peer architecture is addressed: transmission scheduling and peer synchronization. A new decentralized synchronization algorithm is proposed and results obtained from simulations confirm the architecture's feasibility and performance.

Periodic broadcasting is known as an efficient technique for delivering popular videos by reducing bandwidth requirement for transmitting streaming video to simultaneous viewers. The channel transition problem is a noticed issue to be concerned about the variability of popularity of video and resource management in periodic broadcasting. In this paper, we approach a general seamless channel transition scheme for pyramid-based broadcasting protocols and present the stairway channel transition (SWCT) scheme. Compared to the existing schemes that are only designed for dedicated broadcasting protocols, our design possesses more flexibility, while it does not reduce the performance of the original protocols.

Recent technology advances in multimedia communication have ushered in a new era of personal communication. Users can ubiquitously access the Internet via various mobile devices. For the mobile devices featured with lower-bandwidth network connectivity, transcoding can be used to reduce the object size by lowering the quality of a multimedia object. In this paper, we focus on the cache replacement policy in a transcoding proxy, which is a proxy server responsible for transcoding the object and reducing the network traffic. Based on the architecture in prior works, we propose a maximum profit replacement algorithm, abbreviated as MPR. MPR performs cache replacement according to the content in the caching candidate set, which is generated by the concept of dynamic programming. Experimental results show that the the proposed MPR outperforms the prior scheme in terms of the cache hit ratio.

Training a good text detector requires a large amount of labeled data, which can be very expensive to obtain. Co-training has been shown to be a powerful semi-supervised learning tool for solving many problems using a large amount of unlabeled data. However, augmented data from a co-training process could potentially degrade the performance of classifiers due to added noises from unlabeled data. This paper makes two contributions by proposing a modified co-training scheme for text detection. First, to get cleaner augmented data, the new algorithm integrates some authority knowledge of unlabeled data into co-training. Text recognition output of each selected unlabeled image patch is used as the authority that is combined with classifier prediction to decide if the sample will be added to the augmented set. Second, instead of evenly combining predictions of two co-training classifiers, a weighted combination is learned and used to produce the final prediction. Contributions of the new algorithm have been evaluated on a standard text detection dataset.

Since multimedia information is characterized by motley types of media with different properties, multimedia content retrieval in digital libraries requires dynamic reconfigurable architectures. In this paper, mobile agents are being used as the basis of such a dynamic reconfigurable architecture. The proposed architecture is further enhanced with innovative relevance feedback algorithms for interactive multimedia retrieval. In particular, an optimal recursive relevance feedback mechanism is examined by adaptively estimating a parametric correlation-based similarity measure

In this paper an optimized and efficient technique for keyframes extraction of video sequences is proposed, which leads to selection of a meaningful set of video frames for each given shot. Initially for each frame, the singular value decomposition method is applied and a diagonal matrix is produced, containing the singular values of the frame. Afterwards, a feature vector is created for each frame, by gathering the respective singular values. Next, all feature vectors of the shot are collected to form the feature vectors basin of this shot. Finally, a genetic algorithm approach is proposed and applied to the vectors basin, for locating frames of minimally correlated feature vectors, which are selected as keyframes. Experimental results indicate the promising performance of the proposed scheme on real life video shots

This paper investigates the use of unlabeled data to help labeled data for audio-visual event recognition in meetings. To deal with situations in which it is difficult to collect enough labeled data to capture event characteristics, but collecting a large amount of unlabeled data is easy, we present a semi-supervised framework using HMM adaptation techniques. Instead of directly training one model for each event, we first train a well-estimated general event model for all events using both labeled and unlabeled data, and then adapt the general model to each specific event model using its own labeled data. We illustrate the proposed approach with a set of eight audio-visual events defined in meetings. Experiments and comparison with the fully-supervised baseline method show the validity of the proposed semi-supervised approach.

Rotational motion can often be seen in video. However, comparatively little research has been done to investigate rotational motions in video, whose analysis could be useful. For example, if we can efficiently identify the rotation center of a spinning object, extraction and tracking of it can be made easier by grouping points moving at the same radial speed. It could also improve compression by synthesizing analyzed spin transitions, and help tracking of rotating objects. In this paper, we introduce a set of rotation center location methods using only the motion field constructed during video encoding, along with a few methods for improving their performances. These methods can be implemented using integer operations only. They are up to 1.81 times faster than the traditional circulation analysis method with little sacrifice in accuracy, and are not affected by asymmetric fields caused by translational motions.

Text documents, in electronic and hardcopy forms, are and will probably remain the most widely used kind of content in our digital age. The goal of this paper is to overview protocols for text data-hiding based "smart documents", achieving document self-authentication, self-recovery, self-annotation and automatic processing. We argue that document security, recovery and embedded annotation are the most promising data-hiding based frameworks.

This paper presents a new routing strategy, that selects the best network paths in an overlay network, in order to minimize the distortion perceived by the end user. We first propose a model that reports the video distortion as a function of the encoding rate, and the loss process parameters (the packet loss ratio, and the average burst length of errors). We then derive a method to compute an accurate estimation of the end-to-end characteristics of a given path in the network topology, as experienced by the media stream. It allows for reducing a set of streaming paths between the server and the client, to a simple virtual link with equivalent end-to-end parameters. Finally, an algorithm is proposed, that finds the best streaming paths in the overlay network, in terms of media distortion. Our algorithm therefore takes into account not only the conventional network parameters (e.g. end-to-end available bandwidth), but also other application specific metrics (e.g. video distortion measure and loss process). Interestingly, it is shown that the best route in terms of video distortion, does not necessarily use the highest bandwidth links, but carefully trades off channel reliability and bandwidth.

In this paper, an HMM-embedded unsupervised learning approach is proposed to detect the music events by grouping the similar segments of the music signal. This approach can cluster the segments based on their similarity of the spectral as well as the temporal structures. This is not easily done for clustering with the traditional similarity measures. Together with a Bayesian information criterion, the proposed approach can obtain a suitable event set to regularize the complexity of the model structure. The natural product of the approach is a set of music events modeled by the HMMs. Our experimental analyses show that the detected musical events have more perceptual meaning and are more consistent than the KL-distance based clustering. The learned events match better with our experience in spectrogram reading. Its capacity is further evaluated on a task of music identification. The identification error rate is reduced to 1.57%, and 56.3% relative error rate reduction is observed comparing with the system trained using the KL-distance clustering method

This paper presents an experimental implementation of a low-complexity speaker recognition algorithm working in the compressed speech domain. The goal is to perform speaker modeling and identification without decoding the speech bitstream to extract speaker dependent features, thus saving important system resources, for instance, in mobile devices. The compressed bitstream values of the widely used GSM AMR speech coding standard are studied to identify statistics enabling fair recognition after a few seconds of speech. Using Euclidean distance measures on elementary statistical values such as coefficient of variation and skewness of nine standard GSM AMR parameters delivers recognition accuracies close to 100% after about 20 seconds of active speech for a database of 14 speakers recorded in a normal room environment.

Parametric watermarking is effected by modifying the linear predictor coefficients of speech. In this work, the parameter noise is analyzed when watermarked speech is subjected to additive white and colored noise in the time domain. The paper presents two detection techniques for parametric watermarking. The first approach uses the Neyman-Pearson criterion to solve a binary decision problem. In the second approach, discriminant functions based on the minimum-error-rate criterion are used to determine which one of the many watermarks was embedded or if no watermark is present. Experiments with speech data are used to determine the false-alarm and missed detection rates

In this paper, spherical harmonics are proposed as shape descriptors for 2D images. We introduce the concept of connectivity; 2D images are decomposed using connectivity, which is followed by 3D model construction. Spherical harmonics are obtained for 3D models and used as descriptors for the underlying 2D shapes. Difference between two images is computed as the Euclidean distance between their spherical harmonics descriptors. Experiments are performed to test the effectiveness of spherical harmonics for retrieval of 2D images. Item S8 within the MPEG-7 still images content set is used for performing experiments; this dataset consists of 3621 still images. Experimental results show that the proposed descriptors for 2D images are effective

Periodicity is useful for compact representation of periodic motion and a reasonable selection of a proper temporal scale for periodic motion analysis. In this paper, we concern the periodicity detection of local motion within an interesting region and present an approach to automatically detect the motion periodicity inherent to local motion under complex condition. The task is challenging as local motion is usually buried in clutters with global motion and noises. Most existing methods have assumed a static camera and a labeled moving object region. We instead apply robust local motion estimation and an object localization method to extract the object motion. The object motion is characterized by the confidence based motion probability map and the motion vectors obtained by global motion compensation. The autocorrelation series of motion energy is then carried out to locate local maximum points. With the set of indices of local maximum points, we can estimate the basic periodicity through a least-square fitting. This method has been applied to swimming videos and got encouraging results

This paper presents a geometric error concealment scheme for DCT-based image/video data based on the Bezier polynomials (BP). The proposed scheme makes use of the average edge direction and local curvature (extracted from healthy blocks around the damaged block) as boundary conditions to create an n-degree BP. The resulting BP is then used to interpolate the edges of the lost block, as well as to directionally reconstruct its low frequency data. When multiple edges are identified crossing the lost area, a cost function defined in terms of the local and global edge curvatures is used to find the best match for each missing edge. Experimental results show that our approach provides near perfect reconstruction and excellent subjective quality of the restored data, outperforming current linear interpolation schemes in the literature

The ability of a computer to detect and appropriately respond to changes in a user's affective state has significant implications to human-computer interaction (HCI). To more accurately simulate the human ability to assess affects through multi-sensory data, automatic affect recognition should also make use of multimodal data. In this paper, we present our efforts toward audio-visual affect recognition. Based on psychological research, we have chosen affect categories based on an activation-evaluation space which is robust in capturing significant aspects of emotion. We apply the Fisher boosting learning algorithm which can build a strong classifier by combining a small set of weak classification functions. Our experimental results show with 30 Fisher features, the testing error rates of our bimodal affect recognition is about 16% on the evaluation axis and 13% on the activation axis.

A digital image stabilization system compensates the image movement caused by hand jiggle for the image sequence captured by a hand-held video camera. In this paper, a simplified stabilization algorithm based on our previous work is presented. The algorithm performs block-based motion estimation on 16 local 16/spl times/16 blocks and uses a median filter to estimate the global motion. It reduces the complexity by confining the motion estimation to a small number of blocks of the image. This greatly facilitates the implementation of the algorithm on BF561, a DSP processor of analog device. Details of the DSP implementation are described.

Traditional design and test of complex multimedia systems involves a large number of test vectors and is a difficult and time-consuming task. The simulation times are prohibitively long on current desktop computers. Driving actual design scenarios and timing burst behavior which produce real-time effects is difficult to do with current simulation environments. This paper describes a rapid emulation framework for accessing multiple hardware IP blocks on an FPGA. This solution involves an abstraction of the FPGA platform by having a virtual socket layer that resides between the design and the test vehicles. A rapid prototyping platform is thus created, and its use with complex video and multimedia systems is described

Traditionally artistic color concepts play an important role in the analysis of artworks, and provide valuable domain knowledge to guide the analysis and accurate retrieval of paintings. This paper presents an automated approach to analyzing and representing artistic color concepts such as color temperature, color palette and color contrasts in paintings domain. The color concept definitions rely on the widely accepted color theory of Itten. Our approach starts with the extraction of homogeneous color/texture regions from paintings, and employs image processing and machine learning techniques to characterize regions in terms of artistic color concepts. We evaluate the proposed method using collection of 1,000 paintings, which exhibit vast variety of painting styles and artistic color concepts. Experiments show that our method performs well for a vast variety of both spatial and non-spatial artistic color concept queries

In this paper, we present an algorithm for automatically generating pencil-sketch like drawings from personal photos. On top of the core step of gradient computation, some proper transformations are introduced to achieve the best visual effects. It is found that the popular Sobel operator and Laplacian operator are both not desirable for this task, and thus we propose a computationally simple algorithm for gradient estimation. Experimental results show that the proposed method can generate visually appealing pencil-sketch like images from personal photos

In this paper, we introduce an elaborate utterance detection algorithm to enhance speaker segmentation. Silence detector, further divider and audio type classifier are employed in this elaborate utterance detection, to make this algorithm adaptive for both silent and noisy environments. Open-set verification testing has taken on the Hub4-NE broadcasts database. The experiment results show that this enhanced segmentation method can provide better information for speaker models.

In this paper, we propose a butterfly-graph based stream authentication scheme for lossy networks where the streaming packets could be lost in both random and burst ways. Due to the nice properties of butterfly graph, the proposed scheme is quite robust and efficient. Theoretical analysis and simulation results show that the proposed scheme outperforms existing schemes in terms of overhead and authentication probability while maintaining the same levels of sender/receiver delay and robustness.

In this paper, we present an efficient 3D shape rejection algorithm for unlabeled 3D markers. The problem is important in domains such as rehabilitation and the performing arts. There are three key innovations in our approach-(a) a multi-resolution shape representation using Haar wavelets for unlabeled markers, (b) a multi-resolution shape metric and (c) a shape rejection algorithm that is predicated on the simple idea that we do not need to compute the entire distance to conclude that two shapes are dissimilar. We tested the approach on a real-world pose classification problem with excellent results. We achieved a classification accuracy of 98% with an order of magnitude improvement in terms of computational complexity over a baseline shape matching algorithm.

Head pose estimation is a research area which has many applications, e.g. in human computer interfaces design or in the analysis of people's focus-of-attention. The paper addresses the issue of head pose estimation, and makes two contributions. First it introduces a database of more than 2 hours of video with head pose annotation involving people engaged in office activities or meeting discussion. The database is publicly available. The second is an algorithm which couples tracking and head pose estimation in a mixed-state particle filter. The approach combines the robustness of color-based tracking by exploiting skin head/face models with the localization accuracy of texture-based head models, as demonstrated by the reported experiments

Super-resolution is the problem of generating one or a set of high-resolution images from one or a sequence of low-resolution frames. Most methods have been proposed for super-resolution based on multiple low resolution images of the same scene, which is called multiple-frame super-resolution. Only a few approaches produce a high-resolution image from a single low-resolution image, with the help of one or a set of training images from scenes of the same or different types. It is referred to as single-frame super-resolution. This article reviews a variety of single-frame super-resolution methods proposed in the recent years. In the paper, a new manifold learning method: locally linear embedding (LLE) and its relation with single-frame super-resolution is introduced. Detailed study of a critical issue: "neighborhood issue" is presented with related experimental results and analysis and possible future research is given.

It is generally acknowledged that many experts and almost all laypersons have difficulty in formulating requests for information in such a manner that conventional off-line information extraction systems can find optimal answers. Therefore, it is increasingly evident that there is a need for an interactive dialog between information seekers and information extraction systems. In this paper, we describe the demonstrator of an interactive and multimodal information extraction system that is under construction in the NWO funded research program IMIX

The present research deals with audio events detection in noisy environments for a multimedia surveillance application. In surveillance or homeland security most of the systems aiming to automatically detect abnormal situations are only based on visual clues while, in some situations, it may be easier to detect a given event using the audio information. This is in particular the case for the class of sounds considered in this paper, sounds produced by gun shots. The automatic shot detection system presented is based on a novelty detection approach which offers a solution to detect abnormality (abnormal audio events) in continuous audio recordings of public places. We specifically focus on the robustness of the detection against variable and adverse conditions and the reduction of the false rejection rate which is particularly important in surveillance applications. In particular, we take advantage of potential similarity between the acoustic signatures of the different types of weapons by building a hierarchical classification system

This paper presents a novel quarter pel full search block motion estimation architecture for H.264/AVC encoder. The proposed architecture is capable of calculating all 41 motion vectors required by the various size blocks, supported by H.264/AVC, in parallel. The architecture has been prototyped in Verilog HDL, simulated and synthesized for Xilinx Virtex2 FPGA. The experimental result shows that the architecture is capable of processing CIF frame sequences in real time considering 5 reference frames within the search range of -3.75 to +4.00 at a clock speed of 120 MHz. The maximum speed of the architecture is around 150 MHz.

In this paper, we propose a multi-buffer scheduling scheme for streaming video systems. A transmission rate is obtained via a rate control algorithm, which optimally utilizes the network bandwidth and client buffer resources. The server side maintains multiple buffers for packets of different importance levels. It schedules the transmission of each packet based on the source buffer size and playback deadline to reduce the end-to-end distortion. The performance of proposed scheme is evaluated in terms of peak signal-to-noise ratio (PSNR) in the simulations, and the simulation results demonstrate the improvement of the average PSNR values compared with two other scheduling schemes

In this paper an approach for combining online signature authentication experts will be proposed. The different experts are based on one feature extraction method presented in our earlier work, the Biometric Hash algorithm [C. Viehauer, et al., (2002)], to which different distance measurement functions are applied. We will show that by the fusion of several algorithms with an appropriately parameterized strategy an improvement of the recognition accuracy can be achieved. The best fusion strategy results in a decrease of the EER of 12.1% in comparison to the best individual algorithm. The database we used contains 1761 genuine enrollments (with 4 signatures per enrollment), 1101 genuine verification signatures and 431 well skilled forgeries (so-called "brute force attack") by 22 persons. Based on our experimental results, we further discuss usability of alternative handwriting semantics such as pass phrases or PIN

In this paper, we propose a novel technique for model-based recognition of complex object motion trajectories using Gaussian mixture models (GMM). We build our models on principal component analysis (PCA)-based representation of trajectories after segmenting them into small units of perceptually similar pieces of motions. These subtrajectories are then fitted with automatically learnt mixture of Gaussians to estimate the underlying class probability distribution. Experiments are performed on two data sets; the ASL data set (from UCI's KDD archives) consists of 207 trajectories depicting signs for three words, from Australian sign language (ASL); the HJSL data set contains 108 trajectories from sports videos. Our experiments yield an accuracy of 85+% performing much better than existing approaches