Sparse representation for face images.

This thesis address issues for face recognition with multi-view face images. Several effective methods are proposed and compared with current state of the art. A novel framework that generalises existing sparse representation-based methods in order to exploit the sharing information to against pose variations of face images is proposed.

Applications of perceptual sparse representation (Spikegram) for copyright protection of audio signals

Chaque année, le piratage mondial de la musique coûte plusieurs milliards de dollars en pertes économiques, pertes d’emplois et pertes de gains des travailleurs ainsi que la perte de millions de dollars en recettes fiscales. La plupart du piratage de la musique est dû à la croissance rapide et à la facilité des technologies actuelles pour la copie, le partage, la manipulation et la distribution de données musicales [Domingo, 2015], [Siwek, 2007]. Le tatouage des signaux sonores a été proposé pour protéger les droit des auteurs et pour permettre la localisation des instants où le signal sonore a été falsifié. Dans cette thèse, nous proposons d’utiliser la représentation parcimonieuse bio-inspirée par graphe de décharges (spikegramme), pour concevoir une nouvelle méthode permettant la localisation de la falsification dans les signaux sonores. Aussi, une nouvelle méthode de protection du droit d’auteur. Finalement, une nouvelle attaque perceptuelle, en utilisant le spikegramme, pour attaquer des systèmes de tatouage sonore. Nous proposons tout d’abord une technique de localisation des falsifications (‘tampering’) des signaux sonores. Pour cela nous combinons une méthode à spectre étendu modifié (‘modified spread spectrum’, MSS) avec une représentation parcimonieuse. Nous utilisons une technique de poursuite perceptive adaptée (perceptual marching pursuit, PMP [Hossein Najaf-Zadeh, 2008]) pour générer une représentation parcimonieuse (spikegramme) du signal sonore d’entrée qui est invariante au décalage temporel [E. C. Smith, 2006] et qui prend en compte les phénomènes de masquage tels qu’ils sont observés en audition. Un code d’authentification est inséré à l’intérieur des coefficients de la représentation en spikegramme. Puis ceux-ci sont combinés aux seuils de masquage. Le signal tatoué est resynthétisé à partir des coefficients modifiés, et le signal ainsi obtenu est transmis au décodeur. Au décodeur, pour identifier un segment falsifié du signal sonore, les codes d’authentification de tous les segments intacts sont analysés. Si les codes ne peuvent être détectés correctement, on sait qu’alors le segment aura été falsifié. Nous proposons de tatouer selon le principe à spectre étendu (appelé MSS) afin d’obtenir une grande capacité en nombre de bits de tatouage introduits. Dans les situations où il y a désynchronisation entre le codeur et le décodeur, notre méthode permet quand même de détecter des pièces falsifiées. Par rapport à l’état de l’art, notre approche a le taux d’erreur le plus bas pour ce qui est de détecter les pièces falsifiées. Nous avons utilisé le test de l’opinion moyenne (‘MOS’) pour mesurer la qualité des systèmes tatoués. Nous évaluons la méthode de tatouage semi-fragile par le taux d’erreur (nombre de bits erronés divisé par tous les bits soumis) suite à plusieurs attaques. Les résultats confirment la supériorité de notre approche pour la localisation des pièces falsifiées dans les signaux sonores tout en préservant la qualité des signaux. Ensuite nous proposons une nouvelle technique pour la protection des signaux sonores. Cette technique est basée sur la représentation par spikegrammes des signaux sonores et utilise deux dictionnaires (TDA pour Two-Dictionary Approach). Le spikegramme est utilisé pour coder le signal hôte en utilisant un dictionnaire de filtres gammatones. Pour le tatouage, nous utilisons deux dictionnaires différents qui sont sélectionnés en fonction du bit d’entrée à tatouer et du contenu du signal. Notre approche trouve les gammatones appropriés (appelés noyaux de tatouage) sur la base de la valeur du bit à tatouer, et incorpore les bits de tatouage dans la phase des gammatones du tatouage. De plus, il est montré que la TDA est libre d’erreur dans le cas d’aucune situation d’attaque. Il est démontré que la décorrélation des noyaux de tatouage permet la conception d’une méthode de tatouage sonore très robuste. Les expériences ont montré la meilleure robustesse pour la méthode proposée lorsque le signal tatoué est corrompu par une compression MP3 à 32 kbits par seconde avec une charge utile de 56.5 bps par rapport à plusieurs techniques récentes. De plus nous avons étudié la robustesse du tatouage lorsque les nouveaux codec USAC (Unified Audion and Speech Coding) à 24kbps sont utilisés. La charge utile est alors comprise entre 5 et 15 bps. Finalement, nous utilisons les spikegrammes pour proposer trois nouvelles méthodes d’attaques. Nous les comparons aux méthodes récentes d’attaques telles que 32 kbps MP3 et 24 kbps USAC. Ces attaques comprennent l’attaque par PMP, l’attaque par bruit inaudible et l’attaque de remplacement parcimonieuse. Dans le cas de l’attaque par PMP, le signal de tatouage est représenté et resynthétisé avec un spikegramme. Dans le cas de l’attaque par bruit inaudible, celui-ci est généré et ajouté aux coefficients du spikegramme. Dans le cas de l’attaque de remplacement parcimonieuse, dans chaque segment du signal, les caractéristiques spectro-temporelles du signal (les décharges temporelles ;‘time spikes’) se trouvent en utilisant le spikegramme et les spikes temporelles et similaires sont remplacés par une autre. Pour comparer l’efficacité des attaques proposées, nous les comparons au décodeur du tatouage à spectre étendu. Il est démontré que l’attaque par remplacement parcimonieux réduit la corrélation normalisée du décodeur de spectre étendu avec un plus grand facteur par rapport à la situation où le décodeur de spectre étendu est attaqué par la transformation MP3 (32 kbps) et 24 kbps USAC.

Sparse subspace representation for spectral document clustering

We present a novel method for document clustering using sparse representation of documents in conjunction with spectral clustering. An ℓ1-norm optimization formulation is posed to learn the sparse representation of each document, allowing us to characterize the affinity between documents by considering the overall information instead of traditional pair wise similarities. This document affinity is encoded through a graph on which spectral clustering is performed. The decomposition into multiple subspaces allows documents to be part of a sub-group that shares a smaller set of similar vocabulary, thus allowing for cleaner clusters. Extensive experimental evaluations on two real-world datasets from Reuters-21578 and 20Newsgroup corpora show that our proposed method consistently outperforms state-of-the-art algorithms. Significantly, the performance improvement over other methods is prominent for this datasets.

Sparse temporal representations for facial expression recognition

In automatic facial expression recognition, an increasing number of techniques had been proposed for in the literature that exploits the temporal nature of facial expressions. As all facial expressions are known to evolve over time, it is crucially important for a classifier to be capable of modelling their dynamics. We establish that the method of sparse representation (SR) classifiers proves to be a suitable candidate for this purpose, and subsequently propose a framework for expression dynamics to be efficiently incorporated into its current formulation. We additionally show that for the SR method to be applied effectively, then a certain threshold on image dimensionality must be enforced (unlike in facial recognition problems). Thirdly, we determined that recognition rates may be significantly influenced by the size of the projection matrix \Phi. To demonstrate these, a battery of experiments had been conducted on the CK+ dataset for the recognition of the seven prototypic expressions - anger, contempt, disgust, fear, happiness, sadness and surprise - and comparisons have been made between the proposed temporal-SR against the static-SR framework and state-of-the-art support vector machine.

On robust face recognition via sparse encoding : the good, the bad, and the ugly

In the field of face recognition, Sparse Representation (SR) has received considerable attention during the past few years. Most of the relevant literature focuses on holistic descriptors in closed-set identification applications. The underlying assumption in SR-based methods is that each class in the gallery has sufficient samples and the query lies on the subspace spanned by the gallery of the same class. Unfortunately, such assumption is easily violated in the more challenging face verification scenario, where an algorithm is required to determine if two faces (where one or both have not been seen before) belong to the same person. In this paper, we first discuss why previous attempts with SR might not be applicable to verification problems. We then propose an alternative approach to face verification via SR. Specifically, we propose to use explicit SR encoding on local image patches rather than the entire face. The obtained sparse signals are pooled via averaging to form multiple region descriptors, which are then concatenated to form an overall face descriptor. Due to the deliberate loss spatial relations within each region (caused by averaging), the resulting descriptor is robust to misalignment & various image deformations. Within the proposed framework, we evaluate several SR encoding techniques: l1-minimisation, Sparse Autoencoder Neural Network (SANN), and an implicit probabilistic technique based on Gaussian Mixture Models. Thorough experiments on AR, FERET, exYaleB, BANCA and ChokePoint datasets show that the proposed local SR approach obtains considerably better and more robust performance than several previous state-of-the-art holistic SR methods, in both verification and closed-set identification problems. The experiments also show that l1-minimisation based encoding has a considerably higher computational than the other techniques, but leads to higher recognition rates.

Improved image set classification via joint sparse approximated nearest subspaces

Existing multi-model approaches for image set classification extract local models by clustering each image set individually only once, with fixed clusters used for matching with other image sets. However, this may result in the two closest clusters to represent different characteristics of an object, due to different undesirable environmental conditions (such as variations in illumination and pose). To address this problem, we propose to constrain the clustering of each query image set by forcing the clusters to have resemblance to the clusters in the gallery image sets. We first define a Frobenius norm distance between subspaces over Grassmann manifolds based on reconstruction error. We then extract local linear subspaces from a gallery image set via sparse representation. For each local linear subspace, we adaptively construct the corresponding closest subspace from the samples of a probe image set by joint sparse representation. We show that by minimising the sparse representation reconstruction error, we approach the nearest point on a Grassmann manifold. Experiments on Honda, ETH-80 and Cambridge-Gesture datasets show that the proposed method consistently outperforms several other recent techniques, such as Affine Hull based Image Set Distance (AHISD), Sparse Approximated Nearest Points (SANP) and Manifold Discriminant Analysis (MDA).

On robust face recognition via sparse coding: The good, the bad and the ugly

In the field of face recognition, sparse representation (SR) has received considerable attention during the past few years, with a focus on holistic descriptors in closed-set identification applications. The underlying assumption in such SR-based methods is that each class in the gallery has sufficient samples and the query lies on the subspace spanned by the gallery of the same class. Unfortunately, such an assumption is easily violated in the face verification scenario, where the task is to determine if two faces (where one or both have not been seen before) belong to the same person. In this study, the authors propose an alternative approach to SR-based face verification, where SR encoding is performed on local image patches rather than the entire face. The obtained sparse signals are pooled via averaging to form multiple region descriptors, which then form an overall face descriptor. Owing to the deliberate loss of spatial relations within each region (caused by averaging), the resulting descriptor is robust to misalignment and various image deformations. Within the proposed framework, they evaluate several SR encoding techniques: l1-minimisation, Sparse Autoencoder Neural Network (SANN) and an implicit probabilistic technique based on Gaussian mixture models. Thorough experiments on AR, FERET, exYaleB, BANCA and ChokePoint datasets show that the local SR approach obtains considerably better and more robust performance than several previous state-of-the-art holistic SR methods, on both the traditional closed-set identification task and the more applicable face verification task. The experiments also show that l1-minimisation-based encoding has a considerably higher computational cost when compared with SANN-based and probabilistic encoding, but leads to higher recognition rates.

Class-specific sparse codes for representing activities

In this paper we investigate the effectiveness of class specific sparse codes in the context of discriminative action classification. The bag-of-words representation is widely used in activity recognition to encode features, and although it yields state-of-the art performance with several feature descriptors it still suffers from large quantization errors and reduces the overall performance. Recently proposed sparse representation methods have been shown to effectively represent features as a linear combination of an over complete dictionary by minimizing the reconstruction error. In contrast to most of the sparse representation methods which focus on Sparse-Reconstruction based Classification (SRC), this paper focuses on a discriminative classification using a SVM by constructing class-specific sparse codes for motion and appearance separately. Experimental results demonstrates that separate motion and appearance specific sparse coefficients provide the most effective and discriminative representation for each class compared to a single class-specific sparse coefficients.

Low order anti-aliasing filters for sparse signals in embedded applications

Major emphasis, in compressed sensing (CS) research, has been on the acquisition of sub-Nyquist number of samples of a signal that has a sparse representation on some tight frame or an orthogonal basis, and subsequent reconstruction of the original signal using a plethora of recovery algorithms. In this paper, we present compressed sensing data acquisition from a different perspective, wherein a set of signals are reconstructed at a sampling rate which is a multiple of the sampling rate of the ADCs that are used to measure the signals. We illustrate how this can facilitate usage of anti-aliasing filters with relaxed frequency specifications and, consequently, of lower order.

Sparse Correlation Kernel Analysis and Reconstruction

This paper presents a new paradigm for signal reconstruction and superresolution, Correlation Kernel Analysis (CKA), that is based on the selection of a sparse set of bases from a large dictionary of class- specific basis functions. The basis functions that we use are the correlation functions of the class of signals we are analyzing. To choose the appropriate features from this large dictionary, we use Support Vector Machine (SVM) regression and compare this to traditional Principal Component Analysis (PCA) for the tasks of signal reconstruction, superresolution, and compression. The testbed we use in this paper is a set of images of pedestrians. This paper also presents results of experiments in which we use a dictionary of multiscale basis functions and then use Basis Pursuit De-Noising to obtain a sparse, multiscale approximation of a signal. The results are analyzed and we conclude that 1) when used with a sparse representation technique, the correlation function is an effective kernel for image reconstruction and superresolution, 2) for image compression, PCA and SVM have different tradeoffs, depending on the particular metric that is used to evaluate the results, 3) in sparse representation techniques, L_1 is not a good proxy for the true measure of sparsity, L_0, and 4) the L_epsilon norm may be a better error metric for image reconstruction and compression than the L_2 norm, though the exact psychophysical metric should take into account high order structure in images.

Joint multiple dictionary learning for tensor sparse coding

Traditional dictionary learning algorithms are used for finding a sparse representation on high dimensional data by transforming samples into a one-dimensional (1D) vector. This 1D model loses the inherent spatial structure property of data. An alternative solution is to employ Tensor Decomposition for dictionary learning on their original structural form —a tensor— by learning multiple dictionaries along each mode and the corresponding sparse representation in respect to the Kronecker product of these dictionaries. To learn tensor dictionaries along each mode, all the existing methods update each dictionary iteratively in an alternating manner. Because atoms from each mode dictionary jointly make contributions to the sparsity of tensor, existing works ignore atoms correlations between different mode dictionaries by treating each mode dictionary independently. In this paper, we propose a joint multiple dictionary learning method for tensor sparse coding, which explores atom correlations for sparse representation and updates multiple atoms from each mode dictionary simultaneously. In this algorithm, the Frequent-Pattern Tree (FP-tree) mining algorithm is employed to exploit frequent atom patterns in the sparse representation. Inspired by the idea of K-SVD, we develop a new dictionary update method that jointly updates elements in each pattern. Experimental results demonstrate our method outperforms other tensor based dictionary learning algorithms.

On the efficiency of data representation on the modeling and characterization of complex networks

Specific choices about how to represent complex networks can have a substantial impact on the execution time required for the respective construction and analysis of those structures. In this work we report a comparison of the effects of representing complex networks statically by adjacency matrices or dynamically by adjacency lists. Three theoretical models of complex networks are considered: two types of Erdos-Renyi as well as the Barabasi-Albert model. We investigated the effect of the different representations with respect to the construction and measurement of several topological properties (i.e. degree, clustering coefficient, shortest path length, and betweenness centrality). We found that different forms of representation generally have a substantial effect on the execution time, with the sparse representation frequently resulting in remarkably superior performance. (C) 2011 Elsevier B.V. All rights reserved.

The backfilled GEI : a cross-capture modality gait feature for frontal and side-view gait recognition

In this paper, we propose a novel direction for gait recognition research by proposing a new capture-modality independent, appearance-based feature which we call the Back-filled Gait Energy Image (BGEI). It can can be constructed from both frontal depth images, as well as the more commonly used side-view silhouettes, allowing the feature to be applied across these two differing capturing systems using the same enrolled database. To evaluate this new feature, a frontally captured depth-based gait dataset was created containing 37 unique subjects, a subset of which also contained sequences captured from the side. The results demonstrate that the BGEI can effectively be used to identify subjects through their gait across these two differing input devices, achieving rank-1 match rate of 100%, in our experiments. We also compare the BGEI against the GEI and GEV in their respective domains, using the CASIA dataset and our depth dataset, showing that it compares favourably against them. The experiments conducted were performed using a sparse representation based classifier with a locally discriminating input feature space, which show significant improvement in performance over other classifiers used in gait recognition literature, achieving state of the art results with the GEI on the CASIA dataset.

Human identification from video using advanced gait recognition techniques

The solutions proposed in this thesis contribute to improve gait recognition performance in practical scenarios that further enable the adoption of gait recognition into real world security and forensic applications that require identifying humans at a distance. Pioneering work has been conducted on frontal gait recognition using depth images to allow gait to be integrated with biometric walkthrough portals. The effects of gait challenging conditions including clothing, carrying goods, and viewpoint have been explored. Enhanced approaches are proposed on segmentation, feature extraction, feature optimisation and classification elements, and state-of-the-art recognition performance has been achieved. A frontal depth gait database has been developed and made available to the research community for further investigation. Solutions are explored in 2D and 3D domains using multiple images sources, and both domain-specific and independent modality gait features are proposed.