Ammar Assoum

Enhanced and parameterless Locality Preserving Projections for face recognition

In this paper, we address the graph-based linear manifold learning method for object recognition. The proposed method is called enhanced Locality Preserving Projections. The main contribution is a parameterless computation of the affinity matrix that draws on the notion of meaningful and adaptive neighbors. It integrates two interesting properties: (i) being entirely parameter-free and (ii) the mapped data are uncorrelated. The proposed method has been integrated in the framework of three graph-based embedding techniques: Locality Preserving Projections (LPP), Orthogonal Locality Preserving Projections (OLPP), and supervised LPP (SLPP). Recognition tasks on six public face databases show an improvement over the results of LPP, OLPP, and SLPP. The proposed approach could also be applied to other category of objects.

Self-optimized two phase test sample sparse representation method for image classification

Sparse Representation Classifiers and their variants are more and more used by computer vision and signal processing communities due to their good performance. Recently, it has been shown that the performance of Sparse Representation Classifiers and their variants in terms of accuracy and computational complexity can be enhanced by simply including a two-phase coding scheme regardless of the used coding scheme. The two-phase strategies use different schemes for selecting the examples that should be handed over to the next coding phase. However, all of them use a fixed and predefined number for these examples making the performance of the final classifier very dependent on this choice. This paper introduces three strategies for self-optimized size selection associated with Two Phase Test Sample Sparse Representation method. Experiments conducted on three face data sets show that the introduced scheme can outperform the classic two-phase strategies. Although the experiments were conducted on face data sets, the proposed schemes can be useful for a broad spectrum of pattern recognition problems.

Kernel flexible manifold embedding for pattern classification

Flexible Manifold Embedding (FME) has been recently proposed as a semi-supervised graph-based label propagation method. It aims at estimating simultaneously the optimal prediction labels and its linear regression. It integrates the label fitness, the manifold smoothness and a flexible term that forces the linear regression to be as close as possible to nonlinear one. Despite its good performance compared to its counterparts, FME may lead to poor performance when the geometrical structure of data is highly nonlinear. In this paper, we propose a Kernel version of the Flexible Manifold Embedding (KFME). As in classical FME, KFME uses labeled and unlabeled data to estimate the embedding of unlabeled data and its regression function that can map new data samples. Extensive experiments carried out on eight benchmark datasets show that the proposed KFME can outperform FME as well as many state-of-the-art semi-supervised learning methods.

Adaptive Two Phase Sparse Representation Classifier for Face Recognition

Sparse Representation Classifier proved to be a powerful classifier that is more and more used by computer vision and signal processing communities. On the other hand, it is very computationally expensive since it is based on an L 1 minimization. Thus, it is not useful for scenarios demanding a rapid decision or classification. For this reason, researchers have addressed other coding schemes that can make the whole classifier very efficient without scarifying the accuracy of the original proposed SRC. Recently, two-phase coding schemes based on classic Regularized Least Square were proposed. These two-phase strategies can use different schemes for selecting the examples that should be handed over to the next coding phase. However, all of them use a fixed and predefined number for these selected examples making the performance of the final classifier very dependent on this ad-hoc choice. This paper introduces three strategies for adaptive size selection associated with Two Phase Test Sample Sparse Representation classifier. Experiments conducted on three face datasets show that the introduced schemes can outperform the classic two-phase strategies. Although the experiments were conducted on face datasets, the proposed schemes can be useful for a broad spectrum of pattern recognition problems.

Inductive and flexible feature extraction for semi-supervised pattern categorization

This paper proposes a novel discriminant semi-supervised feature extraction method for generic classification and recognition tasks. This method, called inductive flexible semi-supervised feature extraction, is a graph-based embedding method that seeks a linear subspace close to a non-linear one. It is based on a criterion that simultaneously exploits the discrimination information provided by the labeled samples, maintains the graph-based smoothness associated with all samples, regularizes the complexity of the linear transform, and minimizes the discrepancy between the unknown linear regression and the unknown non-linear projection. We extend the proposed method to the case of non-linear feature extraction through the use of kernel trick. This latter allows to obtain a nonlinear regression function with an output subspace closer to the learned manifold than that of the linear one. Extensive experiments are conducted on ten benchmark databases in order to study the performance of the proposed methods. Obtained results demonstrate a significant improvement over state-of-the-art algorithms that are based on label propagation or semi-supervised graph-based embedding. HighlightsA flexible graph-based semi-supervised embedding is proposed.A kernel version of the linear semi-supervised algorithm is also proposed.They simultaneously estimate a non-linear embedding and its out-of-sample extension.Classification performance after embedding is assessed on ten benchmark datasets.We use KNN, SVM, and two phase test sample sparse representation as classifiers.

Advances in Face Image Analysis: Theory and Applications

Description: Advances in Face Image Analysis: Theory and applications describes several approaches to facial image analysis and recognition. Eleven chapters cover advances in computer vision and pattern recognition methods used to analyze facial data. The topics addressed in this book include automatic face detection, 3D face model fitting, robust face recognition, facial expression recognition, face image data embedding, modelless 3D face pose estimation and image-based age estimation.

Machine learning schemes in augmented reality for features detection

Augmented Reality (AR) is a relatively old concept technology, which reached the large public very recently. We can use it to enhance our environments, by augmenting the image, the voice and delivering details and annotations about the surrounding space. Augmented reality (AR) is a growing field, with many diverse applications ranging from TV and film production, to industrial maintenance, medicine, education, entertainment and games. This paper presents an improved approach for image augmented-reality, by acting on two axes in the augmented reality process. First, a machine learning step is added to the detection part. Second, the registration of augmented image is processed by using the following techniques: statistical appearance models, and covariance matrices of dense image descriptors. A tuning of the used techniques and algorithms will be done in order to obtain a reliable and real-time image augmentation. We give a detailed description on how we chose the methods, and we compare our approach with other methods used in this domain. Finally, an evaluation of the proposed technique is presented as well as a performance study for a given use case.

Linear dimensionality reduction through eigenvector selection for object recognition

Past work on Linear Dimensionality Reduction (LDR)has emphasized the issues of classification and dimension estimation. However, relatively less attention has been given to the critical issue of eigenvector selection. The main trend in feature extraction has been representing the data in a lower dimensional space, for example, using principal component analysis (PCA) without using an effective scheme to select an appropriate set of features/eigenvectors in this space. This paper addresses Linear Dimensionality Reduction through Eigenvector selection for object recognition. It has two main contributions. First, we propose a unified framework for one transform based LDR. Second, we propose a framework for two transform based DLR. As a case study, we consider PCA and Linear Discriminant Analysis (LDA) for the linear transforms. We have tested our proposed frameworks on several public benchmark data sets. Experiments on ORL, UMIST, and YALE Face Databases and MNIST Handwritten Digit Database show significant performance improvements in recognition that are based on eigenvector selection.

Data reduction through kernel sparse coding

Sparse Modeling Representative Selection (SMRS) has been recently proposed for finding the most relevant instances in datasets. This method deploys a data self-representativeness coding in order to infer a coding matrix that is regularized with a row sparsity constraint. The method assumes that the score of any sample is set to the L2 norm of the corresponding row in the coding matrix. Since the SMRS method is linear, it cannot always provide good relevant instances. Moreover, many of its selected instances are already in dense areas in the input space. In this paper, we propose to alleviate the SMRS methods shortcomings. More precisely, We propose two kernel data self-representativeness coding schemes that are based on Hilbert space and column generation. Performance evaluation is carried out on reducing training image datasets used for recognition tasks. These experiments showed that the proposed kernel methods can provide better data reduction than state-of-the art selection methods including the SMRS method.

Augmented Reality application: A new implementation chain

Augmented Reality (AR) is a live view of a real-world environment. With advanced AR technology, artificial information about the environment and its objects can be overlaid on the real world. This paper presents a complete augmented reality process for a video sequence captured by a moving camera. The main goal is to construct a full chain composed of 4 blocks that correspond to the main steps of augmented reality process: feature detection, feature extraction, feature matching and image registration. Our work proposes an improved technique for image augmentation, starting from feature detection and ending by image registration. We used the well-known techniques (e.g. SIFT, SURF, etc.) for features detection and extraction in order to compare their performance. Furthermore, we added a features learning step (using SVM, KNN and SRC) to improve the image registration process. The final full chain uses the best method in each block. This best combination is then applied on all video frames taken by the camera. Thereafter, we obtain a video showing the augmented object instead of the real one.