Yongxin Ge

Discriminant Hyper-Laplacian projections with its application to face recognition

Discriminant Locality Preserving Projections (DLPP) is one of the most influential supervised subspace learning algorithms that considers both discriminative and geometric (manifold) information. There is an obvious drawback of DLPP that it only considers the pairwise geometric relationship of samples. However, in many real-world issues, relationships among the samples are often more complex than pairwise. Naively squeezing the complex into pairwise ones will inevitably lead to loss of some information, which are crucial for classification and clustering. We address this issue via using the Hyper-Laplacian instead of the regular Laplacian in DLPP, which only can depict the pairwise relationship. This new DLPP algorithm is exactly a generalization of DLPP and we name it Discriminant Hyper-Laplacian Projection (DHLP). Five popular face databases are adopted for validating our work. The results demonstrate the superiority of DHLP over DLPP, particularly in face recognition in the wild.

Makeup-robust face verification

We investigate in this paper the problem of face verification in the presence of face makeups. To our knowledge, this problem has less formally addressed in the literature. A key challenge is how to increase the measured similarity between face images of the same person without and with makeups. In this paper, we propose a novel approach for makeup-robust face verification, by measuring correlations between face images in a meta subspace. The meta subspace is learned using canonical correlation analysis (CCA), with the objective that intra-personal sample correlations are maximized. Subsequently, discriminative learning with the support vector machine (SVM) classifier is applied to verify faces based on the low-dimensional features in the learned meta subspace. Experimental results on our dataset are presented to demonstrate the efficacy of our approach.

Body-based human age estimation at a distance

Estimating human ages at a distance has many potential applications, especially for visual surveillance in such places as supermarkets and airports. In this paper, we propose a new human age estimation approach from full body images with frontal or back views. Given a body image, we extract local SIFT features from body image patches and learn sparse coefficients for body image feature representation, followed by a regressor for age prediction. Experimental results have clearly demonstrated the feasibility of using fully body images to estimate human age and the effectiveness of our proposed approach.

Activity-based human identification

We investigate in this paper the problem of activity-based human identification. Different from most existing gait recognition methods where only human walking activity is considered and utilized for person identification, we aim to identify people from various activities such as eating, jumping, and weaving. For each video clip, we first extract binary human body masks by using background substraction, followed by computing the average energy image (AEI) features to represent each video clip. Then, a mapping is learned by applying an adaptive discriminant analysis (ADA) method to project AEI features into a low-dimensional subspace, such that the intra-class (activities performed by the same person) variations are minimized and the interclass (activities performed by different persons) are maximized, simultaneously. Moreover, interclass samples with large similarity difference are deemphasized and those with small difference are emphasized, such that more discriminative information can be used for recognition. Experimental results on three publicly available databases show the efficacy of our proposed approach.

Age estimation from human body images

In this paper, we investigate the problem of estimating human ages from full body images. To our best knowledge, this problem has not been formally addressed before possibly due to the great challenges and lacking of such publicly available datasets. However, estimating human ages at a distance has a number of potential applications, especially for visual surveillance in such places as supermarkets, airports, building entrances, and shopping malls. In this paper, we propose a new human age estimation approach from full body images with frontal or back views. Our contributions are three-fold. First, we collect a human body image dataset containing 1500 public figures or celebrities searched from the internet, as well as the age label information of each image. Second, we explore several widely used human local appearance feature descriptors with a regression model to estimate human ages from these body images. Lastly, we apply a multiview canonical correlation analysis (MCCA) method by making use of multiple feature descriptors to exploit complementary information to further improve the age estimation performance. Experimental results have clearly demonstrated the feasibility of using fully body images to estimate human age and the efficacy of our proposed approach.

Improved hypergraph regularized Nonnegative Matrix Factorization with sparse representation

Abstract As a commonly used data representation technique, Nonnegative Matrix Factorization (NMF) has received extensive attentions in the pattern recognition and machine learning communities over decades, since its working mechanism is in accordance with the way how the human brain recognizes objects. Inspired by the remarkable successes of manifold learning, more and more researchers attempt to incorporate the manifold learning into NMF for finding a compact representation,which uncovers the hidden semantics and respects the intrinsic geometric structure simultaneously. Graph regularized Nonnegative Matrix Factorization (GNMF) is one of the representative approaches in this category. The core of such approach is the graph, since a good graph can accurately reveal the relations of samples which benefits the data geometric structure depiction. In this paper, we leverage the sparse representation to construct a sparse hypergraph for better capturing the manifold structure of data, and then impose the sparse hypergraph as a regularization to the NMF framework to present a novel GNMF algorithm called Sparse Hypergraph regularized Nonnegative Matrix Factorization (SHNMF). Since the sparse hypergraph inherits the merits of both the sparse representation and the hypergraph model, SHNMF enjoys more robustness and can better exploit the high-order discriminant manifold information for data representation. We apply our work to address the image clustering issue for evaluation. The experimental results on five popular image databases show the promising performances of the proposed approach in comparison with the state-of-the-art NMF algorithms.

Discriminant Hyper-Laplacian Projections and its scalable extension for dimensionality reduction

Discriminant Locality Preserving Projections (DLPP) is one of the most influential supervised subspace learning algorithms that consider both discriminative and geometric (manifold) information. There is an obvious drawback of DLPP that it only considers the pairwise geometric relationship of samples. However, in many real-world issues, relationships among the samples are often more complex than pairwise. Naively squeezing the complex into pairwise ones will inevitably lead to loss of some information, which are crucial for classification and clustering. We address this issue via using the Hyper-Laplacian instead of the regular Laplacian in DLPP, which only can depict the pairwise relationship. This new DLPP algorithm is exactly a generalization of DLPP and we name it Discriminant Hyper-Laplacian Projection (DHLP). In order to make DHLP can be feasibly applied to big data dimensionality reduction, we adopt the spectral regression framework to reduce the computational complexity of DHLP from cubic-time to linear-time. We named this new DHLP algorithm Scalable Discriminant Hyper-Laplacian Projections (SDHLP). Six popular visual databases are adopted for validating our work. The results not only demonstrate the superiorities of DHLP and SDHLP in comparison with the state-of-the-art algorithms but also demonstrate the efficiency improvement of SDHLP over DHLP.

Two dimensional non-negative sparse Partial Least Squares for face recognition

The Partial Least Squares (PLS) algorithm has been widely applied in face recognition in recent years. However, all the improved algorithms of PLS did not utilize non-negativity and sparsity synchronously to improve the recognition accuracy and robustness. In order to solve these problems, this paper proposes a novel algorithm named Two-Dimension Non-negative Sparse Partial Least Squares (2DNSPLS), which incorporates the constraints of non-negativity and sparse to 2DPLS while extracting the facial features. Consequently, not only do the features extracted by 2DNSPLS contain the label information, as well as the internal structure of image matrix, but they also contain local non-negative interpretability and sparsity. For evaluating the approachs performance, a series of experiments are conducted on the Yale and the PIE face databases, which demonstrate that the proposed approach outperforms the state-of-art algorithms and has good robustness to occlusion.

Two dimension nonnegative partial least squares for face recognition

For benefiting from incorporating the class information, partial least squares (PLS) and its two dimension version (2DPLS) have been widely employed in face recognition when extracting principal components. However, currently popular statistic methods, such as principal component analysis (PCA) and linear discriminant analysis (LDA), only learn holistic, not parts-based, representations which ignore available local features for face recognition. In this paper, we propose a novel approach to extract the facial features called two dimension nonnegative partial least squares (2DNPLS). Our approach can grab the local features via adding non-negativity constraint to the 2DPLS, and can also reserve the advantages of 2DPLS, which are both inherent structure and class information of images. For evaluating our approachs performance, a series of experiments were conducted on two famous face image databases include ORL and Yale face databases, which demonstrate that our proposed approach outperforms the compared state-of-art algorithms.