Wenchao Zhang

Are Gabor phases really useless for face recognition

Gabor features have been recognized as one of the best representations for face recognition. Usually, only the magnitudes of the Gabor coefficients are thought of as being useful for face recognition, while the phases of the Gabor features are deemed to be useless and thus usually ignored by face recognition researchers. However, in this paper, our findings show that the latter should be reconsidered. By encoding Gabor phases through local binary patterns and local histograms, we have achieved very impressive recognition results, which are comparable to those of Gabor magnitudes-based methods. The results of our experiments also indicate that, by combining the phases with the magnitudes, higher accuracy can be achieved. Such observations suggest that more attention should be paid to the Gabor phases for face recognition.

Local Gabor Binary Patterns Based on Kullback–Leibler Divergence for Partially Occluded Face Recognition

The partial occlusion is one of the key issues in the face recognition community. To resolve the problem of partial occlusion, based on our previous work of local Gabor binary patterns (LGBP) for face recognition, we further propose Kullback-Leibler divergence (KLD)-based LGBP for partial occluded face recognition. The local property of LGBP face recognition is thoroughly used in the method, by introducing KLD between the LGBP feature of the local region and that of the non-occluded local region to estimate the probability of occlusion. The probability is used as the weight of the local region for the final feature matching. The experimental results on the AR face database demonstrate the effectiveness of the KLD-based LGBP face recognition method for partially occluded face images.

Robust Head Pose Estimation Using LGBP

In this paper, we introduce a novel discriminative feature which is efficient for pose estimation. The multi-view face representation is based on local Gabor binary patterns (LGBP) and encodes the local facial characteristics in to a compact feature histogram. In LGBP, Gabor filters can extract the feature of the orientation of head and local binary pattern (LBP) can extract the features official local orientation. To keep the spatial information of the multi-view face images, LGBP is operated on many sub-regions of the images. The combination of them can represent well and truly the multi-view face images. Considering the derived feature space, a radial basis function (RBF) kernel SVM classifier is trained to estimate pose. Extensive experiments demonstrate that the facial representation can be effective for pose estimation

LOCAL GABOR BINARY PATTERNS BASED ON MUTUAL INFORMATION FOR FACE RECOGNITION

Appropriate representation is one of the keys to the success of face recognition technologies. In this paper, we present a novel face representation approach using a reduced set of local histograms based on Local Gabor Binary Patterns (LGBP). In the proposed method, a face image is first represented by the LGBP histograms which are extracted from the LGBP images. Then, the local LGBP histograms with high separability and low relevance are selected to obtain a dimension-reduced face descriptor. Extensive experimental results demonstrate that the proposed method not only greatly reduces the dimensionality of face representation, but also outperforms the state-of-the-art approaches for face recognition, such as Fisherfaces, and Gabor Fisher Classification (GFC).

Information fusion in face identification

Information fusion of multi-modal biometrics has attracted much attention in recent years. However, this paper focuses on the information fusion in single models, that is, the face biometric. Two different representation methods, gray level intensity and Gabor feature, are exploited for fusion. We study the fusion problem in face recognition at both the face representation level and the confidence level. At the representation level, both the PCA feature fusion and the LDA feature fusion are considered, while at the confidence level, the sum rule and the product rule are investigated. We show through experiments on FERET face database and our own face database that appropriate information fusion can improve the performance of face recognition and verification. This suggests that gray level intensity and Gabor feature compensate for each other, based on the feasible fusion.

Granularity-tunable gradients partition (GGP) descriptors for human detection

This paper proposes a novel descriptor, granularity-tunable gradients partition (GGP), for human detection. The concept granularity is used to define the spatial and angular uncertainty of the line segments in the Hough space. Then this uncertainty is backprojected into the image space by orientation-space partitioning to achieve efficient implementation. By changing the granularity parameter, the level of uncertainty can be controlled quantitatively. Therefore a family of descriptors with versatile representation property can be generated. Specifically, the finely granular GGP descriptors can represent the specific geometry information of the object (the same as Edgelet); while the coarsely granular GGP descriptors can provide the statistical representation of the object (the same as histograms of oriented gradients, HOG). Moreover, the position, orientation, strength and distribution of the gradients are embedded into a unified descriptor to further improve the GGPs representation power. A cascade structured classifier is built by boosting the linear regression functions. Experimental results on INRIA dataset show that the proposed method achieves comparable results to those of the state-of-the-art methods.

Component-Based cascade linear discriminant analysis for face recognition

This paper presents a novel face recognition method based on cascade Linear Discriminant Analysis (LDA) of the component-based face representation In the proposed method, a face image is represented as four components with overlap at the neighboring area rather than a whole face patch Firstly, LDA is conducted on the principal components of each component individually to extract component discriminant features Then, these features are further concatenated to undergo another LDA to extract the final face descriptor, which actually have assigned different weights to different component features Our experiments on the FERET face database have illustrated the effectiveness of the proposed method compared with the traditional Fisherface method both for face recognition and verification.