BULDP: Biomimetic Uncorrelated Locality Discriminant Projection for Feature Extraction in Face Recog
BULDP: Biomimetic Uncorrelated Locality Discriminant Projection for Feature Extraction in Face Recognition
This paper develops a new dimensionality reduction method, named Biomimetic Uncorrelated Locality Discriminant Projection (BULDP), for face recognition. It is based on unsupervised discriminant projection and two human bionic characteristics: principle of homology continuity and principle of heterogeneous similarity. With these two human bionic characteristics, we propose a novel adjacency coefficient representation, which does not only capture the category information between different samples, but also reflects the continuity between similar samples and the similarity between different samples. By applying this new adjacency coefficient into the unsupervised discriminant projection, it can be shown that we can transform the original data space into an uncorrelated discriminant subspace. A detailed solution of the proposed BULDP is given based on singular value decomposition. Moreover, we also develop a nonlinear version of our BULDP using kernel functions for nonlinear dimensionality reduction. The performance of the proposed algorithms is evaluated and compared with the state-of-the-art methods on four public benchmarks for face recognition. Experimental results show that the proposed BULDP method and its nonlinear version achieve much competitive recognition performance.
Face images lie in a very high dimensional space, which makes the task of recognition very difficult. Dimensionality reduction techniques have been widely used to represent the raw data in a compact way without losing too much useful information. These techniques learn a lower dimensional subspace to represent the face such that the image analysis can be performed more efficiently. Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA are two famous linear algorithms for unsupervised and supervised dimensionality reduction respectively, which have been widely studied and extensively used in many fields such as computer vision, pattern recognition and other biometrics.
We proposed a Biomimetic Uncorrelated Locality Discriminant Projection (BULDP) approach. BULDP is based on UDP, but with a new way of adjacency coefficient construction which is proposed according to the characteristics of imagery thinking. The proposed adjacency coefficient does not only make use of the category information between samples, but also reflect the law between the same samples and the similarity between the different samples. Besides, BULDP introduces the concept of uncorrelated spaces, which makes the last of the vector has no correlation and reduces the redundancy of the extracted vectors. In addition, an extended version of Kernel Biomimetic Uncorrelated Locality Discriminant Projection (KBULDP) is given, which can be considered as a generalization of BULDP in kernel space. To demonstrate its effectiveness, we apply our proposed BULDP methods for face recognition and the experimental results are encouraging.