Research On Face Recognition Based On Biologically Inspired Features

Face recognition can be widely used in many fields where identity recognition or verification is needed, such as in frontier inspection, customs inspection, public security,criminal investigation, intelligent video monitoring system, intelligent entrance guard system, intelligent attendance system, intelligent tickets system, human-computer interaction, access control of computer or other important equipment, thus it has received extensive attentions from academic world and the industrial world and is still a hot research topic in the area of pattern recognition. In this paper, the applications of the partial least square regression and canonical correlation analysis in face recognition is studied, the main contributions are as following:1. An algorithm for face recognition based on biologically inspired features and partial least square(PLS) is presented. First, C1 feature of the facial images are extracted, then principal component analysis(PCA) method is used to reduce the dimension of the C1 features, finally, a class membership matrix is constructed and partial least square is used for classification and recognition. Algorithm is tested on the ORL face database and Yale face database, biologically inspired features is compared with two famous features in the computer vision, LBP features and HOG features, in addition, the PLS method is compared with the nearest neighbor classification and SVM. The experimental results show that the recognition rate of our method C1+PCA+PLS is higher than those of a lot of commonly used face recognition methods.2. An algorithm for face recognition based on biologically inspired features and canonical correlation analysis(CCA) is presented. First, C1 features of the facial images are extracted, then PCA+CCA method is used to reduce the dimension of the C1 features,finally, a class membership matrix is constructed and multiple linear regression method is used for classification and recognition. Algorithm is also tested on the ORL face database and Yale face database The experimental results show that the recognition rate of the algorithm has improved further compared with the C1+PCA+PLS algorithm.3. Considering the C1 features and LBP features has certain complementary to each other, a new algorithm for face recognition is proposed based on feature fusion, the C1 features and LBP features are fused using CCA method and then SVM is used for classification. The experimental results on the ORL database show that the recognition rateof this algorithm is higher than those using the single kind of features.