Research On Dimension Reduction For Face Recognition

Dimensionality reduction methods in the field of face recognition is a kind of very key technology, If directly deal with high-dimensional image data, and the high dimension will not only lead to the “curse of dimensionality” phenomenon, but also makes it difficulty to directly understand and discovery the structure information of the data set.The major algorithms are used as the method of data preprocessing, on the one hand, it can overcome the “curse of dimensionality” phenomenon, on the other hand, it can greatly reduce the calculation complexity and noise, data dimension reduction method has attracted wide attention of the researchers.In this paper, we firstly introduction the typical existing one-dimensional dimensionality reduction methods and two-dimensional one side dimensionality reduction methods in the field of face recognition, then we analysis the bidirectional projection method of tensor subspace analysis(TSA) and discriminant tensor subspace(DTSA). To the problem of the left and right projection matrices are not usually orthogonal in BDSLPP,and the requirement of the orthogonality of the columns of projection matrices is common in that orthogonal projection matrices preserve the metric structure of the facial image space we propose the orthogonal BDSLPP(OBDSLPP), Finally we perform the algorithms on ORL and Yale face data sets. Experimental results demonstrate the recognition algorithm is feasible and has good recognition capability.