Face Recognition Method Based on 2DLDA and SVM Optimated by PSO Algorithm

摘要

Concerning the "Small Samples Size" problem in LDA algorithm and reduce the effects to the SVM face recognition rate caused by random parameters set by human. An algorithm based on combination with the PSO algorithm which was originated form artificial life and evolutionary computation to SVM's parameters election and optimization, and Wavelet Transform, two-dimensional LDA(2DLDA) was proposed. Firstly, the original images were decomposed into high-frequency and low-frequency Components by Wavelet Transform (WT). The high-frequency components were ignored, while the low-frequency components can be obtained. Then, the liner discriminant features were extracted by two-dimensional LDA (2DLDA). Finally, we use the PSO algorithm to SVM's parameters election and optimization. Experimental results based on ORL face database show the validity of the algorithm this paper proposed for face recognition and it can reach the recognition rate of 98%.