Face Recognition Based on Wavelet Transform and Kernel Fisher Discriminant Analysis

A novel face recognition method combination wavelet transform and Kernel Fisher Discriminant Analysis (KFDA) was presented. Firstly, in the proposed method, 2-dimensional wavelet transform was calculated in logarithm domain for face pre-processing. Secondly, KFDA algorithm was used for face feature extraction. Finally, the kNN (k nearest neighborhood) classifier using Cosine distance was adopted for feature classification. The experimental results on Yale B frontal face database show that the proposed approach can identify all test samples accurately, that is, the face recognition rate can attain 100% when wavelet type and wavelet decomposing levels are selected properly. The proposed method can improve the performance of face illumination composition in face recognition algorithm effectively.