Face Recognition Based on Wavelet Transform and Image Principle Component Analysis

Image principle component analysis (IMPCA) is a rapid direct feature extract approach from matrix. We present the basic theory of IMPCA from the view of minimizing the mean reconstruction error. We analyze the feature generated from IMPCA and find they present row characters. Wavelet transform can be used in reducing noise of images and the wavelet image should be more suitable for recognition. Our proposed method firstly transforms face image with wavelet and gets coefficients of different frequency, then horizontal detail coefficient is enhanced. The image generated by wavelet inverse transform is as new object and is recognized using IMPCA. The experiment result on ORL face database presents the proposed method is efficient and the recognition accuracy rate is better than IMPCA only.