The Research Of Face Recognition Algorithm Based On Gabor Wavelet Combined With PCA

In recent years, face recognition technology has attracted more and more attention in many fields and becomes a hotspot. It has great application in file management, personnel security, cards certification,criminal identification, machine learning, and have made some achievements.This paper present the significance of face recognition, application and research status. It does some theoretical and experimental research on face feature extraction and face recognition algorithms, analyzes some classical algorithms such as PCA, LDA and features of Gabor wavelet transform. At the same time, this paper also compares the advantages and disadvantages of PCA and Gabor wavelet. On one hand Gabor wavelet has good spatial locality and direction selectivity while PCA is susceptible to interference illumination, expression and other external factors which lead to a decline on recognition rate. On the other Gabor will bring about high dimensions and the large amount of computation while PCA can reduce the dimension. So we combine the Gabor wavelet with PCA and have complementary advantages.However, before the feature extraction, in order to eliminate the redundant information in image, filter out the noise, enhance the relevant information, improve the reliability of recognition, we proposed a preprocessing method called whitening/low-pass filter. That is, this paper propose a improved algorithm for face recognition based on whitening/low-pass Gabor wavelet and PCA. First of all, we use the whitening/low-pass filter to face image pre-processing. Then let the Gabor wavelet convolution extract feature of each image and reduce the dimension with PCA. Lastly, the nearest neighbor classifier is constructed and the vote decision strategy is used to determine the recognition result.