The Research Of Weighted Face Recognition Method Based On Wavelet Transform

Face recognition technology is a complex and difficult problem that is important for surveillance and security, telecommunications, digital libraries, video meeting, and human-computer intelligent interactions, but FRT is still facing a very big challenge in practical application, because every face image is similar, the varieties of expression , gesture, hair style, especially the change of expressions., which will bring huge trouble for face recognition. How to identify a large number of people correctly and meet the real-time requirement is an urgent question to be resolved. In allusion to this, wavelet transform was applied to eigenface recognition method.As the earlier period of the recognition, face detection is very important. This paper first describes some main current methods of face detection. Images preprocessing is an essential foundation for the whole system. Face recognition would not go on wheels without a reliable preprocessing. Several methods of images preprocessing are given in the paper, including image normalization, light compensation, gray-scale transformation, binary image, image denoising, edge detection.In allusion to insurmountable expression effect in face recognition, wavelet transform was applied to eigenface recognition method. It is a method that through separating the face image into three parts, and every part was decomposed using wavelet transform to make the image dimension reduced. Then, the horizontal high-frequency and low frequency subband was selected, vertical and slash high-frequency subband was removed,and horizontal high-frequency and low frequency subband of every section in divided human face library were parallel processed using eigenface recognition method. Ultimately, the face recognition result was gained through weight—adding summed up, we take the face image with the best effect as the result . We have compared the weighted eigenface method with the eigenface method and the result shows the feasibility and robustness of this algorithm.