Study On Image Processing For Human Face Recognition Based On The Second-Generation Wavelet

The technology of human face recognition was an active subject in the areas of image processing. With the strong development of biometrics identification technology, this dissertation by using modern information processing methods study the human face image recognition. This study not only an exploration in the image processing domain, but also the demand for application of society. This dissertation concentrates on the analysis and designing of algorithms for human face image pre-processing, feature extraction and recognition, using wavelet transform and the theory of High-Dimension space geometry. Some new image processing methods were proposed in this dissertation, and experimental results show that these methods were efficient. The main work was summarized as follows:Firstly, combine wavelet theory with the practice of human face recognition, study the effective usage of the second-generation wavelet transform for human face feature extraction. Then, two keys were proposed: the selection of wavelet bases and the wavelet decomposition levels, and the guard-line were given by experiments. Secondly, a fast algorithm of edge detection for human face images basic on the second-generation wavelet multiscale multiplication was proposed. Compare with the classical edge detection algorithm was more efficient, the simulation shows that the overall recognition rates were satisfied, the method was robust to the facial expressions and glasses.Thirdly, a new human face image pre-processing approach was presented—the theory of High-Dimension space geometry for image enhancement. Through studying the vector and the position between the points in High-Dimension space to study the relationship between the images. The simulation shows the method was efficient and correct.At last, the research work was summarized and foresights were given.