Face Recognition Platform Design And The Complementation Of Face Recognition Algorithms

Face recognition is one of the hot spots in image processing and pattern recognition. It has advantages over other bio-tech recognition methods, and applications of face recognition system have emerged.This thesis mainly discusses the human face recognition algorithm based on Gabor features, and the construction of face recognition test platform. First we discuss basic concepts of human face recognition and the mathematical features of Gabor wavelet and the corresponding facial feature vector extracting methods are also discussed. We put forward an extended optimal Gabor feature sampling algorithm. It combines the method of image masking, energy analizing and key-point selecting to determine which features in face images are helpful for recognition. Then optimal Gabor feature selection is used to get most useful information from the feature vectors and meanwhile, improves the performance of the algorithm. On analysis of the problems we met during recognition, the corresponding solution for improvement, which partly avoids the affect of low quality input images are described. Experimental results are generated from standard database of FERET. The comparisons and analysis with other algorithms are given. The experimental results show that our method is superior to other Gabor wavelet based algorithms.Another topic to be discussed here is the design and implementation of our face recognition test platform. The platform offers an environment of a uniformed poll for achieving and testing face recognition algorithms. It supplies researchers with experimenting tools, and algorithm integrations. When planning a certain method, we keep to the interface...