Research, Face Recognition Based On Gabor Wavelet Transform

With the complicate of social public order and the rigorous situation of international anti-terror,the upcoming of Beijing Olympic Games.National security and city safety have been focused increasingly.System security is on the necessary premise of making our personal identification accurately.Faces are principal organ which are strongly different from others,It is the most natural and direct tool to analyze face images with computer to get effective recognition information.Face recognition can be applied in many aspects,such as criminal recognition of police system,monitor system of bank and custom etc,and also has a bright application future in the field of medicine,video meeting ,family entertainment etc.This paper mainly studies in two parts of face recognition:feature extraction and pattern classification.For feature extraction,this paper first implement two typical methods:PCA and Fisher Linear Discriminant Analysis,analyzes and experiments with them.Then mainly probe into the feature extraction method of Gabor wavelet transform.Its advantage is that Gabor wavelet function can describe the resceptive field simple cells in the visual cortex of human cerebra correctly,so it can extract the face feature effectively.With this advantage,this paper combines Gabor wavelet transform,PCA and Fisher Linear Discriminant to extract the face feature.First getting Gabor wavelet transform feature by Gabor wavelet transform to face image,then applying downsampling and PCA methods to reduce the dimensions of Gabor feature vector,at last applying Fisher Linear Discriminant Analysis to obtain the best projection direction for classification purposes.Projecting the face image onto the transform feature base,transform coefficient can be used to be the input of classifier.About classifier,this paper studies and implements K-NN classifier algorithm,then in order to impove the effect of classification,this paper firstly applies a classifiers combination method which uses Modified Manhattan Distance measurement for first classification and Angle-based Distance measurement for second classification.Then this paper refers to a classifers combination strategy which uses Weighed Angle-based Distance measurement for first classification and Modified Manhattan Distance for second classification .At last,we compare the two classifiers combination ,and also compare the two K-NN classifiers combinations with the NN classifier.Experimental results show that the classifers combination based on Gabor wavelet transform which uses Weighed Angle-based Distance measurement for first classification and Modified Manhattan Distance for second classification gets 99.5% average recognition rate in ORL standard face database.