Face Recognition Study Based On Subspace Analysis

Because of face recognition huge applications in secure authentication systems, credit card verification, medical, archive management, video conferencing, human-computer interaction, the public security system (to identify criminals, etc.), it will increasingly become the study hotspot in current pattern recognition and artificial intelligence field.Face recognition is a technique which making use of computer processing and analysing face picture, then extract feature that has identifying information. Many feature extraction methods have been proposed and among them the subspace analysis has received extensive attention owing to its appealing properties. Now the subspace analysis method has been the most popular technology for feature extraction and face recognition. This thesis studies the face recognition methods of static pictures, especially feature extraction based on subspace analysis.The main research contents include the following aspects:(1) At first, research the basic principles and the realization of the process of the traditional PCA face recognition method, and analyse the advantage and disadvantage of PCA. Considering the traditional PCA method in dealing with the issue of face recognition is the need to transform the image from two-dimensional matrix for the one-dimensional vector, it has long been the high computational complexity of the problem. In recent years, a direct image-based methods 2DPCA matrix, and its advantage is greatly accelerated the speed of feature extraction and recognition rate have also improved. Through experiments gives way to the performance of 2DPCA compared PCA , 2DPCA method is confirmed not only to avoid a large amount of computation, but also increased the recognition rate.(2) Secondly, because of the light sensitive, face recognition rate is relatively low when the light variation. Making use of Gabor wavelet could express the facial local texture feature, a combination of Gabor wavelet and principal component analysis is advanced. This method is good to combine the advantages of both, the experimental results show that the recognition rate is significantly improved.(3) At last, Linear Discriminant Analysis is introduced in feature extraction. LDA is aim to make the sample divided easily, in theory this method is superior to PCA. An improved 2DLDA is presented: Two-dimensional Linear Discriminant Analysis based on image module. Firstly, in the method, the face images through some treatments are divided into block images, which are called sub-images. Then the 2DLDA method is directly employed to the sub-images. Lastly, the recognition results are obtained by the general minimum distance classifier. The advantage of the method is that it can extract part feature of image , which is helpful to discriminate image; dimension reduction of features can be done conveniently. It is based on image matrix, feature extraction easily. A series of experiments are performed on the ORL face image database, the experimental results indicate that the performance of M2DLDA is superior to that of 2DLDA and LDA.