Subspace-based Face Recognition Technology

Face recognition continues to be a hot topic in pattern recognition field due to its wide range of applications such as commercial and law enforcement area. A central issue to a successful approach for face recognition is how to extract discriminant feature from the facial images. 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 investigates the use of subspace analysis for feature extraction from the facial images and recognition.1. The latest development of face recognition methods is briefly introduced .2. This thesis gives a detailed analysis on singular values(SV's) of face images and then reveals why the SV's don't contain enough information for face recognition. Based on this observation, a new singular value's feature extraction method is appealed in this dissertation.3. Principle component analysis(PCA) approach is deeply investigated here. PCA approach can approximate the original data with lower dimensional feature vector. Based on this characteristic, we use it to face recognition. In this approach, choice of feature vector of sample's covariance matrix and distance measure criterion are also discussed deeply.4. The technique of face recognition based on Fisher linear discriminant are analyzed and studied. Then a subspace method based on PCA and Fisher linear discriminant is proposed here.5. A new approach-two dimensional principal component analysis(2DPCA) is developed for face recognition. 2DPCA approach don't need to transform the image matrix into feature vector. It directly uses the image matrix to construct the sample covariance matrix. But 2DPCA need more coefficient than conventional PCA, so a modified 2DPCA is proposed in this thesis. From the experiment result, the modified 2DPCA needs less recognition time than 2DPCA.