GLCM Based LDA for Human Face Recognition

Face recognition system is proposed in the present work depending on the grey level cooccurance matrix (GLCM) based linear discriminant analysis (LDA) method. The GLCM is used to extract eleven effective textural features for the face image, while the LDA is used to discriminant these faces between each other depending on the extracted features. The proposed method requires create some database models, each for specific face. These face models are used to be compared with test ones that input to the recognition procedure. The newly proposed idea is the use of features instead of image pixels in the LDA covariance matrix. The proposed face recognition method consists of two phases: enrollment and recognition. Different face samples are input to the enrollment phase in order to collect the average features of that face, and store them in a database to be comparable models for the recognition phase. The recognition phase compares the extracted features of the unknown test face with that stored in the database. The comparison indicates the similarity between the test face with the database models, which is used to make the recognition decision. The results of frequent tests showed that the used face recognition was about 96%, which ensure the efficiency of the used face recognition method.