Local and global feature extraction for face recognition

摘要: This paper proposes a new feature extraction method for face recognition. The proposed is based on Local Feature Analysis (LFA). LFA known as local recognition since it constructs kernels which detect structures of face. It, however, addresses only image representation and has problem In the paper, we point out propose by modifying LFA. Our consists three steps. After extracting using LFA, construct subset kernels, efficient Then combine to represent them in more compact form. results bases have compromised aspects between eigenfaces images. Through experiments, verify efficiency our method.