Face verification using correlation filters and autoassociative neural networks

Face verification is the process of accepting or rejecting the identity claim of a person using information from his/her face. Representation of the face is an important issue in face verification. This paper propose edge gradient-based representation of face, for correlation-based face verification. The edge gradient based representation of face is obtained using one-dimensional (1-D) processing of the image, which has the advantage of providing multiple partial evidences for a given image. This representation of face is used to recognize the faces, which is performed by a specific type of correlation filter called minimum average correlation energy (MACE). Separate correlation filters are employed for each partial evidence. A method is proposed to combine the output of the filter using an auto-associative neural network (AANN) model to arrive at a decision to accept or reject the claim.