Face authentication across age progression using shifted block matching

A method of personal authentication using face images is proposed for the situation that the person´s face image was registered in his/her childhood and a large change exists between the captured face image and the registered one because of the person´s growth. Here, the positions of the facial elements are different between the two images, but still some similarity exists for the same person. In order to avoid such influence from physical growth, authors proposed before a block matching method which can search a similar facial area in the captured face image, although its position is a little different. However, the performance sometimes gets degraded when the characteristic parts of the face are not appropriately contained in each template block. In this paper, a method of face recognition using shifted block matching is proposed in order to solve the problem of the previous method. The proposed method adjusts the position of the template blocks, so that the characteristic facial parts can be appropriately contained in each block. The experiment results show advantage of the proposed method to the conventional ones.