Video-based face recognition using image averaging technique

Image averaging is equivalent to low-pass filtering and can remove the misleading shadows or occlusion. This paper introduces the image averaging technique to video-based face recognition. The image averaging technique provides an efficient way to perform decision-making over the testing frames in video-based face recognition. Furthermore, the average texture image is morphed back to the original shapes and the re-established faces derive, which remove the misleading information for recognition to some extent. Serving as gallery images, the re-established faces improve the recognition accuracy. The experimental results testify the efficiency of the proposed algorithm.