Face recognition based on correlation of wavelet transform images

Face recognition is a very challenging research topic because variations can be introduced to the pattern of a face by varying pose, lighting, and expression. A new face recognition approach using correlation of wavelet transform images is presented. Using this technique, wavelet transform is applied on images before using images correlation for matching the face representation. The representation used for each face is computed from the mean wavelet transform images and the testing image. The computation time required for recognition is very short. Recognition rate is 98.5% using the ORL database using 5 training images and 5 testing images per person.