Personal identification by multiresolution analysis of lifting dyadic wavelets

This paper proposes a novel method for identifying persons by multiresolution analysis of lifting dyadic wavelets. Our method consists of three procedures: face learning, detection and identification. In the learning procedure, new highpass filters for capturing facial parts are constructed by tuning free parameters in the lifting scheme. By using the learned filters, human faces can be detected from each of video frames. A person whose face is detected in a maximum number of frames is identified as a target person. Experimental results show that our personal identification algorithm is fast and accurate.