PERSONAL IDENTIFICATION USING LIFTING DYADIC WAVELET FILTERS

This paper proposes a new method for identifying persons by multiband lifting dyadic wavelet filters. Our method consists of three procedures as follows: face learning, detection and identification. In the learning procedure, new highpass fil-ters for capturing facial parts are constructed by tuning free parameters in the lifting filters. By using the learned filters for target persons, a face is detected in each of video frames. A person who has the most number of the same face among detected faces in each of the subbands is identified as a target person. Finally, experimental results are reported our per-sonal identification method to be fast and accurate. 1