A multi-resolutional face verification system via filter-based integration

The authors build up a face verification system via multi-resolutional features to identify individuals. In such a system, three fold requirements, including personalization, integration, and adaptive thresholding, should be considered in order to to achieve more accuracy. In order to satisfy the above requirements, R personal feature spaces, an optimal positive Boolean function, and an automatically determined threshold value are obtained from the training samples of a specified individual. Some experimental results are conducted to show the validity of the approaches.