Assessment Of Super-Resolution For Face Recognition From Very-Low Resoution Images

Super-Resolution (SR) involves the registration of multiple images/frames and reconstruction of a single higher resolution image. The goal of this research is to use multiple, very-low resolution images, such as those produced from a video sequence in a wireless sensor network system, as input to the super-resolution process in a face recognition system. The algorithm used for face recognition is the Fisherfaces method with a nearest neighbor classifier used for the recognition decision. Super-resolution consists of two stages, a registration stage and a reconstruction stage. The testing images were segmented using a simple skin color detection approach. After the cropping, the images were combined into groups of four from a sliding window that would take the current image and the following three images repeating this process by moving to the next image in the sequence and the subsequent three images until the end of the current class, or person, is reached. This same sliding window was used for the super-resolution algorithm using faces from the three people or classes. Each group of four images was used as an input to the Keren registration algorithm where the rotational and translation information was saved that was then entered into the robust super-resolution reconstruction algorithm to create a single high quality image, which was processed by the face recognition algorithm. The methods tested to compare were the average of the same groups of four, the centroid shifted average and the minimum of the four faces in the group. The comparison was based on nearest neighbor classifier and on classification rates. The results were not in favor of the super-resolution method but instead, the centroid shifted average was the best in this study