Simultaneous Blurred Face Restoration and Recognition

Blurred face restoration and recognition are closely related problems. Lots of existing works have proven that these two tasks contribute to each other, but only a few consider combining them together. In this paper, we propose Simultaneous Restoration and Recognition (SRR) method by iteratively dealing with these two tasks. Two new models are presented and then integrated into a two-staged method to handle the problems of the state-of-the-art exemplar based restoration method. The better-performing blurring method is chosen for recognition. Experimental results on FERET database demonstrate the efficiency of SRR in terms of restoration qualities and recognition accuracies for various blurs.