Blind Facial Images Deblurring Using Dark Channel Prior

In recent years, terrorism, accidents, violence thefts have increased, leads to install surveillance cameras everywhere in order to identify the people who have done the bad deeds. In most cases, the images are not clear to identify and know the people, so we have developed a system that helps to improve the image level that identify people. To obtain a latent image from a blurred image, effective regularizations are required. In paper, we propose a Contrast-limited adaptive histogram equalization Algorithm and dark channel of blurred images with Downsampling and gradient prior to improving blur kernel estimation (L0 regularized intensity.) and finally use filter to remove Noise. Experimental results demonstrate that the proposed method can better handle various complex face poses, as compared with state-of-the-art approaches.