An adaptive restoration of spatially varying blurred images based on local edge information

We propose an efficient adaptive restoration of the spatially varying blurred images based on local edge information such as variance and contrast in local image block without a priori knowledge, Based on the fact that the regions of sharp transition in an image such as edges and lines undergo maximum intensity change when an image is blurred, the variance and contrast, which depends on the blur measure defined in this paper, are derived using variable sized local window. Then the images are restored adaptively by increasing local contrast and decreasing local variance depending on the blur measure of each block. Experimental results for various spatially varying blurred images show that the proposed algorithm is very efficient in adaptive restoring of the spatially varying blurred images.