Fast no-reference image sharpness measure for blurred images in discrete cosine transform domain

Abstract

Researchers working in the field of image quality assessment are constantly developing algorithms for assessing the quality of a given image. This is a very challenging task as the given image may be affected by different types of distortions. One of the most common distortions is image blurring. In this paper, we propose a no-reference image quality assessment algorithm to assess the quality of images corrupted by blurring in Discrete Cosine Transform domain. The proposed algorithm works faster than the state of the art algorithms for the same problem, and the score computed by the algorithm has more correlation with the human opinion score than existing methods which means that the proposed image sharpness measure mimicks the human visual system closely.