A Blurred Image Quality Assessment Method Based on Content-Partitioned

The theory of structure similarity (SSIM) is a new idea to evaluate the image quality by simulating the function of the human visual characteristics. Because the structure similarity model derived from this theory is very simple, it is widely used. However, the SSIM model only considered that human vision system can only extract the structure information of the image. This model is too limited. So the evaluation result also has a very big disparity compared with ideal results. In this paper, on the basis of deep understanding of SSIM, we proposed a blurred image quality assessment method based on the content-partitioned by combining with image content-partitioned and the gradient information. The prediction results of the method are consistent with the subjective evaluation, and the evaluation results are good.