Image Quality Assessment Based on Structure and Edge Similarity

The quality assessment of images is of fundamental importance for most image applications. Structural similarity measure (SSIM) has been widely used which is very easy to calculate but it has been proved not so good in some cases such as Gaussian noise image and blurred image. In this paper we present a novel objective image quality index which is also very easy to calculate and applicable to various image processing applications. To accord with human visual characteristics, we introduce weight based on edge image of the reference, and calculate the whole index based on edge structure similarity. Experiments demonstrate that the new assessment maintains better consistency with human subjective perception for wider kinds of images.