An Image Quality Assessment Algorithm Based on Dual-scale Edge Structure Similarity

The quality assessment of images is meaningful for most video and images applications. Recently a new metric called structure similarity index (SSIM) is proposed for assessing image quality with better reflection to human visual characteristics than traditional PSNR metrics. However, SSIM exhibits some defects in evaluating the quality of blurred images and Gaussian noise images. This paper proposes a new image quality assessment algorithm based on dual-scale edge structure similarity (MDESSIM) to the issue above. With different from SSIM, the algorithm sufficiently considers image edge information within the structure information, and includes the distortion metric of edge structure in the image quality assessment. Experiments demonstrate that MDESSIM maintains better consistency with human subjective perception for wider kinds of images.