A Wavelet-Domain Structure Similarity for Image Quality Assessment

Recent studies have found that adoption of structure similarity index (SSIM) was successful in reflecting human visual characteristic better compared with traditional PSNR metrics. However this method shows some defects when evaluating the quality of blurred images and compressed degraded images at low bit rate. Good quality results were hardly achieved as they do not match the human vision system (HVS) well. This paper is to introduce a new image quality assessment algorithm based on wavelet-domain structure similarity (WDSSIM). The relative importance of the edge information under different scales is considered sufficiently in image quality assessment. The experimental results have demonstrated better consistency with the subjective perception for a large range of image types.