Research Of Image Quality Assessment Based On Structural Similarity

As digital images are subject to a wide variety of distortions during image processing application, it is necessary to develop objective image quality metric to evaluate the degradation automatically.Considering images are prepared for human eyes,the result of index must consistent with visual results.This paper analysis proposed evaluation algorithms ,human visual systemand two aspects of quality assessment flow ,then presents two improved structural-similarity based image quality assessment: metric of region-partitioned based on wavelet (WR_SSIM) and contend-based weighted metric(WC_SSIM).WR_SSIM considers some characteristic of human visual system, presents a wavelet domain image similarity measure.It make use of the coefficents to construct the SSIM index ,then use the vertical and horizonal coefficents to paritition the region. The new index can evaluation variety of distortions more efficently.WC_SSIM consider another aspect of the evaluation flow—the spatial strategies, assign variety weight according to image contents which have diffient sensitivities to human eys, It revise the SSIM index would be under zero with some condition. The new weight function based on 'arcctg' curve, consider the situation that significance for eyes of those region that have great variance are almost the same, it means the value of weight function is finite which is more consistent with human visual system. and more acceptable.Experimental comparisons with PSNR, SSIM, MS_SSIM on LIVE Quality Assessment Database shows that newly prorosed index can achieve high performance when it comes to variety types of distortions...