基于梯度信息的图像质量评判方法的研究

图像质量客观评判标准广泛应用于图像处理中,基于人眼视觉系统的客观评判方法一直是图像处理领域的研究热点.最近,Zhou Wang等人提出了基于结构信息的评判方法—结构相似度(SSIM),它的理论基础是人眼视觉系统能高度自适应地提取场景中的结构信息.SSIM方法简单、评判性能优于PSNR(或MSE).但随着研究的深入,我们发现SSIM算法存在着一些问题,特别是不能很好地评判模糊失真类的图像.本文提出两种基于梯度信息的图像质量评判方法—基于梯度的结构相似度(GSSIM)和基于边缘的结构相似度(ESSIM).实验结果表明,GSSIM能更好地符合人眼视觉系统特性,而ESSIM对于模糊图像则取得了最好的评判效果.

Gradient Information Based Image Quality Accessment

Objective quality assessment has been widely used in image processing for decades and many researchers have been studying the objective quality assessment method based on Human Visual System(HVS).Recently,the Structure Similarity(SSIM) is proposed by Zhou Wang,under the assumption that the HVS is highly adapted for extracting structural information from a scene,and simulation results have proved that it is better than PSNR(or MSE),furthermore SSIM is very simple.By our studying deeply,it is found that it has some deficiencies in its arithmetic method,and fails in measuring blurred images.Based on this,we develop two improved methods which are called as Gradient-based Structural Similarity(GSSIM) and Edge-based Structural Similarity(ESSIM).Experiment results show that GSSIM is more consistent with HVS,while ESSIM gets the best performance for blurred image.