基于结构相似度(SSIM)的图像质量评价方法是从视觉区域提取图像的结构性信息,但在评价模糊较严重的图像时存在其局限性,因此本文将图像边缘和SSIM相结合,提出了基于多尺度边缘结构相似度(MESS)的图像质量评价方法。实验结果表明,由于MESS考虑了边缘信息对于人眼感知结构信息的重要性,评价结果比SSIM更加符合人眼视觉感知特性。
The structural similarity(SSIM) based image quality assessment extracts the structural information from a visual region, but the evaluation result is limited especially when the image is vague. Therefore,an image quality assessment based on multiscale edge structure similarity(MESS) using image edge combined with the structural similarity is presented. The results show that since the importance of edge information for the human eyes is considered in perceiving structural information,the MESS is more accordant with the characteristics of human visual perception than SSIM.