边缘信息奇异值分解的图像质量评价方法

摘要

The traditional image quality assessment algorithm can not meet the actual requirement since the visual information and the essential features of the image are not taken into consideration. We studied the image edge information extraction and singular value decomposition method, and proposed a new method for objective quality assessment, which can give consideration to both of the information. The singular value information of the image shows the essential information of image and human eyes are sensitive to the edge information of image. Theoretically, the algorithm of image quality assessment based on edge information and singular value decomposition is better than traditional methods. The simulation experiment results show that:compared with the tradition methods of SSIM, MSE and PSNR, the proposed algorithm is more consistent with human visual system, and has better correspondence between subjectivity and objectivity.%针对传统的图像评价方法没有同时考虑图像的视觉信息和基本特征,不能满足实际需求的问题,在研究了图像边缘信息提取和奇异值分解方法的基础上,提出了一种能兼顾这两种重要信息的图像质量客观评价方法.人眼对边缘结构信息相对比较敏感,图像的奇异值反映了图像的一些重要结构信息,因此,从理论上讲,基于边缘信息奇异值分解的图像质量评价方法优于传统的图像评价方法.仿真实验表明,与基于结构相似度(SSIM)评价方法以及均方误差(MSE)、峰值信噪比(PSNR)等传统方法相比,该算法的主、客观一致性更好,更加符合人眼的视觉特性.