基于图像信息相关性的质量评价方法  

Quality Assessment Based on Image Information Correlation

作　　者：李民权[1] 朱大龙[2] 明军[1] 

机构地区：[1]安徽大学计算智能与信号处理教育部重点实验室 [2]光电信息获取与控制教育部重点实验室,安徽合肥230039

出　　处：《安庆师范学院学报（自然科学版）》2007年第2期21-23,共3页Journal of Anqing Teachers College(Natural Science Edition)

基　　金：安徽省教育厅重点项目(KJ2007A080)

摘　　要：图像质量评价是图像工程的重要内容之一,在图像通信过程中起着举足轻重的作用。评价方法一般来说可分为客观质量评价方法和主观评价方法。文中分析了传统的MSE和MOS图像质量评价方法及其一些缺点,在分析图像信息和相关性理论基础之上,通过对基于结构失真的图像质量评价方法的深入分析,提出了一种对结构失真评价方法进行改进的图像质量评价方法,实验证明所提出的改进方法与人的主观视觉质量具有一致性。Image quality assessment is one of the important tasks in the field of image project. There are some crucial effects in image communications. Generally, the measures of image quality assessment have objective quality assessment measure and subjective one. In the paper, we introcluee and point out some deficiencies of conventional measures such as MSE, MOS. Based on the analyzed image information and correlation theory and by analyzing quality assessment profoundly for the image with structural distortion, we propose a ameliorative one for the structural distortion measurement, and verify that the proposed measure has consistent with human visual quality by experiment.

关 键 词：结构失真评价(SSIM) 图像信息相关性 图像质量评价 均方误差(MSE) 

分 类 号：TN911.73[电子电信—通信与信息系统]