人眼视觉特性压缩图像质量评估方法  

HVS Based Compressed Image Quality Assessment Method

作　　者：田金沙[1] 韩永国[1] 吴亚东[1,2] 赵小乐[1] 

机构地区：[1]西南科技大学计算机科学与技术学院,四川绵阳621010 [2]西南科技大学核废物与环境安全国防重点学科实验室,四川绵阳621010

出　　处：《西南科技大学学报》2015年第3期62-70,共9页Journal of Southwest University of Science and Technology

基　　金：国家自然科学基金(61303127);核废物与环境安全国防重点学科实验室项目(13zxnk12);中国科学院"西部之光"人才培养计划项目(13zs0106)

摘　　要：图像质量评价是图像和视频处理系统中的基本算法。通过对人眼视觉特性及图像的边缘特征的分析研究,认为梯度矢量更能反映图像边缘的本质特性,结合人眼视觉的多尺度特性及视觉感兴趣区域的加权,提出了一种基于梯度矢量的压缩图像质量评估方法。仿真实验表明,该评估方法符合人眼的视觉特征,与主观评价结果具有更好的一致性,可广泛应用于JPEG和JPEG2000压缩图像质量评估。Image quality assessment is fundamental importance for image or video processing system.Through the research and analysis human visual system features and the edge feature of image,this paper consider that the gradient vector better reflect the essential characteristics of image edge feature. Combined with the multi- scale feature and weighted interest region of human visual system,a new compressed image quality assessment method based on gradient vector is proposed. The proposed method is simple in calculation. Simulation results upon image databases show that the evaluation method is in accordance with characteristics of human visual system and consistency with subjective evaluation results better. In addition the method can be widely used to assess the quality of JPEG and JPEG2000 compressed images.

关 键 词：图像质量评估 人眼视觉(HVS) 压缩图像 梯度 JPEG JPEG2000 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]