一种基于视觉特性的图像质量评价指标  

An Image Quality Assessment Based on property of HVS

作　　者：杨迪威[1] 沈远彤[1] 李宏伟[1] 熊超[1] 沈永珞[2] 

机构地区：[1]中国地质大学数理学院,武汉430074 [2]中国地质大学机电学院,武汉430074

出　　处：《信号处理》2011年第11期1691-1695,共5页Journal of Signal Processing

基　　金：国家自然科学基金(11026145;61071188)资助项目

摘　　要：在分析多通道结构、遮掩效应、敏感度带通特性等人类基本视觉系统特性(HVS)的基础上,提出了一种新的图像质量评价指标的构造方法。指标构造先借助多尺度几何分析(MGA)中的Contourlet变换对图像进行子带分解以此模拟视觉多通道结构特性,然后在各子带中使用一种视觉遮掩效应模型获取失真图像和参考图像之间的视觉误差,最后通过对比度敏感函数(CSF)取得不同子带的视觉误差加权系数,从而得到图像质量评价值。本文在具有5种类型降质图像的LIVE图库进行性能比较实验,结果表明本文所设计指标比峰值信噪比PSNR、结构相似度SSIM算法具有更好的稳定性和主客观评价一致性;同时试验表明通过对比度敏感函数(CSF)获取的加权对本图像质量评价指标改进较小。A novel image quality assessment algorithm based on property of HVS was proposed to evaluate the quality of image The algorithm exploits the important property of HVS such as multi-channel structure,masking effect and band -pass property of contrast sensitivity.The proposed algorithm can be summarized to three steps;firsdy,the muli-channel behavior of HVS can be simulated after decomposing the original image to subband images by contourlet transform of Multiscale Geometric Analysis（MGA）.Secondly,a model is employed to evaluate the visual error between the reference image and the distorted image in each decomposed subband.Finally,the image quality evaluation results are calculated from the visual error weighted factor which is the comparing results of CSF in various sub-band. The assessment was tested by using the LIVE image database.The result shows that the proposed algorithm is better than PSNR and SSIM in stability and it correlates well with the judgment of human observers.

关 键 词：图像质量评价 人类视觉系统(HVS) CONTOURLET变换 对比度敏感函数(CSF) 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]