基于视觉感兴趣区的视频质量评价方法  

Method of Video Quality Assessment Based on Visual Regions-of-interest

作　　者：卢国庆[1] 李均利[1] 陈刚[1] 章颖[1] 满家巨[2] 

机构地区：[1]宁波大学信息科学与工程学院,宁波315211 [2]湖南师范大学,长沙410083

出　　处：《计算机工程》2009年第10期217-219,共3页Computer Engineering

基　　金：国家自然科学基金资助项目(60672072);浙江省自然科学基金资助项目(Y106505);湖南省自然科学基金资助项目(08JJ3130)

摘　　要：针对基于边缘结构失真的图像质量测量方法不能很好地应用在视频领域和对HVS考虑不充分的缺陷,提出一种基于视觉感兴趣区的视频质量评价方法,包括考虑HVS特性、场景结构信息和运动信息。实验结果证明,该方法在压缩、模糊方面明显好于PSNR,MSSIM和ESSIM的评价方法。Aiming at the the disvantage that the video quality assessment based on structural distortion measurement can not apply directly on the video field and think over HVS, this paper proposes a method of video quality assessment based on visual regions-of-interest. It thinks about HVS, the structural distortion measurement and motion information. Experimental results show that the method is better than the method based on PSNR, MSSIM and ESSIM on the compress losing, smooth losing.

关 键 词：视频质量评价 人眼视觉系统 运动信息 

分 类 号：TN949.6[电子电信—信号与信息处理]