基于视觉特性的视频噪声估计研究  

Investigation on video noise estimation based on perceptual metrics

作　　者：陈吉林[1,2] 赵飞[1] 孙成明[1] 杨新[1] 

机构地区：[1]中国科学院光电研究院,北京100094 [2]中国科学院大学,北京100049

出　　处：《计算机工程与设计》2014年第6期2037-2041,共5页Computer Engineering and Design

基　　金：国家自然科学基金项目(61308101);中国科学院光电研究院创新基金项目(2011);中国科学院光电研究院自主部署研究生创新基金"雏鹰计划"课题基金项目(Y30B02A18Y)

摘　　要：为了获得与主观视觉感知高度一致的视频噪声估计结果,提出了基于人眼视觉特性的噪声估计算法。分析了人类视觉系统特性,深入考虑了视频连续帧间图像的结构相似性和噪声同分布性,建立了人眼视觉感知的噪声估计模型,将视频连续帧的图像变换到人眼感知域,进行差分计算以估计出视频噪声。进行了实拍视频噪声估计实验,采用上述算法与传统算法分别估计并对比结果,验证了基于视觉特性的视频噪声估计算法具有更高的主观一致性,在噪声强度极低的情况下仍适用性良好,应用前景比较广阔。To obtain a video noise estimation result conforming to humans＇ subjective estimation,a perceptual metrics based noise estimation algorithm was presented.By analyzing the characters of human visual system （HVS）,considering the structure similarity and noise independence and identically distributed property between the two sequential frames,a mathematic model was constructed to simulate the HVS＇s perceptual model,then the video frames were transformed into HVS pattern,the estimation was based on the difference between the transformed sequential frames.Lastly by estimating several of videos＇noise and contrasting to traditional methods,the obtained outcomes were better correlated with human＇s subjective estimations,and the algorithm worked well when the images were just slightly contaminated.

关 键 词：人类视觉系统 视觉特性 视频噪声估计 对比度敏感度 掩盖效应 

分 类 号：TP37[自动化与计算机技术—计算机系统结构]