摘要
基于结构相似度的视频质量评价方法(VSSIM)结构简单,性能优于峰值信噪比(PSNR),但是在研究中发现VSSIM不能很好地评价严重模糊的低码率视频。该文提出一种基于频域的结构相似度的视频质量评价方法。该方法将频域信息作为主要的结构信息,根据人眼对不同频率分量的敏感程度不同,对离散余弦变换后的各频率分量加权得到频域函数。由频域函数、亮度函数和对比度函数经过视频内容的加权计算得到结构相似度。实验结果表明:该方法比VSSIM和PSNR更符合人眼视觉系统特性,能较好的评价视频质量。
Although the structural similarity based video quality assessment VSSIM (Video Structural SIMilarity) is simple and has been proven to be better than the PSNR (peak signal to noise ratio), there are still some difficulties in assessing badly blurred low bit rate video. This paper presents a frequency domain based structural similarity standard for video quality assessment. The method uses frequency domain information as the main structural information and calculates a frequency domain function for the frame from weighted frequency components after a discrete cosine transform based on the human eyes with different sensitivities in the frequency domain. The structural similarity is calculated using the frequency domain, the brightness, and the contrast functions with video content based weights. Tests show that this model is more consistent with the human visual system and can assess the video quality more precisely than the VSSIM or PSNR.
出处
《清华大学学报(自然科学版)》
EI
CAS
CSCD
北大核心
2010年第1期121-124,共4页
Journal of Tsinghua University(Science and Technology)
基金
国家自然科学基金资助项目(60572081)
关键词
视频质量评价
结构相似度
频域
人眼视觉系统
video quality assessment
structural similarity
frequency domain
human visual system (HVS)