符合人眼视觉特性的视频质量评价模型

Video Quality Evaluation Models Based on Human Visual Properties

视频技术的发展为其质量评价提出了新的课题 ,但由于评价图象质量的关键在于所用视觉模型是否符合人的感知特性 ,因此评价图象质量必须考虑以视觉锐度、对比度敏感度、多通道结构和掩盖特性为基础的人眼视觉特性 (HVS) .为了使人们对基于人眼视觉特性的视频质量评价模型研究现状有所了解 ,介绍了几种目前比较成功的基于 HVS的视频质量评价模型 ,并分析和总结了它们的性能 。

Picture quality evaluation methods could be utilized to assess image algorithms' performance. The evaluation of image quality includes subjective evaluation and objective test. The traditional subjective evaluation depends on observer's experiences and motivations, and therefore it is complex and its results always change. With the development of the video compression technology, it is very urgent to research new video quality evaluation method. The key point of the video quality evaluation is that its model should match the perceptive characteristics of human. Based on the visual acuity, contrast sensitivity function, multi channel structure and masking effect, this paper summarizes main properties of Human Visual System (HVS) taken into account in the assessment of the video quality. It also presents several comparatively satisfactory video quality models based on HVS, including spatio temporal model, RAI model, three dimension weighted signal to noise model and three layered bottom up noise weighting model. Their performances are also analyzed respectively. The perspective of video quality evaluation is given finally.