无参考视频传输质量评价模型

Null-reference video transport quality evaluation model

根据受众对光源的各类察觉特征,将物理刺激量全域分为多个部分,并依据人眼视觉暂留效果分别映射到感知心理量的区段,以此修正了Weber-Fechner对数法则。并在此基础上提出一种宏块丢失比率与主观保真度的映射模型,该模型以受众的视敏度为依据,利用视星原理计算在最佳观测位置处可察觉的最小失真程度和最小未失真区域,不但体现了失真现象在人眼中的暂留效果,而且反应了损伤在时空中的传播,实现了对损伤全域的评测。模型计算代价小且可部署在任意网络节点中,可得到更贴近受众感受的视频质量。

According to the various characteristics of light source that can be perceived by the audience, the whole domain of stimulation is divided into multiple parts. Each part is mapped to the perception scales by means of the effect of visual staying phenomenon. Thus, in this paper the Weber-Fechner＇s law is improved first. Then, a mapping model is proposed on the MOS and loss ratio of the macroblock. The model calculates the minimum distortion and undamaged area of which audience awarded by standing at the best observation position ahead of the monitor. It achieves an evaluation of the full field damage by not only reflecting the effect of impairment, which can visually stay in the eyes, but also indicating the spatial extent and temporal propagation of loss. Experiments show that the proposed method can be deployed in any network nodes for its low computational complexity and achieve better video quality in close touch with the audience.