三网融合下基于压缩感知的半参考视频质量评估方法研究

摘要

针对三网融合条件下的视频质量模型、评估体系和设计方案,提出了一种基于压缩感知的半参考视频质量评估方法.该算法首先建立视频质量模型,并通过压缩感知算法获取视频图像的感知系数；然后利用视频图像的稀疏边缘分布去表征不同图像失真模型的相关性；最后利用一个信息距离测度测量这些概率分布并进行量化.在标准测试图像库上的实验验证了该算法的有效性和对于不同失真模型的适应性,特别是更适合于运营商在多屏、多制式、多终端、多分工界面等特点的三网融合复杂环境下的应用.%Toward the video quality assessment problem in the field of triple-play,this paper proposed a semi-reference video quality assessment method based on compressed sensing.The algorithm first established the model of the video quality and obtained the perception coefficient of video images by compressed sensing.Then by using the sparse marginal distributions of video images,it characterized the correlation of different image distortion models.Finally,it used a distance measure of information to measure the probability distributions and quantify them.Experiments on a standard test image library confirmed the effectiveness of the algorithm and the adaptability for different distortion models.Moreover,it was more suitable for operators in a complicated environment application of multi-screen,multi-system,multi-terminal,multi-division interface in triple-play.