H.264视频传输中一种改进的端到端失真估计模型

An improved end-to-end distortion estimation modle for H. 264 video transmission

视频传输中端到端失真的估计可为视频编码与传输过程采用更为有效的编码与传输策略提供依据。通过分析包含错误掩盖的H.264视频传输的端到端失真过程以及图像结构相似性估计(structural similarity index measurement,SSIM),给出了帧间错误继承系数估计值,并在此基础上推导出了一种结合一阶矩的端到端失真估计的递归计算模型。该模型可由原始视频序列特性及编码参数、传输系数给出端到端失真估计值,而无需统计编码器编码信息,进一步,通过引入失真反馈,可获得更准确的失真估计值。仿真实验表明,该模型在无反馈与结合反馈的情况下均能较准确地估计端到端失真。

The end-to-end distortion provides the gist of optimized strategy in video codec and transmission. This paper analyzes the end-to-end distortion system combined with error concealment. Then the coefficient of error propagation between frames is estimated by employing structural similarity index measurement （SSIM） based on the analysis results. Finally an improved end-to-end distortion estimation model is proposed. The proposed model could estimate the distortion with the characteristic of the original video sequence, parameters of encoding and transmission and no detailed encoding information is needed. Furthermore, with the information of previous real distortion feedback through transmission channel, the proposed model could estimate the distortion more precisely. The simulation results show that the proposed model could esti- mate the distortion very well with and without the feedback information.