Abstract: The video quality impairments are mainly caused by noisy transmit channel. No-reference evaluating method is an important way to predict the video quality due to without any other loader to the transmit channel. Aiming to improve the accuracy and effectiveness of no-reference video quality evaluating model, a novel variable-weight evaluating model is proposed. Motion intensity is introduced to qualify the variation of video content. And a weight control function is obtained based on the statistic studying method. Then the weight of clearness and smoothness can be adjusted in real time through the weight control function. The variable-weight evaluating model takes into the human perceptual properties. The experimental results show that the variable-weight evaluation model outperforms the existing evaluation models and achieves high correlation with the subjective test results.

Key words: no-reference video quality metric, human visual system, motion intensity, variable-weight

摘要： 传输信道失真是导致视频图像质量损失的重要原因，而不需要任何附加传输信息的无参考客观评价是监测视频传输质量受损的主要方法。为了提高无参考客观评价模型的准确性和效率，提出模拟人类视觉特性的变权评价模型。变权评价模型综合考虑视频的空域和时域两类质量指标，引入运动强度来量化视频内容中的运动变化程度，根据统计学习的非线性回归法建立变权控制函数；通过变权控制函数动态调整清晰度和平滑度权值来模仿人类视觉特性。实验结果表明该变权评价模型与主观评价符合度高，优于现有评价方案。

关键词: 无参视频质量评价, 人类视觉系统, 运动强度, 变权