基于背景模型的AVS-S编码

为提高监控视频的编码效率，在AVS监控伸展档次的编码框架下，提出一种新的背景帧更新方法，即迭代更新均值法.根据某一像素在前景区域和背景区域出现频率不同的原理来确定该像素点在某一帧中的性质.通过在监控视频Crossroad，Overbridge和Classover的测试比较，得到不同性质的像素点对应的背景模型像素值.结果表明，慢速运动物体难以从背景中消除的缺点得到一定程度的改进，同时加强了背景信息的完整性和提高了背景帧的质量.在同等码率和相同峰值信噪比的条件下，相比于AVS基准档次方法平均节省码率为43.283%，平均提高峰值信噪比1.186dB，其监控视频的编码效率也得到了提升.

AVS-S Coding Based on Background Model

In order to improve the coding efficiency of surveillance video， a new background frame update method was proposed， namely， iterative update mean value method， under the coding framework of AVS surveillance extension grade. The nature of the pixel in a certain frame was determined according to the principle of different frequencies of occurrence of a certain pixel in the foreground area and in the background area. Through the test comparison of the surveillance video Crossroad， Overbridge and Classover， the pixel values of the background model corresponding to the pixels of different nature were obtained. The results showed that the shortcomings that slow moving objects are difficult to eliminate from the background are improved to a certain extent in this experiment. At the same time， the integrity of the background information is strengthened and the quality of the background frame is improved. Under the conditions of the same bit rate and the same peak signal-to-noise ratio， compared with the AVS benchmark method， the average code saving rate is 43.283% and the peak signal-to-noise ratio is increased by 1.186dB on average. The coding efficiency of its surveillance video is also improved.