基于对象的监控视频编码方法研究

Surveillance Video Coding Algorithm Based on Object

针对监控视频存在大量视觉冗余问题,提出一种将目标和背景单独进行处理的视频编码方法。首先利用背景建模方法——ViBe算法将视频中的运动目标与背景分离,再对目标与背景以不同的码率分别进行编码。该方法可去除监控视频大量的背景冗余信息,在保证目标视觉质量的同时提高视频压缩率。以不同尺寸、不同内容的监控视频为对象进行测试,实验结果表明,与传统H.264编码方法相比,该方法可提高40-90%的视频压缩率,同时保持较好的视觉质量。

Focused on the problem of substantial visual redundancy in surveillance video,proposes an algorithm that processes the object and background separately for surveillance video coding.Firstly,the moving object and background in video are separated by using the background modeling algorithm,ViBe.In addition,the object and background are coded with different code rates respectively.This method can reduce a large amount of background redundancy information in surveillance video and greatly improve the compression rate of video while ensuring the visual quality of the object.Tested with different sizes and contents of surveillance videos,the experimental results demonstrate that the proposed method can improve the video compression rate by 40%to 90%compared with the traditional H.264,while maintain a good visual quality.