一种运动区域敏感的监控视频压缩算法

A motion-sensitive surveillance video compression algorithm

海量监控视频存储是目前监控行业面临的一个亟待解决的难题。文章针对静态背景的监控视频,提出了运动区域敏感的视频压缩算法。首先,针对视频帧,在运动目标检测的基础上结合运动信息和边缘信息,保证剪裁前后边缘能量代价最小并设计相应的代价函数,实现了对视频的单帧剪裁;其次,针对相邻视频帧,引入时空相关性原则并设计相应的代价函数,保证剪裁后相邻帧间的连续性。实验结果表明,该算法不仅对静态背景监控视频有较高压缩效果,还能保证兴趣区域的高质量和剪裁前后视频结构的形似性和完整性。

Mass surveillance video storage is a pressing problem in the industry of surveillance.A motion-sensitive compression algorithm for surveillance video against static background is presented.First,on the basis of the moving target detection as well as the moving and edge information,and making sure that the difference of the front and rear seams is minimum,the compression of one frame is achieved.Then taking the relationship of adjacent frames into account,the time and space consideration is introduced and the corresponding cost function is designed,ensuring the continuity of the adjacent frames after clipping.The experimental results show that this method not only shows good compression results on static background surveillance video,but also ensures the high quality of region of interests and the structure and integrity of video.