针对稀疏对象的监控视频的快速浏览,提出了一种基于目标分割和融合的快速视频浏览的方法.首先,利用背景差分法得到前景目标,并对背景模型不断更新;然后,滤除无目标出现的视频帧,并以一定的密度对目标对象进行融合播放.实验结果表明,该方法鲁棒性好,能取得较好的效果.%A quick video browsing algorithm is proposed for videos of sparse objects. Background subtraction is applied to get foreground object and the background is updated constantly. Those frames without objects are first discarded and the remaining frames are further fused. Experiment shows our proposed algorithm is robust and has good performance.
展开▼