摘要
提出了一种当视频中的运动物体有较大范围的运动时,对其进行分割的方法.首先利用帧差法来快速定位运动物体的位置,并以此作为初始分割结果.然后利用均值偏移法准确估计运动物体的边缘并利用图切割方法建立两者之间联系.考虑到视频的运动连续性,同时引入前一帧分割结果来约束当前帧的分割.该算法同时利用了帧差法和均值偏移法的优点,能够快速准确地分割在视频场景中出现的运动物体.
A moving object segmentation algorithm is proposed. It can deal with the object which moves quickly and widely in videos. Firstly, the frame difference to estimate the initial positions of the moving objects was computed. Secondly, the mean-shift to segment the video frames and combine the segment results by refining the initial esti- mated moving object with the computation of a graph cut was used. During the calculation, the segmentation result of the previous frame is introduced as a restriction of the current frame segmentation, and thus it assures the contin- uous and smooth segment results. The approach incorporates the advantages of the frame difference and mean-shift, which can quickly locate the moving object and precisely segment the moving objects from video frames.
出处
《浙江大学学报(理学版)》
CAS
CSCD
北大核心
2007年第6期636-639,共4页
Journal of Zhejiang University(Science Edition)
基金
国家自然科学基金资助项目(No.60502006)
关键词
帧差法
运动物体分割
图切割
分水岭
frame subtraction
moving object segmentation
graph cut
watershed