Abstract: In order to overcome the shortage of global motion compensation algorithm in video object segmentation, a new algorithm, which is based on local motion similarity is proposed. The motion vectors are processed by accumulation and median filter; the first segmentation of moving video object is executed according to local motion similarity and the spacial relativity is used for second segmentation; the edge of video moving object is made more exactly. Experimental results of different video sequence show the proposed algorithm has a better veracity of segmentation compared with other algorithms.

Key words: segmentation, video moving object, motion vector

摘要： 由于全局运动参数在一些情况下不能很准确地被估计，因此基于全局运动补偿算法存在一定的局限性。该文提出了一种基于运动相似性的视频运动对象的分割算法。该算法对运动矢量场进行累加和滤波处理，根据运动相似性对运动对象进行初次分割，并利用空间相关性进行二次分割，细化运动对象边缘。实验结果表明该算法避免了全局运动补偿算法的局限性，并使分割的精确性有明显提高。

关键词: 分割, 运动对象, 运动矢量

CLC Number: 

• TP312