嵌入式系统中视频运动对象分割

Moving object segmentation of video sequence in embedded systems

提出了一种基于嵌入式系统的视频运动对象分割算法。首先利用差图像法抽取出运动的像素点,然后通过统计像素点的状态变化频率来区分运动物体和动态背景,并配合一权值状态矩阵将全局光照突变和动态背景像素自适应融合到背景中,从而分割出运动对象并进行跟踪。实验结果表明,该算法在嵌入式系统中实时跟踪运动目标取得了很好的效果。

A video sequence segmentation algorithm based on embedded systems was presented. Firstly, moving pixels was extracted by using difference image algorithm, then motion object and moving background were distinguished by the change frequency of pixel state. Secondly, using a dynamic matrix, the pixels of global lighting change and moving background was updated to background adaptively. So the moving objects could be extracted and tracked from video sequence. Experiment results demonstrate that this method is effective to track video moving objects in embedded systems.