Abstract: This paper presents an extracting method of moving object based on vision system of mobile robot. The global motion estimation, extraction of object background, three-frame-difference technology and region growth method are used to extract moving object from video. The motion compensation is computed by neighbor searching and region growth with constraints is adopted to extract the moving objects for low time cost. The results show that it has lower time cost and good results.

Key words: moving object, vision system, global motion estimation, region growth

摘要： 提出了基于移动机器人视觉系统的运动目标提取方法。以视觉系统获取的视频作为研究对象，通过全局运动估计、目标背景的提取、三帧差技术和区域生长方法提取运动目标。采用邻域搜索运动补偿和带约束的区域生长的方法，降低了时间开销。实验结果表明，在时间开销和目标的提取上取得了较好的效果。

关键词: 运动目标, 视觉系统, 全局运动估计, 区域生长

CLC Number: 

• TP391