摘要
针对动态目标检测中的图像背景实时提取问题,采用红绿蓝三分量综合考虑的方法,进行了比较深入的算法研究。研究表明,如果图像序列时间段过短,背景中就会有目标空洞,如果图像序列时间段过长,则不能反映实时背景,在后续目标检测中就会引起较大误差。考虑合理的视频场景区域的动态目标密集程度因素,就可以较好地兼顾提取背景的实时性和背景中目标空洞的消除。
Based on the frame-difference of color video stream during a time interval, we can extract the real-time background image of the video with moving objects in it. By using comprehensive means of red, green and blue com- ponents, this paper conducts a more in-depth algorithm to real-time extraction of image background problem of moving object detection. The research shows that, if the time interval is too short, the extracted background will have holes in it; if the time interval is too long, the extracted background will not be real time. This will cause errors to follow processing of moving objects detection. Both the real-time and the holes of the extracted background can has a good result if we select a reasonable time interval according to the size and the moving velocities of the objects in video.
出处
《成都信息工程学院学报》
2010年第2期167-171,共5页
Journal of Chengdu University of Information Technology
基金
成都信息工程学院引进人才科研启动项目(KYTZ20060603)
关键词
信息处理
智能监控
动态目标检测
背景图像提取
帧间差分
information processing
intelligent surveillance
detection of moving objects
background image extrac-tion
frame-difference