Detecting Motion Regions using Statistic Parameters

摘要：

Background subtraction has become a popular method for video based motion detection. In this paper, we present a novel statistic parametric model by doing statistical analysis for history samples, incorporating the parameters of the sample number forming the models, the sampling time centroid and the last time point, which are ignored by existed background models. With these parameters, the model can be updated in time and accurately. The experimental results show that the presented model can suppress false detections from tail phenomenon, shadows, illumination change, repetitive motion, cluttered areas, and so on.

关键词： motion detection statistic parametric model tail phenomenon

作者: Yun GAO Xuejie ZHANG Hao ZHOU Jidong LI

作者单位: School of Information of Science and Engineering,Yunnan University Kunming, China High Performance C School of Information of Science and Engineering,Yunnan University Kunming, China

会议类型: 国际会议

会议名称: 2010 International Conference on Software and Computing Technology(2010年软件与计算机技术国际会议 ICSCT 2010)

会议地点: 昆明

会议语种:英文

页码: 206-209

在线出版日期: 2010-10-17（万方平台首次上网日期，不代表论文的发表时间）