监控视频的前景运动物体提取方法  

Foreground extraction in surveillance scene

作　　者：肖碧波[1] 胡伟[1] 

机构地区：[1]北京化工大学信息科学与技术学院,北京100029

出　　处：《计算机工程与设计》2016年第3期695-699,共5页Computer Engineering and Design

基　　金：国家973重点基础研究发展计划基金项目(2011CB706900);北京市青年英才计划基金项目(YETP0501)

摘　　要：为能在包含复杂场景的监控视频中准确提取出前景运动物体,提出一种结合Vibe前景提取和背景减除的迭代算法。通过对Vibe前景提取算法和背景减除法进行研究与分析,指出当前前景运动物体提取方法在提取过程中的难点和不足;利用Vibe前景提取算法和背景减除法的优点,设计迭代算法;结合Vibe前景提取法的漏而不丢与背景减除法的多而无孔的特性,实现在监控视频中准确提取出前景运动物体的目的。实验结果表明,该算法可行有效,为监控视频的后续内容分析提供了准确的前景运动物体数据。To accurately extract foreground moving objects in surveillance videos,an effective iteration algorithm combining the Vibe foreground extraction method and a simple background subtraction method was proposed.Limitations of these two methods were analyzed,and the corresponding reasons were also studied.Based on the advantages of the Vibe and the simple subtraction methods,an iterative algorithm was presented to avoid problems caused by these two methods,and the presented method can accurately extract foreground objects efficiently.Results of experiments show the accuracy and efficiency of the proposed method which can provide accurate moving objects data for further analysis of surveillance videos.

关 键 词：监控视频 前景物体提取 Vibe 视频内容分析 背景减除法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]