基于分级的MRF视频运动检测算法及其实现  

Research and Implementation on Multiresolution Algorithm Based on MRF for Video Motion Detection

作　　者：乔予杰[1] 胡俊[1] 苏祥芳[1] 孙洪[1] 茹国宝[1] 

机构地区：[1]武汉大学电子信息学院,湖北武汉430072

出　　处：《武汉大学学报（理学版）》2002年第1期99-102,共4页Journal of Wuhan University:Natural Science Edition

基　　金：国家自然科学基金资助 ( 6 0 0 72 0 41)

摘　　要：在阐述马尔可夫随机场 (MRF)视频运动检测算法理论的基础上 ,采用分级的方法 ,形成图像序列的时空多分辨率结构 ,大大优化了单一分辨率 MRF算法的初始化过程 ,并通过软件进行了实现 ,有效改进了单一分辨率算法检测的结果 .该算法应用于数字视频监控系统 ,可明显提高检测精度 。Intrduce video motion detection algorithm based on Markov Random Field(MRF). The MRF based motion detection algorithm has more robustness than the others, but it has its defaults yet. Multiresolution is used in the MRF based algorithm .The results show that it can improve the initialization of the algorithm and decrease the mistakes of video motion detection.So,the multiresolution algorithm based on MRF is bat around.Its realization in software is set forth.The multiresolution algorithm based on MRF has wide foreground of application in many fields.

关 键 词：图像序列分析 分级 马尔可夫随机场 图像处理 视频运动检测算法 数字视频监控系统 

分 类 号：TN911.73[电子电信—通信与信息系统]