利用相邻帧和背景信息的运动对象检测  

A Novel Motion Object Detection Algorithm Using Adjacent Frame and Background Frame

作　　者：李波[1] 姚春莲[2] 李炜[1] 郑锦[1] 

机构地区：[1]北京航空航天大学计算机学院数字媒体室,北京100083 [2]北京工商大学,北京100037

出　　处：《电子学报》2008年第11期2154-2159,共6页Acta Electronica Sinica

基　　金：国家自然科学基金(No.60505007);教育部博士点基金(No.2003006034);北京教委项目(No.KM200810011004);新世纪优秀人才支持计划资助

摘　　要：运动对象检测是智能视频监控的基础,现有的运动对象检测算法在实时性和检测精度方面仍有待提高,本文在分析相邻帧间的亮度变化基础上,提出了基于三阶中心矩统计分布的背景点阈值计算方法,给出了抗噪声干扰的背景维护模型,最后综合利用时间差分法和背景减法进行运动对象检测,为了提高检测精度使用了形态学的腐蚀与扩张处理.对室内、室外序列以及标准序列的测试结果表明,本文算法速度快、抗噪声能力强,能够检测出被遮挡的运动对象.Motion object detection is the basis of intelligent video surveillance;the speed and the detection accuracy of the existing algorithms need to be improved. By analyzing the luminance change between adjacent flames, this paper presents threshold calculation method based on the third central moments of frame difference, and proposes an anti-noise background maintenance model and finally we combine the ideas of background subtraction with frame difference algorithm to detect motion object. Furthermore, morphological post-processing method is introduced to reduce noise and improve detection precision. Experiments on standard sequences and outdoor/indoor sequences, the algorithm has the advantage of high-speed, anti-noise and it can detect the obstacle motion object.

关 键 词：运动对象 背景维护 视频监控 阈值 

分 类 号：TN919.8[电子电信—通信与信息系统]