一种基于手机平台的运动人体检测算法的研究  

A Human Movement Detection Algorithm on Mobile Phone Platform

作　　者：陈朝钢[1] 吴春明[1] 叶秉威[1] 

机构地区：[1]浙江大学人工智能研究所,浙江杭州310027

出　　处：《电子学报》2009年第4期771-775,共5页Acta Electronica Sinica

基　　金：浙江省科技计划项目(No.2007C21031)

摘　　要：针对智能手机平台计算能力弱、存储空间有限的特点,本文提出了一种适用于该平台的快速运动人体检测算法.在背景光线慢变的条件下采用背景更新法以快速提取前景目标;针对光线剧变的情况,设计了一个光线剧变检测法,以检测前景目标.在此基础上,结合背景差与帧间差方法,给出了适合在室内环境下基于手机平台监控报警的快速运动人体检测算法.实验结果表明,算法能有效探测到人体目标,并能应对光线突变之情况.A rapid human movement detection algorithm aiming to solve the problem of low computing power and limited memory of smart phone is proposed in this paper. When the background light changes slowly, the background upgrating method is used to extract the moving object, while another method is designed when light changes dramatically.After the moving object is extracted, a human detection algorithm is designed to identify whether the object is a human body based on the difference of back-ground and flames, which can be used in supervision system. The experimental results show that this algorithm can effectively detect human object regardless of light changes.

关 键 词：运动目标检测 人体探测 光线剧变 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]