基于视频流的运动人体检测方法研究  

RESEARCH ON MOVING PEOPLE DETECTION METHOD BASED ON VIDEO STREAM

作　　者：连晓峰[1] 张弢[1] 戴军[1] 刘载文[1] 王小艺[1] 

机构地区：[1]北京工商大学计算机与信息工程学院,北京100048

出　　处：《北京工商大学学报（自然科学版）》2009年第6期40-44,共5页Journal of Beijing Technology and Business University:Natural Science Edition

基　　金：北京市教委科技创新平台资助项目(2008176);北京工商大学教育教学研究与改革资助项目(jg095204);北京工商大学青年教师科研启动基金资助项目

摘　　要：提出一种基于对称帧间差分与背景减除相结合的运动目标检测和自适应背景更新方法.该方法首先建立背景模型,采用帧间差分法将当前帧图像分别与其相邻两帧图像相比较得到运动目标前景部分,并将三帧中变化率小于某一阈值的像素点以一定的更新率实时更新到背景模型中.同时,采用背景减除法对所建立的背景模型进行前景提取,将两种方法所获得的前景图像进行融合,再通过数学形态学运算去除噪声及小面积非人体运动部分,最终得到完整可靠的运动人体二值图像.实验结果表明该方法准确高效,能很好地克服光线的影响,提高监控系统的稳定性.In this paper,a new method of moving-people detection based on fusion of coterminous frames differencing and background subtraction is proposed.Firstly,the method constructs the auto-adapted background model and updating real-time through temporal difference to the video stream,and simultaneous adapting background subtraction extract foreground region from the established background model.Then the method integration the two foreground regions for people identification,together with utilization of the median filter and mathematical morphology operation eliminating noise and the small area of non-human movement parts.Finally obtaining the complete reliable moving target.The experimental results confirm that the approach is accuracy and effectively,and can satisfy the needs of real-time.So it can be applied to improve the stability of visual surveillance system.

关 键 词：运动人体检测 背景建模 帧间差分法 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]