人体运动目标的检测与跟踪  

Moving Object Detection and Tracking for Human-body

作　　者：张铁力[1] 王然冉[1] 李晓佳[1] 

机构地区：[1]沈阳建筑大学信息与控制工程学院,辽宁沈阳110168

出　　处：《自动化博览》2007年第5期102-105,共4页Automation Panorama1

摘　　要：提出了一种静止背景的情况下人体运动目标的检测与跟踪的新方法。该方法利用背景差分法与粒子滤波器算法相结合。首先,利用背景差分法可检测运算出人体运动区域的大小和形心,在一定的时间间隔t(t<<1s)后,再次利用背景差分法经运算可得到人体运动的速度,然后运用粒子滤波器算法利用背景差分所获得的人体运动区域的大小、形心、速度三个参数建立跟踪模型。实验结果证明,该方法对人体目标跟踪是快速且有效的,并且有很好的鲁棒性。A new algorithm is presented for moving object detection and tracking of human-body in a static background in this paper, This algorithm combines background difference with the particle filter, First, the background difference can be used to detect human-body moving region and centroid, then using the background difference method again to get the speed of moving human-body object after a certain interval t （t 〈〈1s）, finally applying particle filter to set up tracking model according to the parameters, centroid and speed which are obtained from background difference, Experimental results show that the presented method is fast, robust, and effective.

关 键 词：粒子滤波器 背景差分 检测与跟踪 人体运动目标 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]