摘要
对公共场所的人流统计系统研究由来已久,系统主要包括运动目标检测和目标跟踪两部分。本文对公交车这一特定场所,提出了一种基于多次区域分割和头部特征提取的算法来检测人头作为计数目标,采用质心跟踪和归一化形状匹配,并且通过采集公交车上的实时图像进行实验,准确率超过84%,结果证明该算法简单易行,能够有效的统计上下车的人数。
It has been a long time for the research of the flow statistic system in the public place.The system mainly include two parts—the detection of the moving objects and the tracking of the objects.For the special place of the bus,this paper proposes an algorithm based on several regional growings and the extraction of the head features to detect the heads as the counting objects.The centroid tracking method is used in this paper.Many images taken by the bus camera were used for the experiment.The result proves that the algorithm is easy-realising and is effective to count the passengers getting on or off the bus.
出处
《微计算机信息》
2010年第35期231-232,225,共3页
Control & Automation
关键词
客流统计
图像分割
目标跟踪
计数
Statistics of the passengers
Image segmentation
Objects tracking
Counting
