摘要
对固定镜头下视频序列中人群的计数方法进行研究。根据人头的轮廓特性和颜色特性,提出一种人头检测的方法,较好地适用于稀疏人群和对遮挡不是十分严重的人群。把检测出的头部的中心点作为运动人体的特征点,利用基于目标运动连续性的匹配算法对人头部进行匹配跟踪,从而实现人数的计数。实验表明,该方法具有很好的实时性和较高的计数准确性。普通PC机即可实现。
This paper discusses the pedestrian counting method in video sequences gathered from a fixed camera.According to the profile characteristic and color feather of a person＇s head,this paper proposes a novel algorithm which can be better used for sparse crowd to detect head.Then the central point of the head area is regarded as the feature point of a pedestrian.With the points,a match algorithm based on object motion continuity is used for head matching and tracking to implements a whole system for crowd counting.Large numbers of experimental results on several real video sequences demonstrates real-time effect and the high precision for various environments.
出处
《机械制造与自动化》
2010年第4期134-138,共5页
Machine Building & Automation
关键词
人数统计
人头检测
椭圆拟合
最小二乘法
人体跟踪计数
pedestrian counting
head detection
ellipse fitting
least square method
human tracking counting
