基于视频的人数识别方法综述  

A Survey on the Video—Based Group Number Recognition

作　　者：赵明瀚[1] 王晨升[1] 

机构地区：[1]北京邮电大学自动化学院,北京市100876

出　　处：《软件》2013年第3期10-12,54,共4页Software

摘　　要：基于计算机视觉的人群行为分析是一个非常活跃的研究领域,视频人数统计是人群行为分析的一个重要内容。本文对视频人数识别近年来的发展作了比较详细的论述,从基于特征点跟踪、基于区域的跟踪和基于模板匹配的跟踪三个方面分析近些年人数识别进展情况。文章通过对各种不同识别算法思想的研究,对当前该研究方向上亟待解决的问题做了比较详细的分析,并对未来人数识别的指出了一些改进方法。Group behavior analysis based on computer vision,has been a quite active research field,in which video population statistics is regarded as an important issue.This paper aims to discuss the recent development of video number recognition in detail and analyze the recent progress situation of number recognition from three perspectives—feature-based,area-based and image-matching trackings.Through comparison among studies of diffirent types of recognition algorithm ideas,problems to be solved at present in this research field are analyzed at length and an improved method of future number recognition is provide as well.

关 键 词：计算机视觉 跟踪算法 综述 人数统计 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]