基于图像处理的人群行为识别方法综述

摘要

Recognition of human action using image-based processing is an active research area in computer vision .At present most of the researchers mainly study the recognition of individual behavior .For recognition of group activities re-search is relatively less while the recognition of group behavior in intelligent monitoring ,virtual reality ,public security ,etc has a broad application prospect ,such as ACTS of terrorism ,suspicious event ,mass violence in real-time monitoring .Hu-man behavior recognition consists of detection ,tracking and object classification of image sequence ,and image understanding and recognition of human behavior .By introducing the recognition of individual behavior ,this paper analyzes the develop-ment and current status of group behavior in recent years and the research problems to be solved .And possible directions of future are discussed .%基于图像处理的人体行为识别是计算机视觉中一个活跃的研究领域，目前大部分研究者主要针对个人行为识别展开研究。对人群的行为识别研究相对较少而人群行为识别在智能监控、虚拟现实、公共安全等方面有着广泛的应用前景。例如，恐怖活动、可疑事件、群体性暴力事件的实时监控等。人体行为识别对包含人的图像序列进行检测、跟踪、目标分类、异常检测以及对人的行为理解与识别。论文通过从一般的单人行为识别出发，介绍了人群识别方面近年来研究发展状况以及当前该研究方向亟待解决的问题，并展望了未来可能的发展方向。