人体行为的识别及发展  

Human Body Motion Recognition and Development

作　　者：张行健[1] 

机构地区：[1]上海电力学院电子与信息工程学院,上海200090

出　　处：《上海电力学院学报》2017年第1期102-106,共5页Journal of Shanghai University of Electric Power

摘　　要：人的动作行为识别是计算机视觉领域中的热点研究课题之一,广泛应用于安防监控系统、医疗诊断监护、人机交互等方面.从运动对象检测、行为特征提取和人体行为识别方法方面阐述了人体行为识别技术的现状和相关难点.与已有相关综述性文章相比,结合了新的研究成果,介绍了基于轨迹能量扩散图的组行为识别方法以及基于稠密光流轨迹和稀疏编码算法的行为识别算法.且指出了未来行为识别的发展趋势.Human motion recognition is one of the hot topics in the field of computer vision, which is used widely in security monitoring system, medical diagnosis and monitoring, human-com- puter interaction, etc. The current state and related difficulties of human motion recognition technol- ogy is presented within the motion object detection, feature extraction and human motion recogni- tion method. Compared with the existing relevant review articles, the new research results are com- bined, such as group behavior recognition method based on trajectory energy diffusion map and be- havior recognition algorithm based on dense flow trajectory and sparse coding algorithm. Further- more, some research direction of motion identification on future work is given.

关 键 词：动作行为识别 运动对象检测 特征提取 识别方法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]