摘要
人的行为理解与描述是近年来被广泛关注的研究热点和难点。本文设计了一个利用视频序列对人的日常行为进行识别的系统,本系统由图像处理模块和图像理解模块组成,图像处理模块主要实现图像的预处理和人体的正确分割,图像理解模块则完成人体动作的识别。利用本系统对人的日常生活中的"站"、"躺"、"蹲"、"坐"、"弯腰"等五种动作进行识别,试验结果表明本方法在实践中是可行的。在虚拟现实、视觉监控、感知接口等领域均有着广阔的应用前景。
Automatic understanding of people from images has increasingly become a central focus of research in computer vision over the past two decades, In this paper we develop a system for human behavior recognition in video sequences, The system is composed of image processing module and action understanding module, Image processing module is designed to realize image pre-processing and body segmentation, Action understanding module is used for human action recognition, Experimental results show that this method is practicable for the human activities analysis.
出处
《微计算机信息》
北大核心
2008年第13期264-266,共3页
Control & Automation
基金
国家自然科学基金资助(60104009)
关键词
计算机视觉
图像处理
行为识别
computer vision
image processing
action recognition
