基于功率谱密度的人体运动状态检测  

Detection of human movement state based on power spectral density

作　　者：蔺广逢[1] 张二虎[1] 顾桓[1] 潘向辉[1] 范引娣[2] 

机构地区：[1]西安理工大学印刷包装工程学院,西安710048 [2]陕西交通职业技术学院经济管理系,西安710014

出　　处：《计算机应用》2008年第5期1269-1271,共3页journal of Computer Applications

基　　金：西安理工大学科学研究计划项目(104-210703)

摘　　要：提出了一种视频图像序列中人体运动状态检测的新方法。针对传统人体状态检测方法利用数据信息的复杂性,对图像中人体外接矩形区域提取位置和尺度稳定的非结构化数据信息——关键点,计算关键点的功率谱密度,在序列视频图像中可获得关键点功率谱密度的分布,利用其分布特点判断人体运动状态。实验证明方法的有效、可行和简易。A new method of human movement state detection in image sequence of video was put forward. Aiming at the complexity of data used by tradition method, the new method firstly extracted key points-unstructured data of position and scale stabilization from circum-rectangle area of human body, computed power spectral density of key point; Then obtained distribution of power spectral density of key point; finally judged human movement state form its distribution characteristic. Experiment shows its efficiency, feasibility and simplicity.

关 键 词：人体运动检测 功率谱密度 非结构化数据信息 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]