一种利用二维时空模型的人体步态运动估计  

Human Body Gait Movement Estimation Using 2D Temporal and Spatial Models

作　　者：董小铨[1] 魏玮[1,2] 申媛媛[1] 

机构地区：[1]河北工业大学计算机科学与软件学院,天津300401 [2]河北省数学研究中心,石家庄050024

出　　处：《计算机工程》2012年第13期172-174,共3页Computer Engineering

基　　金：河北省自然科学数学研究专项基金资助项目(08M006)

摘　　要：以人体步态运动为研究对象,运用二维模型来估计和重建人体步态运动的信息。利用点分布模型训练轮廓点和骨架点的二维图像样本,通过构建多个视角的二维模型解决不易确定方向的问题,在得到总体的二维时空模型框架后,利用基于跨角最大化的时空约束寻找可能所属的类别,进而得到相应的步态信息。实验结果证明,利用该时空约束可以快速地对人体步态运动进行估计。This paper researches on the gait of human body, it makes use of the 2D model to estimate and reconstruct the gait information of human body. Mainly using the point distribution models to train the 2D image samples including the points of shapes and skeletons, the problem of the uncertain orientation of the 2D models can be solved by constructing them corresponding to many views. After getting the global 2D temporal and spatial model framework, temporal and spatial constraints based on the maximum angle between legs which are first proposed are considered to find the most probable models, thus information of the gait is gotten. Experimental results show the proposed algorithm can estimate the human body gait quickly and accurately.

关 键 词：人体步态 二维模型 点分布模型 主成分分析 时空约束 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]