一种单目视频下的人体量测参数计算方法  

Anthropometric features extracted from calibrated image sequences by single camera

作　　者：刘少华[1] 杜奎[1] 

机构地区：[1]海军工程大学装备经济管理系,武汉430033

出　　处：《海军工程大学学报》2013年第3期109-112,共4页Journal of Naval University of Engineering

基　　金：海军工程大学自然科学基金资助项目(HGDQNJJ11042)

摘　　要：提出了一种在单目标定图像序列中提取身体量测参数的方法。该方法将人体掩膜图像简化为人体线模型,从而得到人体特征点,包括头顶点、肩点、重心点和脚点;通过反投影计算出特征点在现实世界中的空间坐标,从而得到身体量测参数,包括身高、肩高、肩宽和步幅。实验结果表明:该算法提取的身体量测参数误差较小,有较强的可用性。This paper proposes a method for extracting anthropometric parameters from the calibrated image sequences by a single camera. The method can reduce the masking image of human body to a human line model so as to deal with the key points of human body, such as the top of head, shoulder, barycentric point and foot. The coordinates based on the key points obtained from antiprojection are used to display the anthropometric features such as stature, shoulder breadth and height, and pace. The experiment results show that the method is available with little error in anthropometric dimen- sion.

关 键 词：人体量测特征 人体线模型 单目视频 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]