摘要
提出了一种利用三维人体点云数据测量人体关键尺寸的方法。应用线结构光激光三维扫描仪对特定站姿的人体进行扫描,获得完整三维人体点云。根据人体视线和脚尖的同向性及两腿的自然分开特征,对点云进行方位调整,使人体朝向z轴正方向。利用手动删除和最大连通域删除相结合删除噪声点。确定人体的6个特征点和20个特征平面,分别应用对应截面y坐标差值法测量长度尺寸,凸壳法测量截面周长,测量了34个人体关键长度和周长,对影响测量精度的因素进行了分析。结果表明,典型尺寸测量误差小于3%,可满足人类工效学等领域快速测量人体特征尺寸的要求。
A measuring approach for figure size from 3D human point cloud is presented. Using a lightstripe 3D laser scanning system to scan the desired pose of body, human's three-dimensional coordinates are obtained without loss. Based on the directions of eyesight and the natural separating of two legs, the human's orientation is adjusted to facing z positive direction. Applying manual method and maximal connectivity domain method to delete noisy points, 6 figure points and 20 figure planes are labeled on human silhouette. Two different section planes' y coordinates are used to obtain one length size of human and a method, called convex hull method, to measure body section's girth, respectively. Using this technique, the 34 key figure lengths and girth sizes can be measured, and the factors affecting measurement precision can be analyzed. Measurement results show that the typical measurement error of this method is lower than 3% and can meet the requirement of industry.
出处
《光学精密工程》
EI
CAS
CSCD
北大核心
2007年第1期84-88,共5页
Optics and Precision Engineering
基金
国家自然科学基金资助项目(No.60277009)
天津市科技攻关计划培育项目(No.05YFGPX05000)
关键词
人体尺寸
测量
凸壳法
三堆点云
激光三维扫描
human figure size
measurement
convex hull method
3D point cloud
laser 3D scanning