作者：孟繁昌, 梁晋, 张桁维, 徐劲澜, 赵鹏亮
作者单位：西安交通大学机械工程学院 机械制造系统工程国家重点实验室, 陕西 西安 710049
关键词：三维人体重建;Kinect传感器;多相机标定;ICP算法
摘要：
为解决目前人体扫描系统价格昂贵、操作复杂、占用面积大等问题，基于微软推出的Kinect相机具有快速、安全、经济的特点，提出利用3个Kinect相机和一个旋转台来构建三维人体扫描系统，研究该系统涉及的深度相机成像原理、多相机标定以及点云配准等技术。首先利用KinectFusion技术重建出单个传感器的三维点云，然后采用多相机标定技术，计算各传感器之间的相对位置关系，实现点云的初始拼接，并使用ICP算法实现点云的精确配准，从而重建出完整的三维人体点云模型。实验结果表明，该方法能够重建出高精度点云，并且该人体扫描设备成本低，占地面积小，可行性与实时性强。
Study on the rotary human body scanning system based on multiple Kinects
MENG Fanchang, LIANG Jin, ZHANG Hengwei, XU Jinlan, ZHAO Pengliang
State Key Laboratory for Manufacturing Systems Engineering, School of Mechanical Engineering, Xi'an Jiaotong University, Xi'an 710049, China
Abstract: In order to solve the problems such as high price, complex operation and large occupation area of current human body scanning system, with the Kinect camera introduced by microsoft with the rapid, safe and economical characteristics, a 3D human body scanning system using three Kinect cameras and a rotating platform is proposed. The principle of imaging of the depth camera, multi camera calibration technique and the method of the point cloud registration involved in the system are studied. Firstly, KinectFusion technology is used to reconstruct the 3D point cloud of single sensor, and then the relative position parameters between the sensors are calculated with the method of multi camera calibration to realize the initial registration of point clouds. Finally, ICP algorithm is used to achieve the optimal registration of point clouds and accordingly a complete 3D human point cloud model is reconstructed. The experimental results show that the method mentioned above can reconstruct the high-precision point clouds and the device is of practicability and feasibility with the advantages of real-time, low cost and less land occupation.
Keywords: 3D human body reconstruction;Kinect sensor;multi camera calibration;ICP algorithm
2018, 44(2): 72-77 收稿日期: 2017-04-29;收到修改稿日期: 2017-06-15
基金项目: 国家自然科学基金资助项目（51675404，51421004）
作者简介: 孟繁昌(1991-),男,山西吕梁市人,硕士研究生,专业方向为三维光学测量。
参考文献
[1] 黄敏,杨念,张剑. 基于雅可比算法的着装人体三维模型重建[J]. 东华大学学报(自然科学版),2007,33(5):607-612.
[2] 刘鑫,许华荣,胡占义. 基于GPU和Kinect的快速物体重建[J]. Acta Automatica Sinica,2012,38(8):1288-1297.
[3] JONES P R, WEST G M, HARRIS D H, et al. The Loughborough anthropometric shadow scanner(LASS)[J]. Endeavour,1989,13(4):162-168.
[4] 李勇,付小莉,尚会超. 三维人体测量方法的研究[J]. 纺织学报,2001,22(4):261-263.
[5] 甘应进,陈东生,孟爽,等. 非接触式三维人体计测现状[J]. 纺织学报,2005,26(3):145-146.
[6] ISTOOK C L, HWANG S J. 3D body scanning systems with application to the apparel industry[J]. Journal of Fashion Marketing and Management,2001,5(2):120-132.
[7] 吴磊. 基于Kinect的室内场景三维重建[D]. 北京:北京理工大学,2015.
[8] 周瑾,潘建江,童晶,等. 使用Kinect快速重建三维人体[J]. 计算机辅助设计与图形学学报,2013(6):873-879.
[9] 崔微,朱民耀,颜柯,等. 基于Kinect的三维重建技术在三维显示中的应用[J]. 电子测量技术,2017,40(3):113-116
[10] MAO Q R, PAN X Y, ZHAN Y Z, et al. 基于Kinect的实时面部情感识别(英文)[J]. Frontiers of Information Technology & Electronic Engineering,2015(4):272-283.
[11] WEISS A, HIRSHBERG D, BLACK M J. Home 3D body scans from noisy image and range data[C]//Interna-tional Conference on Computer Vision. IEEE Computer Society,2011.
[12] NEWCOMBE R A, IZADI S, HILLIGES O, et al. KinectFusion:Realtime dense surface mapping and tracking[C]//IEEE International Symposium on Mixed and Augmented Reality. IEEE Computer Society,2011.
[13] 宋诗超,禹素萍,许武军. 基于Kinect的三维人体扫描、重建及测量技术的研究[J]. 天津工业大学学报,2012(5):34-37.
[14] MUTTO C D, ZANUTTIGH P, CORTELAZZO G M. Time-of-flight cameras and microsoft KinectTM[M]. Springer US,2012:49-68.
[15] ZHANG D H. Exploitation of photogrammetry measure-mentsystem[J]. Optical Engineering,2010,49(3):263-267.
