基于Kinect的三维人体建模研究  

Research on 3D human modeling based on Kinect

作　　者：高丽伟 张元[1] 韩燮[1] GAO Liwei;ZHANG Yuan;HAN Xie(Data Science and Technology,North University of China,Taiyuan 030051,China)

机构地区：[1]中北大学大数据学院,太原030051

出　　处：《中国科技论文》2018年第2期196-201,共6页China Sciencepaper

基　　金：国家科技支撑计划重点项目(2013BAH45F02)

摘　　要：针对三维人体模型在多媒体工业中高精度的要求,提出了1种通过查找近似模型并结合网格变形算法的三维人体重建方法,采用Kinect深度摄像机扫描人体获得目标模型;在可变模型中查找与扫描模型特征信息相近的模型;并利用可变人体模型的网格变形,重建出三维人体模型。实验结果表明,所提方法简单、易操作,且不需任何人体标定,能够重建出高精度的三维人体模型。Aiming at the high-precision requirement of 3D human model in multimedia industry,a 3D reconstruction method is proposed,combining the approximation model and mesh deformation algorithm.Firstly,the Kinect depth camera is used to scan the body to generate the target model.Then,a model similar to the characteristic information of the scanning model is searched from the deformable human models.Finally,the triangle mesh deformable of human model is applied to reconstruct the 3D human model.The experimental results show that the method is easy to implement,and good at 3D human model reconstruct without any calibration.

关 键 词：三维人体模型 Kinect深度摄像机 查找体型相似模型 迭代最近点 网格变形 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]