适用于光学运动捕捉数据校正的拟合算法  

Fitting Algorithm for Optic-based Motion-capture-data Calibration

作　　者：吴琼[1] 黄海明[2] 刘金刚[1] 

机构地区：[1]中科院计算所&首都师范大学计算机科学联合研究院,北京100037 [2]中国科学院自动化研究所,北京100080

出　　处：《计算机工程》2007年第19期221-223,共3页Computer Engineering

基　　金：国家"863"计划基金资助项目(2001AA114201)

摘　　要：为了从软件上提高基于CCD的三维运动捕捉系统的测量精度,减小因CCD成像及装配因素造成的空间扭曲误差,该文在分析了散乱数据拟合原理的基础上,比较了各种拟合方法的优劣,提出了一种基于搜索的分层多重拟合算法,给出了该算法的实现过程及实验结果。结果表明,该算法拟合较精确,时间复杂度低,且具有很高的收敛性,是一种实用的数据拟合算法,达到了运动捕捉数据校正的目的。　In order to improve the precision of the 3D motion capture system based on CCD devices using software and decrease the space-warp error arisen from CCD imaging and assembling,this paper compares several data-fitting methods and proposes a layered and repetitious data-fitting algorithm based on searching by analyzing scattered data fitting theory.The algorithm is explained and the experiment results are presented.Experiments show that its accuracy,time complexity and convergence are better than other approaches.It is a practical data-fitting algorithm and is appropriate for calibrating the motion capture data.

关 键 词：运动捕捉 散乱数据 拟合 线阵CCD 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]