一种摄像机和3维激光雷达外部参数最大似然估计的标定算法  

A Calibration Algorithm for Maximum Likelihood Estimation of Extrinsic Parameters of a Camera and a 3D Laser Radar

作　　者：温景阳[1] 杨建[2] 付梦印[1] 张继伟[1] 杨蔚青[2] 

机构地区：[1]北京理工大学,北京100081 [2]中国兵器科学研究院,北京100089

出　　处：《机器人》2011年第1期102-106,共5页Robot

基　　金：国家自然科学基金资助项目(60773044);教育部长江学者和创新团队发展计划资助项目;中国博士后科学基金资助项目(20080440310)

摘　　要：提出了摄像机和激光雷达外部参数的最大似然估计算法.本算法在计算摄像机和3维激光雷达坐标系转换的旋转矩阵时,考虑了各参数初值的误差情况,根据方差的大小为参数初值赋予相应的权值,使误差大的初值具有较小的权重,而误差小的初值具有较大的权重,从而使参数初值的误差对标定结果的影响能够一致.在此基础上求解旋转矩阵的最大似然估计值,得到了更加精确的结果.此外建立了误差模型,揭示了标定过程中以差的传递机理,为深入研究标定问题提供参考.The maximum likelihood estimation method of extrinsic parameters of video camera and laser radar is proposed. When calculating the rotation matrix for the coordinate transformation of video camera and three-dimensional laser radar,the error of each initial value is taken into consideration,and corresponding weights are assigned to these initial values according to their respective variance,i.e.assigning minor weights to the initial values with major error and major weights to the initial values with minor error,thus to make the effects of various initial values＇ error on the calibration results consistent with each other.On this basis,the maximum likelihood estimation of rotation matrix will be more accurate.In addition,by establishing the error model,this paper reveals the error transfer principle in the calibration process,providing reference for the further research on the calibration problems.

关 键 词：摄像机 3维激光雷达 参数标定 信息融合 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]