约束总体最小二乘在点云拼接中的应用

APPLICATION OF CONSTRAINED TOTAL LEAST-SQUARES TO CLOUD POINT REGISTRATION

在两站点云拼接的大角度空间直角坐标转换问题中,由于拟合靶标球心坐标值在两套扫描测站坐标系下均存在误差,提出基于约束总体最小二乘的点云拼接方法,建立附有约束条件的变量中的误差模型,对观测向量和系数阵同时进行修改。经算例证明,与传统的约束最小二乘法相比,可得到更加合理的模型和更高精度的参数解。

The cloud point registration is a peoblem of three dimensional-datum transformation with big rotation angle.As the fitting coordinates of sphere target’s centers in the two sets of scanning coordinate systems are both influnenced by errors,we introduce constrained total least-squares（CTLS） method for cloud point registration.We established the errors-in-variables（EIV） model with the constraints and revised the observed vector and the coefficient matrix.The calculated example has proved that the more reasonable transformation model and the more accurate parameter solution can be given with CTLS methods as compared with the CLS methods.