一种权值约束的精确配准算法

A New Weight Constraint Precision Registration Algorithm

针对数据与模型的精确配准问题,提出一种权值约束的配准算法,通过对配准点施加不同的权值,利用权值约束保证模型重要区域的配准精度。首先,论文基于经典配准模型,引入权重因子,建立了改进的权值约束的配准模型。针对配准模型的求解问题,通过对现有SVD-ICP算法进行适应性改进,提出并研究了带权SVD-ICP(wSVD-ICP)算法,重点推导了基于wSVD算法求解旋转矩阵R和平移矩阵T的过程。最后,论文利用仿真数据和实测数据对配准模型进行了验证;计算结果表明,论文所提算法通过对精度要求较高区域分配高权值进行约束,可有效提升该局部区域的配准精度;同时,可在一定程度上改进整体配准精度和效率。

Aiming at the precision registration, a registration algorithm with weight constraint is studied in this paper. This algorithm can ensure the registration accuracy by imposing different weights to different registration points. First, the registration model with weight constraint is established by introducing weight factor into traditional registration model. Second, in order to solve the registration model, a weight constraint SVD-ICP algorithm(wSVD-ICP) is studied based on the classical SVD-ICP algorithm. Especially, the detailed solving process of rotation matrix R and transform matrix T based on wSVD algorithm is shown. At last, two applications are presented for demonstration by using simulation data and measuring data respectively. The results indicate that the algorithm employed in this paper can improve the local registration accuracy significantly by assigning high weight to the high precision demand registration points. Moreover, the algorithm can improve the global registration accuracy and efficiency.