曲率约束的激光点云全局优化配准算法

Global Optimization Registration Algorithm of Laser Point Cloud Based on Curvature Features

针对对于多视角下测得的散乱点云数据,ICP算法存在不稳定性和易收敛到局部最优的问题,提出基于曲率特征的ICP改进算法。该方法首先引入随机抽样一致性算法查找特征点,以距离最近为判断依据获得特征点对,然后利用四元数法计算配准参数,最后基于模拟退火法得到全局最优配准参数完成点云精确配准。实验表明,与传统ICP算法相比,改进的ICP算法可以有效提高点云配准的稳定性和精度。

In order to solve the problem of instability and easy convergence to local optimum in the ICP algorithm with scattered point cloud data measured from multiple angles,an improved ICP algorithm based on curvature characteristics is proposed in this paper.The method first introduces random sampling consensus algorithm to find feature points,and obtains feature points from the nearest judgment basis,then uses four element method to calculate the registration parameters.Finally,the accurate registration of the global optimal registration parameters is obtained based on the simulated annealing algorithm.Experimental results show that the improved ICP algorithm can effectively improve the stability and accuracy of point cloud registration compared with the traditional ICP algorithm.