基于SIFT特征点提取的ICP配准算法

ICP Registration Algorithm Based on SIFT Feature Point Extraction

为解决传统迭代最近点(ICP)算法对点云配准的起始点对选择不佳而导致配准时间长、效率低的问题,提出一种基于尺度不变特征变换(SIFT)特征点提取的ICP点云配准算法(ST-ICP)。首先使用SIFT算法进行原始点云与目标点云的SIFT特征点提取,根据提取特征点完成快速点特征直方图(FPFH)特征运算,通过采样一致性初始配准算法(SAC-IA)搜索对应点对、求解变换矩阵,再进一步运用ICP算法进行点云精细配准。实验结果表明:与ICP算法相比较,ST-ICP算法的配准误差在迭代次数为5次时减小了1.019 cm,迭代次数为10次时减小了0.443 cm;在配准误差达到10^(-2) cm级别时,ST-ICP算法所用时间比传统ICP算法减少了12.829 s。ST-ICP算法优化了对应点对的选择,提升了配准精度和配准效率。

The ST-ICP algorithm is proposed to address the issue of poor selection of starting point pairs in traditional iterative closest point(ICP)registration algorithm,which results in long registra-tion time and low efficiency.Firstly,the SIFT algorithm is used to extract SIFT feature points from the original point cloud and the target point cloud,and the fast point feature histogram(FPFH)fea-ture operation is completed according to the extracted feature points.This feature is used to search for the corresponding point pairs and solve the transformation matrix by sample consensus initial alignment(SAC-IA)algorithm,and the ICP algorithm is further used to perform point cloud fine local registration.The experimental results show that compared with the ICP algorithm,the registra-tion error of the ST-ICP algorithm is reduced by 1.019 cm when the number of iterations is 5 times,and 0.443 cm when the number of iterations is 10 times.When the registration error rea-ches the level of 10^(-2)cm,the time taken by the ST-ICP algorithm is reduced by 12.829 s compared with the traditional ICP algorithm.The ST-ICP algorithm optimizes the selection of corresponding point pairs,which has better registration accuracy and improves registration efficiency.