基于特征点匹配的点云配准方法研究

针对快速点特征直方图(fast point feature histogram, FPFH)与迭代最近点(iterative closest point, ICP)算法结合的配准方法达不到精度要求的问题,文章在FPFH的基础上加入特征点的提取与匹配,使得配准精度进一步提升。该方法先通过尺度不变特征变换(scale-invariant feature transform, SIFT)算法和3DHarris算法对点云数据的特征点进行提取,再通过计算FPFH寻找对应点对,使用随机采样一致性(random sample consensus, RANSAC)算法剔除错误点对,通过奇异值分解(singular value decomposition, SVD)算法计算初始旋转矩阵和平移矩阵,最后用传统ICP精配准。结果表明,基于特征点匹配的算法相比基于特征描述的算法精度更高。

Aiming at the problem that the accuracy of the registration method combining the fast point feature histogram(FPFH) and the iterative closest point(ICP) algorithm is not satisfactory, the extraction and matching of feature points are added on the basis of FPFH, which further improves the registration effect. This method first extracts the feature points of the point cloud data through the scale-invariant feature transform(SIFT) algorithm and the 3DHarris algorithm, then finds the corresponding point pairs by calculating FPFH, uses the random sample consensus(RANSAC) algorithm to eliminate the wrong point pairs, and calculates the initial rotation matrix and translation matrix by the singular value decomposition(SVD) algorithm, and finally uses the traditional ICP to perform fine registration. The results show that the algorithm based on feature point matching has higher accuracy than the algorithm based on feature description.