基于特征点的曲面点云配准方法

Surface point cloud registration method based on feature points

针对当前曲面点云配准质量和效率低的问题,提出基于特征点的曲面点云配准方法。首先通过计算特征点曲率的豪斯多夫距离(HD)获得初始匹配点对;然后利点云局部表面曲线形状特征提取精确特征点对;最后用对偶四元数实现源点云和目标点云的粗配准,获取良好的初始的配准参数,并采用改进的最邻近点迭代算法实现点云的精配准。实验结果表明:该方法获取的点云中误差在0.005 m以内,且配准时间在7 s以内,提高了传统曲面点云配准的精度和配准效率,验证了该方法的稳健性和准确性。

Aiming at the problems of low quality and efficiency of current surface point cloud registration,a method for surface point cloud registration based on feature points was proposed.First,the initial matching point pair was obtained by calculating the Hausdorff distance(HD)of the curvature of the feature point;then the precise feature point pair was extracted by using the local surface curve shape feature of the point cloud;finally,the dual quaternion was utilized to realize the rough registration of the source point cloud and the target point cloud,and the acquisition was good.The initial registration parameters were used,and the improved nearest neighbor iterative algorithm was used to achieve precise registration of point clouds.The experimental results showed that the error in the point cloud obtained by this method was within 0.005 m,and the registration time was within 7 s,which improved the accuracy and registration efficiency of traditional surface point cloud registration,and verified the robustness and accuracy of the method.