March 3, 2026Open Access

基于八叉树与3d-Sift特征提取的点云配准算法

Key Points

The proposed algorithm significantly improves registration accuracy and efficiency for complex point clouds.
Specifically, the average error is reduced by up to 84.8% compared to existing methods like RANSAC+ICP and SAC-IA+ICP.
Preprocessing involves statistical filtering and feature extraction with octree sampling and 3D-SIFT to retain original data features.
Significantly reduces processing time, with the average time dropping to 7.97 seconds, facilitating effective handling of large datasets.

Abstract

增材制造拓扑构型件的三维检测在后处理中至关重要，拓扑构型件的表面几何形状不规则，结构上分布有大量孔洞，现有三维检测方法在配准拓扑构型件点云时，会出现配准效率低下和产生误匹配等问题，对此，提出了一种基于八叉树与3D-SIFT特征提取的点云配准算法。首先，对点云数据进行统计滤波预处理，然后采用八叉树采样和3D-SIFT相结合提取特征点，最大程度地保留点云原始特征；引入相对相异度向量改进SAC-IA粗配准算法，预先排除差异大的点对；最后，使用结合双向KD-tree与Welsch函数改进的ICP算法进行精配准。在斯坦福大学公共点云配准实验中，相比于RANSAC+ICP、SAC-IA+ICP、ISS+3DSC+NDT等算法，本文所提算法的平均误差分别减少了约84.8%、69.5%和54.7%，配准耗时分别减少约86.7%、78.1%和58.2%，在扫描拓扑构型件点云配准实验中，均方根误差为0.0612 mm，配准耗时为7.97 s。实验结果表明：本文算法能有效提高点云配准的准确性和效率，适用于数据量大、结构复杂的拓扑构型件点云配准，为提高后处理效果奠定了基础。

Connected Papers

Building similarity graph...

Analyzing shared references across papers

Discussion

Authors

Shuqi HUANG

上

上海工程技术大学机械与汽车工程学院，上海 201620

Zhifeng ZHOU

Journals

Journal of Applied Optics

Actions

References and Citations

Connected Papers

Building similarity graph...

Analyzing shared references across papers

基于八叉树与3d-Sift特征提取的点云配准算法

Key Points

Abstract

Citation Network

Connected Papers

Discussion

Authors

Journals

Actions

References and Citations

Citation Network

Connected Papers

Discussion

Cite this study