Surface Segmentation of 3D Point Clouds of Electrical Components Based on Gaussian Mapping

In the electrical industry, precise segmentation of 3D point cloud data is crucial for accurate geometric measurement, automated modeling, and defect detection in electrical components. However, existing segmentation methods often struggle to capture sharp features and complex details, limiting reconstruction accuracy. This paper proposes a sharp feature point detection and secondary surface segmentation method based on Gaussian mapping. The method includes three steps: identifying and correcting sharp feature point normal vectors using Gaussian mapping, performing initial segmentation, and applying a secondary fine segmentation in Euclidean space to separate surfaces with similar normal vector orientations. Experimental results on point clouds of 14 electrical components demonstrate that, compared to RG, RANSAC, RSPD, and C2NO, the proposed approach accurately detects sharp feature points and improves surface extraction accuracy.