基于几何增强与混淆约束的输电线路点云分割方法及装置
By explicitly enhancing multi-scale geometric features and highly normalized features, and combining the combined loss function and confusion constraint loss, the problems of difficulty in identifying small linear objects, class imbalance, and easy confusion in point cloud segmentation of transmission lines are solved, achieving high-precision and robust segmentation results that are suitable for power line inspection.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- HARBIN INST OF TECH AT WEIHAI
- Filing Date
- 2026-05-22
- Publication Date
- 2026-07-17
AI Technical Summary
Existing technologies for semantic segmentation of point clouds of transmission lines suffer from problems such as insufficient representation of features of small linear objects, class imbalance, difficulty in distinguishing easily confused spatial structures, and data augmentation strategies that disrupt scene structure, resulting in inaccurate and unreasonable segmentation results.
We employ explicit enhancement of multi-scale geometric features and highly normalized features, combined with a combined loss function and a confusion constraint loss for training. We use a structure-preserving training strategy to segment transmission line point clouds, and improve the segmentation results through structured post-processing.
It significantly improves the recognition accuracy of small linear targets and rare categories, reduces the inter-class confusion rate, enhances the generalization ability of the model and the physical rationality of the segmentation results, and meets the actual needs of power line inspection.
Smart Images
Figure CN122244078B_ABST