Enhanced Automatic Span Segmentation of Airborne LiDAR Powerline Point Clouds: Mitigating Adjacent Powerline...

Area of Science:

• Geomatics Engineering
• Computer Vision
• Remote Sensing

Background:

• Automatic transmission corridor inspection relies on 3D reconstruction of powerline point clouds from airborne LiDAR.
• Extracting accurate powerline data is challenging due to interference from adjacent or crossing powerlines.

Purpose of the Study:

• To develop an efficient, automated method for powerline span segmentation from LiDAR point clouds.
• To provide clean data for automatic reconstruction of powerline catenary curve models.
• To address and suppress interference from adjacent powerlines during data processing.

Main Methods:

• A novel point-counting grid (PCGrid) based fast density clustering algorithm (DBSCAN) is developed.
• The method utilizes spatial relationships between pylons and powerlines in LiDAR data.
• It involves 2D/3D density clustering for pylon extraction and point cloud filtering, pylon connection verification, and span assignment.

Main Results:

• The PCGrid structure significantly accelerates clustering efficiency.
• The method achieves fully automated span segmentation.
• Adjacent powerline interference is effectively suppressed, yielding reliable results for 3D powerline reconstruction.

Conclusions:

• The proposed method offers an efficient and automated solution for powerline span segmentation.
• Integration of PCGrid clustering and spatial-relationship-driven pylon verification enhances 3D powerline reconstruction accuracy.
• This framework is vital for improving automatic transmission corridor inspection processes.