Abstract: In order to improve the effectiveness of power line corridor inspection, a power line extraction method based on airborne point cloud was proposed. Firstly, the gradual triangular network filtering was used to eliminate the ground points, and the height difference from the ground was calculated to obtain a set of points containing power lines. In order to overcome the shortcomings of traditional two-dimensional density segmentation methods that cannot effectively identify the ground object points below the power line, three-dimensional grid division density segmentation method was adopted to achieve rough extraction of power line points. Through the RANSAC method, the line detection and point cloud height difference were used to analyze and eliminate the residual tower points and insulator points, so as to achieve the precise extraction of power line point clouds. In addition, the power lines on the same vertical plane were split by the curve detection of RANSAC method, and finally the K-Means algorithm was used to complete the extraction between the split wire bundles. The experiment showed that the accuracy of this method for extracting power line point clouds was more than 97%, and the misjudgment rate of non-power line point clouds was as low as about 1%, which can fully meet the needs of practical projects.