A Fusion Method for Reconstructed Point Clouds of 3D Laser Point Clouds and Visible Light Images Based on Power Grid Operations

In response to the safety monitoring needs of power grid operators, a high-precision point cloud fusion method for reconstructed point clouds of three-dimensional (3D) laser point clouds and visible light images is proposed in this paper. This fusion method utilizes the camera pose calculated during the visible light image 3D reconstruction process and optimizes the initial value selection of the centroid position of the 3D point cloud model, improving the registration accuracy and efficiency of the iterative closest point (ICP) algorithm. At the same time, based on the characteristics of the two sets of point clouds, the Euclidean distance threshold is used to remove outliers from the edges of the 3D point cloud. Finally, the optimized ICP algorithm is used to obtain a high-precision 3D reconstructed point cloud with scale information fusion. The experimental results show that the fusion method proposed in this paper can effectively improve the point cloud accuracy of 3D reconstruction of spatial targets.