Research on Automatic Registration Method of Three Dimensional Point Cloud for Robot Patrol Trajectory in Substation

摘要

The dynamic changes in the substation environment, such as lighting changes, equipment movement, occlusion, etc., may cause inconsistency in point cloud data and affect the accuracy of matching. Therefore, a research on automatic registration method for three-dimensional point cloud of substation robot inspection trajectory is proposed. By dividing the core trajectory feature area and edge trajectory feature area of the substation robot inspection image, ensuring the correspondence between the center matching points of the associated image, the robot inspection trajectory features are extracted; Calculate the similarity between trajectory 3D point clouds, remove non registered point clouds, calculate the main direction of robot inspection image registration processing, implement detail enhancement processing on trajectory 3D point clouds, and achieve automatic registration of substation robot inspection trajectory 3D point clouds. The experimental results show that the average deviation of the automatic registration angle for the 3D point cloud of the inspection trajectory in this article is 0.42 °, and the average deviation of the automatic registration distance for the 3D point cloud of the trajectory is 0.35cm, which reflects the accuracy of the automatic registration of the 3D point cloud of the trajectory.