Research on Point Cloud Fine Registration Method Combined with Colored Grid Projection

Abstract

Aiming at the problem of low accuracy and large position deviation of component point cloud registration in the process of robot spraying, a point cloud fine registration method combined with colored grid projection is proposed. First, the topology of the point cloud is established; secondly, the overlapping part of the point cloud is obtained by finding the nearest point based on the established topology; third, color cluster the point clouds to find the key points; Finally, the corresponding point is found by the similarity between the position of the key point and the RGB value, and the point cloud fine registration is completed. Take the point cloud of the flat class parts and the freeform surfaces class parts as an example, the average distance between the corresponding points before the point cloud is finely registered is about 20 mm, and the average distance between the corresponding points after the completion of the fine registration is about 3 mm, and the accuracy of the point cloud meets the requirements of the robot intelligent spraying for the component point cloud model. The effectiveness and applicability of the proposed algorithm are verified.