LiDAR point cloud registration based on improved ICP method and SIFT feature

With the increasing demand of unmanned combat system in the information war in the future and the development of sensors, LiDAR is widely applied to acquire environmental information for unmanned ground system as an important sensor. In this paper, an improved iterative closest point (ICP) algorithm for solving the registration problem is researched. It is found that the existence of error corresponding points has a serious impact on the registration results after analyzing the traditional ICP algorithm. The image information is fused into the improved algorithm and the SIFT feature of the image is collected. The SIFT feature point is used as corresponding point in the process of the improved ICP algorithm in order to reduce the error corresponding points. The reducing of the closest point search improves the accuracy and efficiency of the ICP algorithm. From simulations, the better performance of the proposed method is achieved in terms of the registration results.