An Improved ICP Algorithm Based on 3D-Harris Extraction of Key Feature Points

Point cloud alignment is an important step in spatial 3D reconstruction technology. The traditional point cloud alignment ICP algorithm is studied, and the 3D-Harris-based ICP algorithm is proposed for its problems such as large influence by the initial position of the point cloud, low iteration efficiency and large alignment error. First, the input point cloud is downsampled to simplify the point cloud data; then the 3D-Harris algorithm is used to extract the key feature points of the point cloud and calculate the fast point feature histogram feature description values of the key feature points; then the optimal rigid body transformation matrix of the point cloud to be aligned and the target point cloud is obtained by the sampling consistency initial alignment algorithm; finally, the ICP algorithm combined with the Kd-Tree data structure is used to accelerate the finding of the corresponding point pairs of the point cloud to complete the point cloud alignment. The experimental results show that the alignment error of the algorithm in this paper is smaller when the same number of iterations are performed, while the required alignment time is significantly reduced. This algorithm reduces the dependence of the ICP algorithm on the initial position of the point cloud, while ensuring good alignment, and can play an active role in practical use.