Method for Efficient Point Cloud Registration and Stitching Based on FPFH Descriptors

To improve the accuracy of point cloud stitching, an improved point cloud registration method based on fast point feature histogram (FPFH) descriptors was proposed, addressing the challenge of high keypoint requirements on smooth and flat object surfaces. The method involves performing voxel down-sampling on the point clouds to be registered, using the down-sampled points as keypoints, and calculating FPFH descriptors to describe local features. The point-to-point correspondences between the two point clouds are then estimated, followed by applying the random sample consensus (RANSAC) algorithm to identify inliers, remove outliers, and estimate the initial pose. Fine registration of the point clouds is achieved using the iterative closest point (ICP) method, completing the point cloud registration process. By sequentially registering adjacent point clouds, the rotation and translation matrices are obtained, enabling point cloud stitching. Comparative simulation experiments were conducted using public dataset point clouds and self-scanned stone point cloud data to validate the effectiveness of the proposed method against the traditional ICP method. The results demonstrate that, compared to the traditional ICP registration method, the fitness score of the proposed method is 1.579e−05, significantly lower than the fitness scores of point-to-point and point-to-plane ICP registration methods. For point cloud stitching at different angles in the public dataset, the proposed method achieves an average fitness score of 2.058e−04 and an average MSE of 0.075 for adjacent point cloud registration. For the self-scanned stone point cloud stitching, the correspondences retained after removing mismatched points exhibit a nearly parallel effect, and the stitching results align with expectations. Using down-sampled points as keypoints and calculating FPFH descriptors for point cloud registration not only effectively reduces computational complexity, but also significantly improves registration accuracy. This method, particularly in handling smooth and flat surfaces, provides a solid foundation for point cloud stitching and has practical applicability.