Study on a Road Surface Point Cloud Stitching Method Based on Improved Euclidean Clustering Matching

To address the contradiction between large-area data acquisition and high-precision measurement requirements in the process of road 3D measurement, this paper proposes a road 3D point cloud stitching method based on line-structured light image acquisition; first, artificial feature objects are placed in the overlapping regions of the point cloud, feature information is obtained through 3D data acquisition, the Random Sample Consensus (RANSAC) algorithm is used to achieve effective separation of ground objects and background, and an information entropy-improved Euclidean clustering algorithm is introduced to complete accurate classification of target features; then, a matching strategy is formulated based on spatial relationships to perform feature matching on the clustering results; finally, appropriate matching pairs are used as input data for point cloud registration, and the Super-4PCS algorithm is adopted for coarse registration while the ICP algorithm is used for fine registration to complete the road point cloud stitching.