Automatic registration of laser point cloud using precisely located sphere targets

Abstract

Sphere targets are used extensively in terrestrial laser scanning registration; however, in practice, it is still a time-consuming and labor-intensive task. This paper proposes an automatic registration method for laser point clouds based on sphere targets’ detection. First, a modified eight-neighbors check method is applied to mark occluding edge points. Then, for the sphere targets in the raster structure, occluding edge points are clustered, and circle and sphere detections are sequentially implemented in the cluster node and circular area, respectively. The sphere models that pass through multilevel constraints are considered the final results. Next, triangles constructed using three arbitrary noncollinear sphere centers in each scan station are selected as registration primitives and the area and interior angles of each are selected as similarity measures. Finally, the congruent sphere centers between two scan stations are matched in an iterative manner and used to calculate the transformation matrix. The results of experiments in which a lab was scanned from two locations indicate that our method can effectively detect four sphere targets in more than 10 million point clouds within ∼1.5 min, with the largest position error between congruent points <2 mm.