A spherical targets fitting method for terrestrial laser scanning data

摘要

The dominating measurement error of Terrestrial Laser Scanner comes from laser ranging, which is also the major cause of "tailing points" appearing at the edge of targets scanned by coaxial laser scanner. Since the existence of such abnormal points, general least squares sphere fitting method cannot correctly position spherical targets or determine the dimension of geometries. Base on the working mechanism of coaxial laser scanner, a surface fitting model is designed in this paper for the spherical targets, and utilize robust least squares method to get model parameters from scanning data with obvious edge noise, then conclude the sphere radius and centre point coordinate. In the end, this paper accomplishes two experiments separately for really scanning data and synthesized data subjecting to scanner's error distribution. The result demonstrates that proposed fitting method can weaken the impact of abnormal points on the edge of spherical targets effectively, get geometric parameters of spherical targets accurately, and exactly locate the spherical targets.