Sphere Target Fitting of Point Cloud Data Based on M-estimation

In terrestrial laser scanning, sphere target is an important accessory. With it, multiple point clouds data obtained form different viewpoints can be transformed into a common coordinate system. In order to complete the task mentioned above, we must determine centre of sphere target. Least square method is commonly used to locating sphere center coordinates, but it has no ability to resist outliers, and thus results in poor estimated parameters. A new method to determine centre coordinates of sphere target is presented. The method is based on M-estimation, and can eliminate the effects of outliers on sphere center determination. Experiments with simulation data and real data are conducted. Results show that the proposed method has the advantage of robustness, and can overcome the effect of outliers, improving the accuracy of sphere centre coordinates.