Registration based on Lie group representation for point sets

We propose a registration method for two point sets. Given two point sets, i.e. the source and target sets, our method first finds the point pairs in corresponding datasets, then maps the source to the target dataset via scaling, rotation and translation transformations by minimizing an energy functional. The novelties of this algorithm lie in that we generalize the Lie group framework to the trimmed iterative closest point algorithm. The experimental results of LiDAR data acquired from Mount St. Helens demonstrate that, compared with some state-of-the-art algorithms, our algorithm is more accurate and robust.