Point Clouds Preprocessing for Better Registration

Abstract

In this article, we review point clouds preprocessing algorithms that enhance the results and performance of point clouds registration. Registration is a problem of merging two or more point clouds into one common. If a point cloud is obtained from a real physical object, it definitely has some noise and outliers. This noise can negatively affect the registration process and often prevent from good results. Point clouds preprocessing reduces noise and helps to get better registration results. Also we present several experiments on real dataset and results.