Methods for Fitting CSG Models to Point Clouds and their Comparison

T. Rabbani and F. van den Heuvel (The Netherlands)


Surface fitting, Reverse Engineering, 3D Modelling, Point cloud processing, Object Recognition


We present three different methods for fitting CAD models to point clouds. Constructive Solid Geometry (CSG) is used for representing the models because of its flexibility and compactness. A given CSG tree is converted to a Boundary Representation (B-rep) or a triangular mesh or a point cloud for approximating the orthogonal distance of a given point from the model surface. We name the resulting methods as ICS (Iterative Closest Surface-point), ICT (Iterative Closest Triangle point) and ICP (Iterative Closest Point-cloud-point) respectively. Levenberg-Marquardt algorithm is used for the minimization of the sum of orthogonal distances. We introduce the notion of Internal Constraints representing the geometric relationships among constituent components of a CSG tree. We show that their inclusion reduces the number of free parameters and results in a more faithful and robust estimation as complete a priori geometric information is utilized. The results of applying these methods to different objects are presented and compared, which bring out the necessity of using a hybrid methodology for achieving accuracy at a practical speed.

Important Links:

Go Back