Abstract
This paper deals with a reverse engineering system which automatically constructs the mathematical model of a physical model based on measurement data. Four methods are proposed; the first one is to smooth measured data by fitting a quadric surface to measurement data, the second one is to uniform measurement data density by converting measurement points into 3D lattice point, the third one is to segment measurement data based on maximum curvature and normal direction, and the last one is to define NURBS surfaces for each segment creating high quality boundary lines between surfaces based on analysis of designer's intention. In order to validate the effectiveness of the proposed method, reverse engineering system is developed and the mathematical model is constructed through basic experiments.
