Adaptive algorithm for accurate 3D reconstruction of solid surface from unorganized data

In this paper, a simple method for direct acquisition of a cloud of points from an object surface is presented together with a novel surface reconstruction algorithm used to obtain an explicit 3D model for the measured object. This technique is aimed at relatively accurate but simple and reliable 3D reconstruction of anatomical parts for biomedical applications. The whole system is based on acquisition of unorganized range data by laser scanning and successive image processing, by expanding and fitting a regular geometrical model within the range data, for surface reconstruction. The prototype system proved to be working finely, with an estimated resolution <10 mum, a repeatability <50mum, an acquisition noise of ~17 mum (rms value), and accuracy of ~80 mum