Semi-automatic 3D object digitizing system using range images

Abstract

Manual object digitizing is a tedious task and can be replaced by 3D scanners which provide an accurate and fast way to digitize solid objects. Since only one view of an object can be captured at once, several views have to be combined in order to obtain a description of the complete surface. In this paper a digitizing system is proposed which captures and triangulates views of a real world 3D object and semi-automatically registers and integrates them into a virtual model. This process is divided into three steps. First, an object is placed at different poses and its surfaces are sensed by a range scanner. Then, the different surfaces are aligned automatically starting from a pose estimate entered interactively. Finally, the overlapping triangle meshes of the registered surfaces are fused in order to obtain one unique mesh for the entire object.