STEREO-IMAGING NETWORK DESIGN FOR PRECISE AND DENSE 3D RECONSTRUCTION

This paper presents a novel stereo-imaging network design strategy comprising four steps: datum definition, optimum distance estimation, candidate camera position generation, and lastly clustering and selection of vantage viewpoints. An approximate 3D mesh model of an object from a quick low-cost system is used in the first step of the method. An ellipsoid is fitted to the rough model in order to find a suitable imaging distance regarding range- and visibility-related constraints for a stereopair. Next, a set of viewpoints is provided taking into account the ellipsoid and the calculated distance. Finally, camera positions and orientations are defined by clustering and choosing the vantage viewpoints. A system, called imaging network designer (IND), has been developed based on this method and is evaluated with two different tests. Results show that IND can provide an imaging network, which fulfils the expected precision for the entire object.