Abstract
This paper proposes a flexible technique framework for optimizing the scanning plans of outdoor constructions using photogrammetry-based modeling and mathematical optimization. Our framework is capable of accepting onsite feedbacks and flexibly reflecting them on the recalculation to complete the scanning through an optimized manner, thus the users can make the most use of the executed scans during the ongoing work. Based on a structure-from-motion (SFM) technique and a patch-based meshing technique, taking photos of the target site during a preliminary survey makes it possible to acquire three-dimensional (3D) point clouds and dense 3D surface meshes systematically. Such 3D information enables elaborate estimation of precise visible characteristics of the targeted scanning object from multiple different scanner viewpoints, so that a mathematical programming method can then be used to derive an optimal scan with the minimum number of measurement points and their layout necessary to scan all surfaces of the target object. On the hand, since on-site ground conditions at the planned scanner positions are sometimes unavailable when setting up the scanner, prepared plans may need to be modified or improved. The proposed optimization scenario can accept such modifications by editing the prepared 3D mesh models so as to fit the physical conditions. The proposed method was applied to an actual field of a castle ruin, where the stone walls need to be managed for conservation. A unmanned aerial vehicle (UAV) was used to collect photos for photogrammetry-based 3D modeling and initial optimization of the scanning plan, which needed to be modified to change the number of scans and a scanner position due to the ground conditions and requests from collaborating investigators on the spot. Eventually, the quality of the acquired point cloud was adequate for further investigations in terms of coverage and density with minimum scans.
