Abstract
For recording of the disaster site and cultural asserts, laser sanning and/or SFM (Structure from Motion) with UAV are often utilized, since a laser scanner is capable of capturing “as is” situation by sampling the surface shapes of objects as “point cloud” sets and SFM is a powerful image-based modeling technique whici can utilize the mobility of unmanned areal vehicles (UAVs) equipped with high-resolution camera, recently. However, laser scanning may has limitations in both visibility from fixed viewpoint and footholds for installation of the tripod on the ground, while SFM is not capable of obtaining real dimension in nature. Therefore, we propose a method to integrate to utilized both measurement in complimentary way. A laser sanner is usually equipped and calibrated with a color digital camera in order to obtain color information corresponding to scanned points. This feature allows the scanner viewpoint to be integrated into the 3D geometory reconstructed by SFM. More specifically, the 3D geometry produced by the SFM and the laser-scanned data from scanner position viewpoints are overlaid upon each other. Then, selecting the proper mumber of points from the scanner view enables to pick up the proper corresponding point pairs to find the 3D transformation to fit them together without incremental optimization. In this study, we conducted an experiment using a UAV (DJI phantom II) and Laser scanner (Riegl LMS-Z420i) equipped with a hi-resolution camera, and found that the resultant 3D data set consisting of complementarily scanner point cloud data and SFM process data covered the entire surface shape of the scene.
