In this paper, the accuracy of 3D models of both laser scanning and photogrammetry is discussed for terrestrial applications. The models constructed by photogrammetry are based on digital stereoscopic restitution and structure from motion. These models, and the models generated by processing laser scanning data, are statistically compared with topographic measurements by a total station. The ‘Sint-Baafs’ Abbey (Ghent, Belgium) is used as a test case in this paper. This historic site contains a series of buildings and ruins, and from these objects, a number of exterior walls are selected for virtual 3D reconstruction. Firstly, the outlines and characteristic points of these walls are measured by a total station. Secondly, a series of extra targets are measured with total station as well, in order to correctly orientate and register multiple point sets from the terrestrial laser scanner. Based on previous research in comparable projects, the angular resolution of the laser scanner must be at least equal to the image resolution, and the distance between the scanner and the wall should be limited. Using the phase-based Leica HDS 6100 laser scanner, the angular resolution was set to 20 mgon, and the distance between the scanner and the walls was limited to 15-20 m. Thereafter, the walls are photographed with a high resolution digital camera. The acquisition points of the images were not recorded, but the measured characteristic points on the walls are used for orientation of the images and for the 3D geometry derivation from the images. The different images and laser scanning data sets were processed independently, using the total station data as geometric reference. In order to compare the final results, an extra series of check points from this total station were used. An iterative closest point algorithm is also performed to compare the models.
No takes yet. Share an insight, caveat, or question.
Stal et al. (2012) studied this question.