Co-Registration of Large Volume Laser Scanner Point Clouds: The Pinchango Alto (Peru) Data Set (original) (raw)

2005, Internal Technical Report at IGP - ETH, Zurich, July, 21 pages.

CO-REGISTRATION OF LARGE VOLUME LASER SCANNER POINT CLOUDS: THE PINCHANGO ALTO (PERU) DATA SET Internal Technical Report by Devrim Akca Group of Photogrammetry and Remote Sensing (P+F), ETH Zurich www.photogrammetry.ethz.ch Presented to Prof. Dr. Armin Gruen Prepared for the project 3D Modeling of the pre-Inkaic site Pinchango Alto, Peru a cooperation of Group of Photogrammetry and Remote Sensing of ETH Zurich and German Institute of Archaeology (KAVA, Bonn, Germany) Swiss Federal Institute of Technology (ETH) Zurich Institute of Geodesy and Photogrammetry ETH-Hoenggerberg, CH-8093 Zurich July 2005 Conclusions The disadvantages of the target based registration of the laser scanning point clouds are well known. Adopting a target based registration approach requires more fieldwork and personnel, i.e. setting up those targets to the site and measuring them using a theodolite or a GPS system. It is apparent that additional geodetic measurement devices increase the equipment cost as well. The target based registration methods cannot exploit the full accuracy potential of the data, due to additional errors introduced by the geodetic measurements. Although the laser data naturally has very high level of redundancy, the target based registration techniques use only a very small portion of the data. This is the second reason causing to degrade the accuracy potential. Surface based registration techniques stand as efficient and versatile alternative to the target based techniques. They offer better registration results while keeping the project cost lower. In this study we showed the capabilities our surface based registration method applying to the Pinchango Alto laser scanning data set. Our proposed method, the Least Squares 3D Surface Matching (LS3D), estimates the transformation parameters between two or more fully 3D surfaces, using the Generalized Gauss Markoff model, minimizing the sum of squares of the Euclidean distances between the surfaces. The mathematical model is a generalization of the least squares image matching method and offers high flexibility for any kind of 3D surface correspondence problem. The least squares concept allows for the monitoring of the quality of the final results by means of precision and reliability criterions. The Pinchango Alto data set stands as a special case due to the huge volume and many occlusions on the data. The practical example shows that our proposed method can provide successful matching results in reasonable processing times. It exploits the full accuracy potential of the data owing to its powerful mathematical model. The following up step surface modeling was performed by use of commercial software packages. However, it was not possible to model in the original resolution. Due to not efficient memory management capabilities of the software packages, the modeling had to be performed at reduced resolution. The modeling of 3D laser scanner point clouds is still a troublesome step and sophisticated algorithms need to be developed with real 3D capabilities.