Peter DORNINGER1 / Robert KALASEK2 / Ingrid ADENSTEDT3
(1TU Wien , Institut für Photogrammetrie und Fernerkundung, Vienna, Austria / 2TU Wien, Department für Raumentwicklung, Infrastruktur- und Umweltplanung, Vienna, Austria / 3Österreichische Akademie der Wissenschaften, Institut für Kulturgeschichte der Antike, Vienna, Austria)
Outline: We present methods for automated point cloud processing and subsequent analysis for documentation purposes in archeological applications. The results comprise 2D floor map generation, the extraction of scaled, color coded plots of wall scenes and 3d model generation. We compare the results of manual and automated documentation methods based on results achieved at an archaeological excavation site. Additionally, we figure out the qualitative advantages of manual documentation and evaluate the economical advantages of automated data processing.
Abstract: In June 2010, the Dwelling Unit 7 of the Terrace House 2 at Ephesos, Turkey, was scanned from 172 scanning positions covering the object completely. Altogether, approximately 10 billion laser scanning points and a set of control points were acquired. In order to enable a global registration of the given datasets, we applied a robust 3d filtering method on the individual scans to reduce the amount data. After the registration, the points were merged to one point cloud by means of the same 3d filtering process. The main task of the project was the determination of 2D floor maps and the generation of scaled projections of numerous walls for further interpretation. For this, we applied manual as well as automated methods. Within this contribution, we discuss the differences in quality and quantity of the two approaches. Special focus is drawn on the qualitative benefit of manual methods with respect to the quantitative benefit of the automated processing methods. While the manual analysis is based on commercial software tools, the automated processing is realized by innovative methods for data selection, projection and filtering in order to achieve the expected results. Additionally, hybrid approaches aiming at introducing as much automation to support the manual work while ensuring high qualitative results are discussed. The conclusion of this contribution is that automated data processing for documentation purposes at archaeological sites is economically preferable to manual documentation, if large scale projects are to be dealt with while certain restrictions on the quality requirements are acceptable, as automated processing serves a reliable foundation for subsequent large scale interpretation of archaeological sites and higher repetition rates.
Keywords: Laser scanning, prospection, automation, mapping, 3d modeling