Clearing the wood - evaluation of different software packages for ALS filtering

Abstract

During the last decade high quality digital terrain models (DTM) and digital surface models (DSM) became very important information sources for archaeological interpretation. This is due to the fact that the terrain structures represented in the models give not only knowledge about the contemporary world but include also features remaining in the topography of past landscape. Airborne laser scanning (ALS) proved to be very useful for generating dense and accurate DTMs and DSMs. The data collected by the ALS systems represent the whole landscape in a very detailed manner: vegetation, terrain, buildings etc. In order to obtain the information essential for the archaeologists the point cloud must be filtered. The aim of the filtering is to separate all man-made objects and vegetation from the terrain i.e. a classification of the point cloud into terrain and off-terrain points. This process is usually done automatically using various filtering algorithms implemented in numerous software packages. However, its outcome is highly critical for further processing and interpretation of the data for the intended purpose and thus deserves a more detailed examination. Currently many ALS data processing software packages, which are able to perform the automatic ground-points extraction, can be found on the market. The goal of the presented research is to evaluate the available software packages by testing: the filtering procedures used by the software, the accuracy of the results and the difficulties that occurred during the work. The outcome is a library of filtering parameters that gave good results for chosen type of point cloud and terrain. The experimental comparison is based on three reference point clouds, which have been filtered in a manual way. They are located in Leitha Hills (Lower Austria), a region in the focus of integrated archaeological research since many years with a broad temporal and typological variety of remnant archaeological structures. The data used in the project was collected over this area using a Riegl Airborne Laser Scanner LMS-Q560 on the 8th of April 2006. The information gathered during the research will be published in a dynamic way by creating a web-page. The web-page should facilitate the access to various filtering methods, software packages and parameters. It would be used as an exchange platform in order to share the experience in ALS filtering. The outcome is dedicated not only to ALS beginners but to the whole ALS community.