Title: Refinement of Digital Surface Models through Constrained Connectivity Partitioning of Optical Imagery
Authors: POLI DANIELASOILLE Pierre
Citation: ISPRS Hannover Workshop 2011 "High-Resolution Earth Imaging for Geospatial Information" vol. XXXVIII
Publisher: ISPRS
Publication Year: 2011
JRC N°: JRC63527
URI: http://www.isprs.org/proceedings/XXXVIII/4-W19/paper/papers.htm
http://publications.jrc.ec.europa.eu/repository/handle/JRC63527
Type: Contributions to Conferences
Abstract: In the last years the number of Earth-observation platforms equipped with high and very high resolution imagers (VHRSI) has increased. Those sensors are characterized by a number of geometric, radiometric, spectral and operational specifications, and in most cases allow the acquisition of multi-angle imagery for automatic digital surface models (DSM) generation and 3D feature extraction [1]. One of the tasks of ISFEREA action at JRC in Ispra (Italy) is to assess changes in land cover after events like natural disasters, using satellite imagery. To describe the pre-event and post-event scenarios the available heterogeneous data are analyzed and eventually fused in order to achieve the most accurate and reliable information. With respect to automatic 3D information extraction, the availability of accurate and detailed DSMs is a crucial for automatic building detection and subsequent damage estimation. This paper investigates the quality of DSMs obtained from high or medium resolution stereo satellite imagery with commercial photogrammetric software packages and proposes a methodology for the geometric refinement of the DSM using high or very high resolution single scenes. The main idea for DSM enhancement is to fuse it with the scene at higher resolution through an advanced hierarchical image partitioning. The method is applied on different datasets at various resolutions. The first dataset includes a stereopair acquired by IKONOS sensor in 2007 and a single scene by GeoEye1 in 2009. After the refinement of the IKONOS image orientation in a block adjustment, different DSMs with 2m grid spacing were generated automatically using commercial software packages available at ISFEREA. The data were compared in order to evaluate global and local height differences, feature details and behavior of geomorphological parameters according to the land cover. Among the available DSM, the one that was showing the best software performance was chosen for further processing steps. The GeoEye1 (GE1) scene was used for the DSM refinement. The scene was first oriented using GCPs extracted in the IKONOS DSM and orthoimages. For each detected segment the statistical and morphological properties of the covered DSM are computed and a suitable function is chosen to refine the DSM. As a consequence, in the resulting DSM features are more sharpen and better reflect the segment extent. The second dataset consists of stereo scenes of WorldView-1 (WV1) and Cartosat-1 over Catalonia, Spain, provided by ISPRS WGI/4 through the project ¿Benchmarking and quality analysis of DEM generated from high and very high resolution optical stereo satellite data¿. A DSM at 5m grid spacing will be extracted from Cartosat-1 scenes and refined using the Cartosat-1 scenes themselves and the WV1 scenes. Another DSM from WV1 will be generated with 2m grid spacing. Then the produced surface models will be compared to a Lidar DSM on the same area for quality analysis.
JRC Institute:Institute for the Protection and Security of the Citizen

Files in This Item:
There are no files associated with this item.


Items in repository are protected by copyright, with all rights reserved, unless otherwise indicated.