الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 144 |  |  |
| | | from | 144 | | |
Abstract
The purpose of this study is to investigate the impact of Digital Elevation Model (DEM) vertical accuracy on the orthoimage production process and how to remove relief displacement caused by buildings’ heights when DSM data is not available. The study took place on a QuickBird image through using four different DEMs with different vertical accuracies, 9 GCPs, three sensor models and land surveying data for eight buildings having different shapes, sizes and heights. The four DEMs are: GPS based DEM with vertical accuracy 2.13m; topographic map based DEM with vertical accuracy 3.58m; Shuttle Radar Topography Mission (SRTM) based DEM with vertical accuracy 6.23m; and Advanced Spaceborne Thermal Emission and Reflection (ASTER) based DEM with vertical accuracy 9.78m. The used sensor models are: the rigorous model; Rational Functional Model RFM and refined RFM. First, the GPS based DEM was used with rigorous, RFM and Refined RFM sensor models and the planimetric accuracies of the three generated orthoimages were 0.56m, 13.68m and 0.50m, respectively. After that the topographic map based DEM was used and the planimetric accuracies of the generated orthoimages were 0.88m, 13.83m and 1.00m, respectively. The planimetric accuracies of orthoimages produced using SRTM based DEM were 1.65m, 12.91m and 1.33m, respectively. Finally, the ASTER based DEM was used and the produced orthoimages were at planimetric accuracies of 2.69m, 13.03m and 2.59m, respectively. The results show that the accuracy of DEM has great impact on the produced orthoimage since the planimetric accuracy of the orthoimage is around quarter of the vertical accuracy of the used DEM. The results show that the RFM (without GCPs) gives inaccurate results in case of high spatial resolution satellite images. The results of buildings evaluation shows that moving and scaling the digitized rooftops improved the average completeness by 25%, while improved the average correctness by about 23%. The research recommends the use of rigorous model and the refined RFM for orthorectifying images accurately as the results showed that these two models give accurate, precise and stable results also their results do not contain systematic errors.