International Science and Technology Journal

City models extracted from high resolution satellite images have huge potential for supporting the smart city concept, such as urban mapping analysis, telecommunications applications, and environment evaluations etc. The combination between semi global matching algorithm and Pleiades triplet images may improve the correlation accuracy of city models extraction, and reduce the consuming time for urban mapping. The objective of this paper is to test the new combination of Pleiades triplet images and advanced semi-global matching algorithm for semi-automatic buildings model extraction. Well- distributed control points, covering the study area were used for images orientation, and the obtained geolocation results were improved, with planmetric accuracy of 0.50m. The oriented Pleiades images with resolution of half meter were processed and new Epopilar Pleiades images with resample of half meter was obtained and used for geocoded digital surface model extraction. OrthoEngine has included new state-of-art of advanced matching algorithms was used for 32 bits DSM extraction from Pleiades Epipolar images. The obtained DSM was visually tested against real buildings 2D boundaries and shapes, whereas the narrow streets were appearing in DSM. Buildings model height accuracy was also evaluated against the real heights, with height accuracy of 0.30meter. This was substantial improvement of reducing hidden pixels and improve better matching results for 3D urban extractions ″Buildings model boundaries limits and height details″. This means, the combination between semi global matching algorithm and Pleiades triplet images can satisfy the required accuracy for different applications in urban areas.