DBPapers
DOI: 10.5593/SGEM2016/B22/S10.130

SYSTEM OF AUTOMATIC 3D-RECONSTRUCTION OF URBAN SCENES FROM SATELLITE IMAGES BASED ON A PRIORI INFORMATION

A.Zakharov, A. Tuzhilkin, A. Zhiznyakov
Thursday 8 September 2016 by Libadmin2016

References: 16th International Multidisciplinary Scientific GeoConference SGEM 2016, www.sgem.org, SGEM2016 Conference Proceedings, ISBN 978-619-7105-59-9 / ISSN 1314-2704, June 28 - July 6, 2016, Book2 Vol. 2, 1019-1026 pp

ABSTRACT
The development of the system of automatic 3D-reconstruction of urban scenes from
satellite images is presented. Brief review of methods for three-dimensional
reconstruction from images is shown. The structure of the system is given. Threedimensional reconstruction process consists of the following steps: image
preprocessing, image orientation, epipolar image generation, image matching, digital
surface model generation, image orthorectification, object detection, removal of
vegetation and shadow areas, building detection, decomposition of the building contour, recognition and reconstruction of 3D-models. The reconstruction is based on image recognition and search of the three-dimensional object model in the database. The reconstructed objects are detected previously. The detection is based on joint use of height and color. Selecting and decomposition of the contour of the building are
performed at an early stage. Recognition is based on comparing the geometric
histograms of standard object and images. Image roofs are used for reconstruction. The results of the system are given.

Keywords: 3D-reconstruction, satellite image, urban scenes