images

Abstract — High resolution IKONOS stereo imagery provides an important data source for three- dimensional scene extraction. In urban areas, it has been long a challenge topic to automatically extract urban objects from images due to the high object density and scene complexity. In this paper, we propose a new method for automatic 3d building reconstruction from IKONOS stereo images. Our approach is based on extraction and characterisation of objects from the scene imagery. Objects are characterised by their contours and Texture. The matching process is object oriented and combines objects geometrical features and textural characteristics. Results show that, in the case of 3D urban scenes reconstruction, combining texture features with geometrical features is well suited for matching the stereo images. I