Rectangular building extraction from stereoscopic airborne Radar images

From the recent availability of images recorded by synthetic aperture radar (SAR) airborne systems, automatic results of digital elevation models (DEMs) on urban structures have been published lately. This paper deals with automatic extraction of three-dimensional (3-D) buildings from stereoscopic high-resolution images recorded by the SAR airborne RAMSES sensor from the French Aerospace Research Center (ONERA). On these images, roofs are not very textured whereas typical strong L-shaped echoes are visible. These returns generally result from dihedral corners between ground and structures. They provide a part of the building footprints and the ground altitude, but not the building heights. Thus, we present an adapted processing scheme in two steps. First is stereoscopic structure extraction from L-shaped echoes. Buildings are detected on each image using the Hough transform. Then they are recognized during a stereoscopic refinement stage based on a criterion optimization. Second, is height measurement. As most of previous extracted footprints indicate the ground altitude, building heights are found by monoscopic and stereoscopic measures. Between structures, ground altitudes are obtained by a dense matching process. Experiments are performed on images representing an industrial area. Results are compared with a ground truth. Advantages and limitations of the method are brought out.