Automatic DEM generation from low B/H stereoscopic acquisition

The knowledge of ground elevation is essential in most remote sensing applications especially for very high resolution images. This ground elevation information can be retrieved from a pair of stereoscopic images, by correlation methods. The improving resolution of Earth observation systems and their increasing stereoscopic capabilities open up new horizons for automatic Digital Elevation Model (DEM) generation and allow buildings reconstruction to be considered. To reach this goal, the correlation methods used for computing disparities has to be improved and adapted to urban scenes. This paper aims to describe a new method of DEM generation from a stereoscopic pair of high resolution images, whether aerial or satellite, fitted to urban scenes, focusing on correlation improvement. The proposed framework relies on a multiscale dense correlation method with a step of regularisation/interpolation and a step of filtering. It efficiently improves the quality of disparity maps, by reducing the level of noise, and allows us to generate high quality DEM from high resolution images.