DSM extraction from SPOT5 stereo-pairs

The generation of a digital surface model (DSM) from remotely sensed images is an important task for various applications such as telecommunications, cell planning, the environment monitoring.... In general, extracting DSMs consists of: camera modelling, stereo matching and interpolation. However, in the case of satellite imagery and because of the satellite characteristics, it is difficult to apply the algorithms and the model generally used with aerial images. In this paper, we adopted a model which uses all parameters of acquisition: it is the full model. The matching algorithm used takes into account geometry of the image. It is base on a hierarchical approach followed by a relaxation process. The full model of SPOT5 sensors is then used for the calculation of the 3D-co-ordinates on the ground. Results show good quality and accuracy of the obtained DSM compared to ground control points and DSM data of superior quality.