An inverse problem approach for the segmentation of snow cover in satellite images

A new model for the correction of topographic effects in satellite images of rough terrain is described. The model simulates a synthetic image of the scene using a computer graphics approach which combines ray-tracing techniques with radiosity methods. Computation is structured on three levels: a macro level in which the image is described by the Digital Elevation Model and the light source, a meso-scale in which the model simulates the integration effect of the imaging sensor and a micro-scale which is characterized by the reflectance of the snow cover (specular and diffuse). The parameters of the model are tuned with a gradient search to fit real images acquired by the Landsat-TM sensor. The results show a better accuracy than the classical "cosine of incidence" and Minnaert models. Additionally a new technique based on maximum entropy estimation is used to determine the reflectance function of snow and compare it with the one predicted by our model.