Framework for multi-sensor data fusion using template based matching

A general approach or framework is proposed for multi-sensor data fusion using template based matching (TBM). The main advantage of TBM is that it allows defining features/templates using a priori information from the scene/image. The approach works as follows: first quite complex features e.g. roundabouts/junctions are extracted in the optical image, then these features are simulated in SAR and finally they are compared or matched on patch/area basis with the SAR image. The proposed framework is illustrated for common tie points extraction in very high resolution satellite optical WorldView-2 and radar TerraSAR-X imagery.