Sequential Bayesian methods for Resolution Enhancement of TIR Image Sequences

The availability of remotely sensed image sequences characterized by both spatial and temporal high resolution is crucial in many applications, ranging from agriculture to Earth surface hazard monitoring. To date, image sequences presenting such desirable characteristics in both domains are not directly obtainable by a single device and thus a viable solution is represented by the joint use of multisensor information. In this work, we propose a solution, based on Bayesian sequential estimation, for fusing two image sequences characterized by complementary features. Together with the assessment of two different sequential estimation approaches, a novel method for constructing a sharpened observations is presented here. The proposals are then evaluated by employing different datasets acquired by the SEVIRI and MODIS sensors, showing remarkable improvements with respect to classical approaches.