An Energy-based Paradigm for Multimodal Images Fusion L. Brunie and P. C in q u in

One of the most basic abilities of any human or artificial intelligence is the inference of knowledge by matching various pieces of information [20]. Thus, in computer vision, the analysis and the interpretation of a scene can rarely be performed using a single source of information. In order to remove remaining uncertainties and ambiguities, additional information is required. Two main approaches can be used: either reconstruct the scene from a large set of data [11,19]; or, when only a few data are available, introduce a priori knowledge to compensate for the lack of information and match it with the data. Such is the frame of model-based computer vision.