Image understanding using fuzzy isomorphism of fuzzy structures

We propose a system architecture able to classify objects into models. Each object is represented by 2D color image. The fuzzy sets theory has been a fundamental base to build algorithms presented here. Each image is segmented into semantically annotated regions. In a second step, we extract structural information which are coded into graphs. At the end, we obtain a semantic graph representing the image. The classification will be done after finding the isomorphism between the 2D image graph and the available model graphs