A similarity measure of fuzzy attributed graphs and its application to object recognition

Attributed graphs (AG) or attributed relational graphs (ARG) and fuzzy attributed graphs (FAG) have been used for representing objects or scenes in image understanding. Based on, matching of two AGs or FAGs, similarity between two objects or two scenes can be measured. Since the matching belongs to a NP-hard problem, there have been many attempts to reduce the computational complexity of the matching. This paper uses FAGs to represent objects and gives several algorithms to measure the similarity of two FAGs. Its approach for matching is based on Eshera and Fu´s method (1984). Our algorithm is extended to the case of FAGs and is simpler and reasonable. As an example, some line drawings of chairs are used for verifications. A method for representing a line drawing with a FAG is given. With two indexes for similarity and compatibility between two FAGs, identification and inclusion relations between two line drawings are studied