Phi-Descriptor Based Fuzzy Modeling for More Spatial Relations

In numerous fields of computer vision and pattern recognition, spatial relations between image objects play an important role. Descriptor serves as a base for molding spatial relationships. Numerous descriptors have been proposed that have applications in areas like image recognition, extraction, indexing of images, image comparison, description of the scene, and map-to-image conflation, but these descriptors are immature to explore complex relations like “between”, “surround” and “among”. A recently proposed Phi-descriptor having many good features as compared to the other position descriptors is extended for these relations. Phi-descriptor allows the extraction of a wide range of spatial relations including distance and directional relations, the set and non-set topological relations. The important property of the Phi-descriptor is its linear time complexity for raster objects which make the descriptor computationally fast. In this paper fuzzy functions are designed for the extraction “between”, “surround” and “among” relations using Phi-descriptor. The novelty of this model is the use of the Phi-descriptor for the extraction of more complex relations that are not taken into account yet. Results of this model are compared with those of existing models on standard image data and are found satisfactory as per expectations.