Coordinate grammars are discussed and alternative notation variants are given. The lack of generality and feasibility of Rosenfeld's later notation leads back to the original notation of productions. The interaction of the productions is depicted by production nets. From the structure of these nets a hierarchy of computational complexities for coordinate grammars can be inferred. As an example a grammar is given for the recognition of certain vehicles in ground-based visual spectrum domain picture sequences. It uses coordinates in 2D picture space and 3D scene space. The analysis process runs purely bottom-up and data-driven. Relations to some nonsyntactic paradigms of pattern recognition like production systems or semantic nets are mentioned. Emphasis is more on the practical use of such structures for complicated pattern recognition tasks and less on the theoretical survey.
«