Recognition of motion events of directed points using a language based framework

In this paper, we have proposed a framework for representation and recognition of motion events of spatial objects abstracted as directed points. The framework combines qualitative spatial reasoning with formal grammars in order to recognize motion events in an input stream. Motion events are represented by three qualitative attributes, namely, orientati on, direction and distance. It has been shown that by using spatio-temporal continuity of qualitative relations, it is possible to use regular grammars for event recognition. The novelty of the approach is that any number of qualitati ve relation sets can be combined with formal grammars and hierarchies of abstractions can be created to make the knowledge representation closer to human cognition.