Fuzzy Geometric Relations to Represent Hierarchical Spatial Information
This work addresses the need for handling vague spatial data in domains like military applications, but it appears incremental as it builds on existing fuzzy constraint and pattern recognition concepts.
The paper tackles the problem of representing imprecise spatial information by introducing a model based on fuzzy geometric relations and spatial templates, and shows how to match templates against crisp situations to recognize instances with a proximity measure for similarity.
A model to represent spatial information is presented in this paper. It is based on fuzzy constraints represented as fuzzy geometric relations that can be hierarchically structured. The concept of spatial template is introduced to capture the idea of interrelated objects in two-dimensional space. The representation model is used to specify imprecise or vague information consisting in relative locations and orientations of template objects. It is shown in this paper how a template represented by this model can be matched against a crisp situation to recognize a particular instance of this template. Furthermore, the proximity measure (fuzzy measure) between the instance and the template is worked out - this measure can be interpreted as a degree of similarity. In this context, template recognition can be viewed as a case of fuzzy pattern recognition. The results of this work have been implemented and applied to a complex military problem from which this work originated.