Applications Of Model-Based Image Interpretation By Line Structure Description
Eliane Egeli · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 1984
A general scheme is given for the processing of images containing patterns that can be described by line structures: first lines or/and edges are extracted from the original gray-level picture. Then topological operations are applied to reduce the resulting line structure to an unambiguous skeleton representation. This preprocessed version is translated into an equivalent graph, retaining only the geometrical and the topological information. Finally these transformed data are analysed to yield a set of model-dependent attributes that constitute the base for the interpretation. The practical use of this scheme is demonstrated by means of two different applications: restoration of cartographic data and automated analysis of blood samples.