Geometric modeling of raster images of documents with weakly formalized description of objects

In this paper, we analyze graphic documents with a weakly formalized description of objects (WFGD) and reveal their main features that influence the choice of models, methods and algorithms for processing such documents. In the framework of the development of the combinatorial-geometric approach, a geometric model for describing WFGDs with a pronounced orientation of linear objects is proposed. We also propose a technology for vectorization of raster images of WFGDs in the presence of noise in the source data. The effectiveness of an extended class of vector models (linear and segment-node models) used for describing WFGDs with a distinctive linear orientation of objects is shown, which was revealed during practical experiments on real WFGDs.