Grouping Line Drawing Elements Based upon Their Area Voronoi Tessellation

We present an algorithm for grouping multipart symbols, dashed lines, and character strings for extraction from line drawings. Initially, the image undergoes a lossless raster-to-vector conversion creating as its vector representation an undirected graph, a so-called run graph. Next, during localization, the connected components of the run graph are extracted and classiied prob-abilistically from their geometric features using a decision tree. The result of the classiication are sets of dashes, dots, circles, symbols, and graphics. An area Voronoi tessellation of the members of the sets is constructed, from which a neighborhood graph is derived. In this graph, the nodes represent the image elements and the edges the connections to their nearest graph neighbors. The graph is then traversed to group the members of the various sets into candidate character strings, dashed lines, and multipart symbols for extraction and for input to diierent recognition modules. No a priori font or other domain speciic information is required for the grouping: neighborhoods are guaranteed to be minimal and complete, and no special geometrical relationships among the elements are assumed. Results from the localization and grouping are presented with example images taken from those used by our Swiss cadastral map understanding system.