Optical Graph Recognition on a Mobile Device
Christopher Auer, Christian Bachmaier, Franz Josef Brandenburg, Andreas Gleißner, Josef Reislhuber · 2013
In [1] we proposed Optical Graph Recognition (OGR) as the reversal of graph drawing. A drawing transforms the topological structure of a graph into a graphical representation. Primarily, it maps vertices to points and displays them by icons, and it maps edges to Jordan curves connecting the endpoints. In reverse, OGR transforms the digital image of a drawn graph into its topological structure. The recognition process is divided into the four phases preprocessing, segmentation, topology recognition, and postprocessing. In the preprocessing phase OGR detects which pixels of an image are part of the graph (graph pixels) and which pixels are not. The segmentation phase recognizes the vertices of the graph and classifies the graph pixels as vertex and edge pixels. The topology recognition phase first recognizes edge sections. Edge crossings divide edges into edge sections, i. e., the regions between crossings and vertices. The edge sections are merged into edges in the most probable way based on direction vectors. The postprocessing phase concludes OGR with tasks like converting the recognized graph into different file formats, like GraphML or adding coordinates to the