Point Object Extraction from Scanned Topographic Maps for the Digital Topographic Maps Production

Topographic maps published by the Military Geographical Institute are an important cartographic source in the process of creating a digital topographic map in the scale of 1:25 000. Existing topographic maps are scanned and georeferenced, but vectorization of the required content from these maps is done manually, which requires a lot of time. The homogeneity of cartographic content gives the potential for the application of algorithms for automatic vectorization of geospatial data. This paper presents a tool programmed in the Python programming language that extracts point symbol - well objects (as a hydrographic object) from a georeferenced map, by identifying objects based on assigned samples and generating vector point spatial data. Recognition is performed on a processed input raster containing only shades of blue. The output data are the geographical coordinates of the identified objects and the initial raster with markings at the places where the objects are recognized. 1074 wells on 24 map sheets in the wider area of Novi Sad and Zrenjanin were generated. Proposed Points from Corner method showed satisfactory positional accuracy results.