Vectorising Building Footprints From Historic Maps

Historic maps provide fascinating insights into the growth and development of cities over large time periods, and often the places they document have changed dramatically since their creation. However, rasterised image data is limited in that it is not easily manipulatable, for either analysis or presentation of data, and it is not easy to use the data as a basis for reconstructions. If it were possible to extract vector outlines for building footprints from the map then these limitations would be removed, and the vector outlines could easily be used as a basis for three dimensional models representing the buildings. Manually defining building outlines from maps is a laborious process and so this paper investigates the use of image processing techniques to enable a semi-automatic process to detect vector outlines of buildings from a variety of maps and to optimise the output for use as a basis for creating large scale three dimensional reconstructions.