Automating Information Extraction from Large Historical Topographic Map Archives: New Opportunities and Challenges

Abstract

Historical maps constitute unique sources of retrospective geographic information. Recently, several archives containing historical map series covering large spatial and temporal extents have been systematically scanned and made available to the public. The spatial-temporal information contained in such archives represents valuable information for a myriad of scientific applications. However, this geographic information needs to be unlocked and provided in analysis-ready geospatial data formats using adequate extraction and recognition techniques that can handle the typically very large volumes of complex data and thus, requiring high degrees of automation. Whereas traditional approaches for information extraction from map documents typically involve a certain degree of user interaction, recently, a number of methods has been proposed aiming to overcome such shortcomings and to fully automate these information extraction tasks based on machine learning methods and the automated generation of training data, among others. In this chapter, we provide an overview of these recent trends, on existing, publicly available map archives, and the opportunities and challenges associated with these developments.