Abstract

Historical maps contain valuable information on settlements, land cover, and transportation infrastructure in the past. However, the information contained in such documents needs to be converted into analysis-ready data formats in order to be used for retrospective landscape assessments or to inform long-term land cover/land use models. The recent availability of historical topographic map archives as collections of georeferenced, digital raster data has catalysed the development of methods for information extraction from historical maps using advanced image processing or machinelearning based computer vision methods (Uhl & Duan 2021). Most of these studies, however, focus on relatively small datasets and thus, it remains unknown whether such methods are applicable and scalable for the extraction of historical geographic features across large (eg, country-level) study areas and massive data volumes. We …

Date

January 1, 2021

Authors

Johannes H Uhl, Stefan Leyk, Weiwei Duan, Zekun Li, Basel Shbita, Yao-Yi Chiang, Craig A Knoblock

Journal

Abstracts of the ICA

Volume

3

Total_citations

Cited by 1