In this paper we propose a general method for contour recognition and DEM (Digital Elevation Models) generation from commercially available printed topographic maps. Automatic contour recognition of a scanned topographic map is a difficult problem due to the presence of closely spaced or broken contours, overlapping data layers and complexities arising from textured background etc. Beginning with a scanned map our approach utilizes a multistage process which broadly includes contour identification, cleaning and vectorization. While the identification and vectorization steps are based on calibrating existing image processing techniques, the cleaning step uses a modified geodesic distance based approach. The proposed technique was tested on contour regions of varying complexity scanned from a Survey of India toposheet of scale 1:50,000, scanned at 300 DPI. It was found that the proposed technique resulted in clean segmentation, crisp contour borders in the case of regular contour regions. However, differing levels of manual interventions were required to vectorize complex contour regions represented in the printed map. © 2011 Elsevier B.V. All rights reserved.