The Impact of Graphical Quality on Automatic Text Recognition in Digital Maps
ISBN 978-85-88783-11-9
Authors
1Chiang, Y.; 2Leyk, S.; 3Honarvar Nazari, N.; 4Moghaddam, S.
1SPATIAL SCIENCES INSTITUTE,UNIVERSITY OF SOUTHERN CALIFORNIA Email: yaoyic@usc.edu
2DEPARTMENT OF GEOGRAPHY,UNIVERSITY OF COLORADO, BOULDER Email: stefan.leyk@colorado.edu
3SPATIAL SCIENCES INSTITUTE,UNIVERSITY OF SOUTHERN CALIFORNIA Email: nhonarva@usc.edu
4SPATIAL SCIENCES INSTITUTE,UNIVERSITY OF SOUTHERN CALIFORNIA Email: khashkha@usc.edu
Abstract
Converting geographic information (e.g., place names, contour lines) in scanned maps into vector format is usually the first step for incorporating map content into a geographic information system (GIS). With the advancement in computational power and algorithm design, map processing systems have been considerably improved during the last decade. However, the fundamental techniques that support automatic map processing such as color segmentation, layer separation, and object recognition are sensitive to minor changes in the input image (e.g., different scanning parameters such as resolution). As a result, most map processing systems would fail if the user does not "properly" conduct a digital scan of the map of interest and train a map processing system. Not only this can create a discouraged user community but also slows down further advancement of map processing techniques as less sophisticated tools would be perceived as more viable solutions. Thus what kinds of maps are suitable for automatic (or semi-automatic) map processing and what types of results can be expected are critical points. In this paper, we attempt to shed light on these questions by using a typical map processing task, text recognition, to dis-cuss a number of map instances that vary in the degree of suitability for map processing. We also describe a text recognition experiment on Ordnance Survey historical map scans to provide measures of baseline performance of standard recognition tools under varying map conditions. This outcome helps the user to understand what to expect when a fully or semi-automatic map processing system is used to process a scanned map image with certain properties of graphics and map content. This discussion could be further extended to other processing techniques such as line detection or symbol recognition in scanned maps.