Separation of character strings and high quality vectorization for digitized Korean cadastral map images

When making an automatic map interpreter for a geographic information system (GIS) using printed maps, some difficulties lie in the interconnections between character strings and lines, and the distortions of vectorization at the line crossings and junctions. In this paper, the authors focus on the fact that the problems are only found at the line crossings and junctions, which can be extracted using the conventional thinning algorithm. They propose a new method to overcome the two difficulties at the same time. The proposed method is applied to Korean cadastral maps, and the experimental results show that the method is effective in separating the character strings from the lines and getting high quality vectorization for the digitized map images. At the same time the processing time can be reduced.