Mathematical Morphology : A Tool for Automated GIs Data ~ cquisition from Scanned Thematic Maps

Over several years, the demand for digital data has grown as geographical information systems were im- plemented. Because of the cost of database creation, data acquisition should unquestionably be automated as much as possible. Today, scanning devices produce a huge amount of data that still needs complex processing. This paper proposes an original approach using image processing tools coming from Mathematical Morphology theory to acquire GIs data from scanned thematic maps. These tools are used in order to obtain a segmentation prior to radiometric analysis. To illustrate the methodology, a subset of the Belgian soil map is treated. URING THE LAST FIFTEEN YEARS, dramatic growth in the use of geographical information systems (GIs) occurred in many disciplines. This technology provides multipurpose land infor- mation systems as a means of dealing with the land records modernization problem. Natural resources modeling and man- agement also benefit greatly from this technology by integrating a tremendous amount of data. By far, data capture is the most expensive task in a GIs ap- plication. To make full use of GIs capabilities, accurate and up- to-date information must be provided. The usual method of acquiring data from existing maps is by manually following lines on a digitizing table. Though advertising brochures usually show an operator happily at work, this method is actually a very tedious and boring operation with a high probability of errors due to either duplicating or omitting information (Doyle, 1978). To overcome this predicament, techniques for automated and semi-automated data capture have been developed during the past several years. Two major "automated processes being used today for data acquisition are line following and map scanning. Line following techniques are used for digitizing lines such as stream networks or contour lines while map scanning allows the extraction of point, line, and surface information. A recent opinion of map scanning (Faust, 1987) stated that "the existing multistep-inter- active procedure for creating point, line, and polygon structures with the appropriate attributes must be automated to a Kreater extent." 1i ibis paper, we present a method to automaLe data acquisition with the Belgian soil map which is of interest for natural resources oriented GIS as an example. The major feature of the method is its ability to segment the map into its different areas prior to radiometric analysis. This process makes use of Mathematical Morphology, a new image processing technique. The results indicate that the method is consistent with human interpretation of the map. However, it is not the claim of the authors to suggest that a methodology 'could be applied without distinction to any kind of thematic maps, because each has its own peculiarity. Nonetheless, our opinion is that some tools used here could be widely employed for processing other scanned maps.