Automated georeference of the 1:20 000 Romanian maps under Lambert-Cholesky (1916–1959) projection system
2010
Old cartographic documents represent a valuable resource in reconstructing the natural and built environment evolution. In order to integrate such maps with recent geospatial datasets, in GIS environment, some preparatory actions need to be done. This includes scanning, color enhancement, georeferencing and reprojection. From our experience, when it comes to collections with a large number of map sheets, georeferencing is the most meticulous and time consuming process. Traditionally, this is done manually, by carefully selecting points with known coordinates. To overcome this disadvantage and to increase the rectification precision an automated procedure was created. The whole process of map sheet georeferencing is done by a specially developed tool, called “findlines”, which relays on radon transform to extract, even in degraded and noisy conditions, all the straight lines from the map graticule network. Then, by knowing the distance between the graticule lines and the relation between the map name and lower-left corner coordinates, it is possible to automatically associate real coordinates to each intersection points in the graticule. Finally, the points are used along with GDAL to rapidly georeference each map sheet. The method was first successfully tested with modern topographic maps. To evaluate the procedure with old cartographic documents we select the 1:20 000 Romanian maps collection, under Lambert-Cholesky (1916–1959) projection system, as test data. The basic map, called “Plan Director de Tragere” was drafted under 1:20 000 scale in 2 118 drawings, covering all the Romanian territory. Each map sheet is 75 cm wide and 50 cm high and has a direct relation between the name and the lower-left corner coordinates. The maps were also successfully georeferenced using the findlines utility. Depending on the selected resample method, the time needed to georeference each map sheet, using 150 control points, ranged between 3 and 15 minutes on a normal computer. That proved the method efficiency and flexibility. Any map series with a regulated graticule grid and a logical relation between the name and the spatial position can be georeferenced using this approach, without any human intervention.
Keywords:
- Correction
- Source
- Cite
- Save
- Machine Reading By IdeaReader
7
References
4
Citations
NaN
KQI