Possibilities of vectorgraphic information extraction

OData support
Supervisor:
Lukovszki Csaba
Department of Telecommunications and Media Informatics

Today converting raster graphics into vector graphics is more and more important. To produce vector files manually is a long process, so its automatization is of great importance. I created a test application that can produce vectors from input images of inside pictures and floor plans of buildings. With small changes the programme could process other types of maps and images.

During preprocessing, I used and tested several digital filters. I detected the edges successfully on pictures of different quality and sizes and used a thinning based algorithm to localize the edges.

I completed the raw raster-vector conversion with line-tracking algorithm. Generated and displayed the edges, defined the nodes. Corrected the vectorization with local process, detected straight line segments, joined the edges and the nodes, deleted the unnecessary edges. I checked the algorithms efficiency on the test pictures. The vetcor-graphics picture is acceptable if there is not much difference between the input and output pictures with the last human constribution. The results of testing was satisfactory.

Downloads

Please sign in to download the files of this thesis.