Adaptive Vectorization of Line Drawing Images

second, they capture the gross structure of the lines in the original image. They describe the end points, corners, A novel method for vectorizing line drawing images is presented. The method is based on a sequence of a standard vectorijunctions, and other special points in the image, i.e. the zation algorithm and maximum threshold morphology, which obviously important (dominant) points of a vectorization. can be iterated until a fitting criterion is met. First, ‘‘anchor The algorithm has been developed in the context of a points’’ are found in a coarse vectorization. The position of map interpretation system. Therefore, examples are illuseach anchor point is corrected by morphological operations, trated with this kind of line drawing. Of course, the algoafter which it is fixed. Next, the vectorization is refined between rithm can be applied to other types of line drawings such the anchor points. The method is adaptive because the original as CAD/CAM drawings, schematic diagrams, and lines input image is used for correcting and improving the vectorizaextracted from an image with an edge detector. Indeed, tion. Postprocessing modules can be added to anticipate specific one of the experiments is with an electrical diagram. There properties of the line drawings to be vectorized. The vectorizais no significant difference between the performance of the tion method is evaluated using line drawings differing in scale and resolution.  1997 Academic Press algorithm on maps or on diagrams. Between the different modules in the algorithm there are possibilities to anticipate specific properties of the line drawings.