A region-based vectorization method

Abstract

Lines and junctions are principal features for a line image. The spatial relationships established among them are usually employed in applications such as OCR and architectural blueprint vectorization. The conventional thinning techniques often suffered the pitfall of spurious junction points that are crucial features to derive. The liability of the shape of the skeleton resulting from multiple fork points connected through several short branches will impede the further recognition stage. Also, different thinning algorithms often lead to different results and are very sensitive to the presence of noise. Since morphological thinning operation proceeds in a step-by-step peeling of outer-layer pixels down to a unit-pixel skeleton, the computation requirements are high and very time-consuming. An algorithm which differentiates from the conventional thinning algorithms in vectorizing a raster line image called maximal inscribing circle (MIC) is used. This region-based vectoriation algorithm takes the ensemble of pixels within the line segments collectively as legitimate candidates in deciding the vectoriaed representation. An iterative procedure is outlined and the criteria for merging short straight line segments into the corresponding curve representation are described. This method can not only segment the lines and junctions, construct their spatial relationships in a computation efficient manner, but also retain their line width. Experimental studies comparing the MIC algorithm with a conventional thinning algorithm are performed using scanned raster diagrams. Engineering blueprints with various sizes and complexities are uses as test raster images to demonstrate the versatility of our method. The results show promising performance in comparison with that of the traditional approaches. Finally, conclusions are made and future works are discussed.