Generation Of A Line Description File For Graphics Recognition
Ching-chuan Shih, Rangachar Kasturi · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 1988
A typical office document or an engineering drawing consists of regions of text, graphics, and halftone images. Developing algorithms for automating the input of such documents is the goal of this research. The scope of this paper is the design of algorithms to process raster-oriented binary images of paper-based graphics to obtain vector-oriented description files. The image is preprocessed to suppress noise and other digitization artifacts. The graphical components are then represented as a union of maximal squares using the maximal square moving algorithm. The pointers describing the connectivity of the maximal squares are analyzed to generate a linked list of squares. Straight line segments, curves, junctions, and large areas are then identified after extensive processing of this linked list. The output of the algorithm is a graphics description file which is then used as input to a graphics recognition system.