2-D modeling and Hough transform for document analysis

The authors study how discrimination between text and graphics can be made in images binarized or not. The method used assumes that a block composed of text has privileged directions. In this case, the discrimination becomes a detection of lines by the Hough transform. The algorithm used is relatively independent of changes in text font, style, and size, as well as of string orientation. The images that are recognized either are nonbinarized, issue from a scanner, or issue from a charge-coupled device (CCD) camera