Page decomposition and signature finding via shape classification and geometric layout
John D. Hobby · 1999
Consider the problem of decomposing a page image into text ruling lines, signatures, other line art, and other material. A fast classifier based on a skeletonization of the image and various curve-fitting techniques gives an initial labeling, followed by Baird's language-free layout analysis and a post-processor that uses the geometric layout to refine the decisions about text versus non-text.