Document image segmentation and text area ordering

摘要

A system for document image segmentation and ordering text areasis described and applied to both Japanese and English complex printedpage layouts. There is no need to make any assumption about the shape ofblocks, hence the segmentation technique can handle not only skewedimages without skew-correction but also documents where column are notrectangular. In this technique, on the bottom-up strategy, the connectedcomponents are extracted from the reduced image, and classifiedaccording to their local information. The connected components aremerged into lines, and lines are merged into areas. Extracted text areasare classified as body, caption, header, and footer. A tree graph of thelayout of body texts is made, and we get the order of texts by preordertraversal on the graph. The authors introduce the influence range ofeach node, a procedure for the title part, and extraction of the whitehorizontal separator. Making it possible to get good results on variousdocuments. The total system is fast and compact