CHINESE DOCUMENT LAYOUT ANALYSIS BASED ON ADAPTIVE SPLIT-AND-MERGE AND QUALITATIVE SPATIAL REASONING

The ultimate goal of automatic document processing is to understand th e semantics of a document. Towards such an end, one of the primary ena bling steps has been to first reason about the layout of the document by means of page segmentation and segment spatial reasoning or labelin g. This, in turn, allows for the derivation of document logical organi zation. This paper describes a generic document segmentation and geome tric relation labeling method with applications to Chinese document an alysis. Unlike the previous document segmentation methods where text s pacing, border lines, and/or a priori layout models based on template matching processing are performed, the present method begins with a hi erarchy of partitioned image layers where inhomogeneous higher-level r egions are recursively partitioned into lower-level rectangular subreg ions and at the same time lower-level smaller homogeneous regions are merged into larger homogeneous regions. Furthermore, the derived segme nt data structure readily enables efficient search for geometric relat ionships between identified document segments. (C) 1997 pattern Recogn ition Society. Published by Elsevier Science Ltd.