Using Page Breaks for Book Structuring

Hervé Déjean

Abstract: We report on the XRCE participation to the Structure Extraction task of the INEX/ICDAR Book Structure Extraction 2011. We wanted to assess a simple method for structuring a book: using leading and trailing page whitespace. The detection of such large whitespace occurring at the top of leading pages and at the bottom of trailing pages is based on the detection of the type area zone. Evaluation shows as expected a very good precision. Since this approach aims at detecting high level book structures (parts, chapters), structures not marked a page break are not detected (thus a lower recall).