A method of document image structure analysis of books index information

摘要

Analysis of the document image structure of a book by making use of theindex information enables us to grasp its logical structure. When developing an application software whose purpose is information search, index information in a book should include its optimal keywords. The page numbers in an index should be optimally used by automatically linking them to their respective keywords, in order to facilitate improved information search. Employing this method for hundreds of books would be useful for building a large-scale hypertext database such as an electronic library. This paper deals with the method of analyzing index information available as keywords for information search when converting the whole text of a book into hypertext. Our experiment with ten different kinds of index information succeeded in analyzing 98% of all the information, except for those whose characters were unrecognizable because of typefaces of numbers.