عنوان مقاله [English]
In any type of document, whether book, magazine, dissertation or likes, the table of content expresses concisely its logical structure. By using table of contents, the document structure is easily reviewed and the desired topic could be readily accessed. The present paper presents for the first time a method for automated recognition of the table of content in dissertations written in Farsi, Arabic and any other western script. In this method the content pages are recognized given their patterns without employing OCR and merely using image processing techniques. The method can recognize the table of content pages regardless of the language or text justifications. Since it does not use OCR, it is independent of the document scan quality. The method was tested over a number of IRANDOC Farsi, Arabic and Western Dissertations. Recognition accuracy of 99.7 percent was achieved.