Detection of numbered captions

A system for detection of numbered captions comprising:

• a graphical object detector for identifying illustrations on pages of a document comprising a sequence of document pages;
• a related text detector for identifying associated text for each identified illustration;
• an imitation page generator for generating an imitation page for each of the identified illustrations, each imitation page comprising a single illustration and its associated text;
• a sequence detector which, for a sequence of the imitation pages identifies a sequence of terms, each term derived from a text fragment of the associate text of a respective imitation page, the terms of a sequence complying with at least one predefined numbering scheme which defines a form and an incremental state of the terms in a sequence;
• and
• a numbered captions identifier for construing the terms of the identified sequence of terms as numbered captions for respective illustrations in the document;
• and
• a processor for implementing the graphical object detector, related text detector, imitation page generator, sequence detector, and numbered captions identifier.

A method of detection captions in a document comprising:

• receiving a document comprising a sequence of document pages, the document pages comprising graphical elements and text elements;
• identifying illustrations on pages of the document based on the graphical elements, comprising segmenting the page into illustration zones without considering any of the text elements on the pages, each zone comprising a respective illustration;
• for each identified illustration, identifying associated text from the text elements;
• generating an imitation page for each of the identified illustrations, each imitation page comprising a single illustration and its associated text;
• for a sequence of the imitation pages, with a processor, identifying a sequence of terms, each term derived from a text fragment of the associate text of a respective imitation page, the terms of a sequence complying with at least one predefined numbering scheme applicable to page numbers of a document;
• and
• identifying numbered captions for respective illustrations in the document based on the terms of the identified sequence of terms.