4. Final discussion

Structured documents play an important role in electronic publishing, and their value is well established. While the structured document approach lends itself well to the creation of new documents, built with the help of specialized editors, it also creates new difficulties when it comes to retrieving and reusing existing documents, whether in electronic or paper form.

Originally, document recognition was mainly applied to old paper documents, for which no electronic form existed, or at least was not accessible. The document image was captured using a camera or scanner and then analyzed. Today, we realize that some of these techniques can also be applied to document images obtained synthetically from existing, poorly structured electronic documents.

Despite the enormous efforts made to advance structural document recognition, it has to be...