The JBIG algorithm for loss-less compression of bi-level images, especially scanned documents, has been implemented, optimized and tested as a portable C library. This implementation can now easily be included into image and document handling systems. Specialized data compression methods for scanned text and document images are going to play an increasingly important role in the future for archiving scientific publications and keeping them available for library users. Several widely used basic mechanisms for data compression with special emphasis on those used in the JBIG algorithm are described as well as details of this JBIG implementation which is especially designed for the needs of interactive applications. The performance of JBIG on a number of sample documents is evaluated demonstrating that compression ratios between around 4 and 60 depending on the type, image quality and character density of the document are achieved. Typical compression ratios obtained on scanned scientific publications are in the range 8 to 20. With the presented implementation, a fast personal computer can encode and decode at data rates over 64 kbit/s as required by real-time compression for fax transmission on ISDN channels.

Kuhn, M. Effiziente Kompression von bi-level Bilddaten durch kontextsensitive arithmetische Codierung. Studienarbeit, Lehrstuhl für Betriebssysteme, IMMD IV, Universität Erlangen-Nürnberg, Erlangen, July 1995.

The full text (German, 62 pages) is also available on-line.