Setting a New Standard

by Doug Henschen

"Standards are great,"one joke has it, "because there are so many to choose from."

While the world of technology can seem chaotic, you can usually boil standards down into two groups. First, there are the "true" standards backed by internationally recognized organizations. Second, there are de facto standards, which often start out with support from a single company, yet manage to gain broad acceptance.

To get a sense of standards in document imaging, you need look no further than this month's Services Directory (page 53). Of the 54 companies listed, every last one supports TIFF. While TIFF was originally developed by Aldus, it is most commonly used in combination with Group III or Group IV compression, standards sanctioned by the Comite Consultatif Internationale de Telegraphique et Telephonique (CCITT) - one of those impartial bodies that few question.

A close second in popularity is PDF, a de facto standard originally developed by Adobe. The format now enjoys such broad support it can hardly be described as proprietary, yet there are foibles related to its commercial roots.

There are now dozens of third-party PDF document imaging tools, but they vary widely in cost, performance and image quality. Everyone's output will pass the minimal test of opening up in Acrobat Reader, but variability undermines the idea of a standard.

The next-most-popular format in our directory is JPEG, devised by the Joint Photographic Experts Group. This standard is just the thing for compressing color images, but has its drawbacks. JPEGs are very large with minimal compression, and very lossy and crude in appearance with high compression.

JPEG's answer for next-generation imaging is JPEG 2000, which uses wavelet compression to encode graphical images accurately yet efficiently. Now there's a proposal afoot to create a document-oriented version of JPEG 2000 dubbed "JPM."

JPM would add image segmentation to JPEG 2000 so that hard-edged areas of an image (such as text) could be retained losslessly on one layer. The remaining areas would be encoded on a separate layer with wavelet compression.

If you'd like a preview of what the combination of segmentation and wavelet compression can do, read "A Closer Look at Compact Image Formats" (page 63). Our lab-based review pits TIFF Group IV, JPEG and PDF against DjVu and LuraDocument. The latter two are proprietary wavelet/segmentation formats from Seattle-based LizardTech and Menlo Park, CA-based LuraTech, respectively.

The results of our test were impressive, as DjVu and LuraDocument both delivered dramatically smaller files - some more than 98 percent smaller than the current standards (although not without compromises).

Color imaging and the Internet are the key drivers of interest in this technology. It remains to be seen whether the proprietary formats will become de facto standards. Many, including this editor, would love to see adoption of JPM within the year. The next milestone would be browser support from Microsoft and Netscape. As it is, it might take years for new browsers and plug-ins to make it into the field. But without that last ingredient - particularly support from Microsoft - any "standard" would be little better than a proprietary alternative.