Imaging Standards

© 2000 by Charles A. Plesums, Austin, Texas

Standards Programs

The Association for Information and Image Management International (AIIM International) is a professional and trade organization for Document and Image management including workflow, and has recently expanded into e-commerce. In addition to the well known annual AIIM show and conference, they are active in the development of standards. In the United States, they are the "Standards Developing Organization" designated by the American National Standards Institute (ANSI) responsible for microfilm and electronic imaging standards. As designated by ANSI, they determine the US position on international standards in these areas.

In addition to the domestic standards program, AIIM and AIIM members are active participants on many of the international standards organizations and committees.

Formal standards are developed by consensus, with requirements for widespread publicity and consideration of every opinion. Approvals go through many layers of votes and publication. Development of an official standard normally takes several years. In the United States compliance with approved standards is normally voluntary, so sometimes an official standard is less relevant than a de facto industry standard. For example, when the ASCII standard was approved for data on magnetic tape, IBM already dominated the market, so most tape (probably 80-90% or more) was written in IBM's EBCDIC form rather than ASCII.

Sometimes the time required to create a formal standard allows a non-standard technique to achieve industry popularity. Industry consortia, or the dominant vendor, sometimes develop these de facto standards that meet the test of use, even though they may not have international endorsement and concensus. Specifications that have all the required features, are well defined, and are published and freely available to anyone, often succeed.

Imaging Standards

Image processing standards deal with many different aspects of the image system. Some are primarily of concern to the manufacturers, such as the information that must be included in product specifications or the electrical interfaces between devices. Others deal with the data, and are of more interest to the user.

Black and White (Document) Images

Most office documents can be reduced to pure black and pure white (or dark and light) without gray or color. These are called bi-level images, and are similar to the microfilm, fax, and copier technologies that have been used for years.

When an image is initially scanned, there are many shades of dark and light, based on imperfect lighting in the scanner and flaws in the paper. Early in the process, each pixel (picture element, or dot of ink) is converted to a pure white or pure black. This information must then be compressed. The simplest form (and the underlying technology of the more sophisticated techniques) consists of counting the number of white pixels, then counting the number of black, the number of white, black, white, etc., then encoding the series of numbers in the most efficient way.

A facsimile (fax) document is an image; the compression technology used by fax has been widely used and optimized for many years. The common Group 3 fax uses either the T.4 or the T.6 compression, defined as standards by the International Telecommunications Union (ITU, formerly CCITT). The rare Group 4 fax does not support T.4, but does use standard T.6 compression. Therefore the T.6 image compression is often referred to, incorrectly, as Group 4 compression. T.6 compression (Group 4) is widely used for bi-level (black and white) document images.

Another compression technique for black and white documents is JBIG, approved in March 1993 as ISO standard IS 11544, and adopted by the ITU as telecommunications standard T.82. Rather than counting the number of black and white spaces, or the differences between the current and previous space, it compares the current space to another reference position, typically slightly above and left. By making the reference space a nearby dot in a half tone picture (like a newspaper picture), JBIG can efficiently compress half-tone photographs, which are very inefficient with T.4 or T.6 compression. Overall JBIG was computationally more intense than T.6, so has not been widely adopted.

Image file formats

The Group 3 or 4 fax standard includes the exchange of information necessary to interpret the images in the hand-shake at the beginning of the telephone call. Therefore when the T.6 compression is used outside of a fax, that extra information must be recorded elsewhere. Likewise, as information is stored in a computer, rather than just being sent along a telephone wire, it needs to be broken into blocks for input/output and into files for storage. Therefore another standard is required to define how the compressed image will be stored - a file format.

Many image formats only support a single page, not a multi-page document. Some image systems support each page as a separate file, but others combine the several pages of a document into a single file, with a single reference to the entire document from the index database.

One of the early standards that was well defined, met the requirements, was published and freely available was IBM's MO:DCA (Mixed Object: Document Content Architecture). Each page in the defined "mixed object" could be many things, one of which was a scanned image, IOCA, for "Image Object Content Architecture." The IOCA standard defined a page with all the necessary header information, a code to define the type of compression (including T.6), and the compressed image. The several pages of a document, each in IOCA format, were combined into a MO:DCA document. This was an excellent format except for one small problem - the manual was possibly the worst IBM ever published - nobody could understand it. Despite it's technical virtues, MO:DCA never caught on.

Version 6 of the Aldus TIFF image file specification, published in 1992 (and now owned by Adobe), included support for multipage documents, along with many other options. TIFF stands for Tagged Image File Format, and each page has multiple tags (typically 10 to 25 tags), each specifying a different option, value, or set of values. The specification was concise and understandable, so programmers quickly adopted the format, and it has become a de facto standard. If anything, the hundreds of TIFF options have been a weakness, since there are so many features in TIFF that not every TIFF reader can interpret every TIFF file. To be assured of compatibility, it is wise to specify the options that a TIFF file supports.

Photographic Images

Fax compression is based on counting the number of white pixels, then the number of black pixels, then the number of white, etc., and encoding those counts. Therefore the T.4 or T.6 compression only makes sense for documents that can be represented as pure black and pure white. Some scanners have the option to convert gray into a series of black and white dots, but this is terribly inefficient during image compression. Therefore if a "gray" document, like a "black and white" picture, or a color document must be stored, it is better to switch to a compression that is designed for "continuous tone" images.

The Joint Photographic Experts Group (JPEG) has defined a compression toolkit that has become ISO standard IS 10918-1. It was adopted for telecommunications use (color fax) as standard T.81. As a toolkit there are many options. A mathematically lossless compression is defined, but the compression ratio is often only 2 or 3 to 1. A mathematically lossy compression, with so little loss that it is indistinguishable to even expert viewers, often gives a compression of 30:1 or more. Compression may be sequential (all the detail is presented for the beginning of the picture before continuing to the rest of the picture - ideal for printers), or progressive (the entire picture appears quickly, but the quality improves progressively). The codes to define which options have been used are included in the compressed image. The JPEG standard defines a subset of functions that must be implemented in every decompression routine; any image that only uses the subset should be usable through every JPEG program. A single image compressed with JPEG can be stored in a JPEG file (defined as part of the standard), or JPEG can be used with either MO:DCA or TIFF file formats.

The fractal technology for compression has received a lot of publicity because it appears to have a much higher degree of compression than JPEG. The real question for both is what happens to the image if "too much" compression is specified. JPEG produces an image that shows unsightly square blocks, but on close examination most of the detail is still present. Fractals mathematically seek areas of the picture that are similar. The scar on one cheek of a person may appear on both cheeks and the forehead, or may disappear altogether. The right side of the car with the dent may suddenly look like the left side without the dent. Fractals create a very attractive picture as they degrade, but may lose the detail that is required. As one expert said, "If I want a picture of a pretty tree, I need fractals, but if I want a picture of MY tree, I need JPEG."

Back to the home page

Send e-mail comments to

©2000 by Charles A. Plesums, Austin, Texas USA. ALL RIGHTS RESERVED. You may license additional copies of this document through a nominal royalty payment as specified on