Kodak i840 Document Scanner
The Association for Information and Image Management has a trade show and conference each year, with about 30,000 attendees and 300 exhibitors. The focus has evolved from microfilm to electronic images, storage (such as optical discs) to this year’s new hot topic, Enterprise Content Management. Don’t tell anyone that many of the vendors are the same, with new brochures and names for very similar products. The changing target and lack of consistent focus had both vendors and attendees confused, so the show was less exciting than in previous years.
Color scanners were common among most of the vendors this year. Some of the software vendors were offering tools to do image enhancement, bar code recognition, and form drop out from the full-color image, but Kodak was the only vendor concentrating on production of both color and b/w from the same scanner pass. (Normally the black and white image produced from a color scanner is inferior quality, which is why Kodak produces both together.) Bell and Howell may survive despite earlier predictions to the contrary.
|
Kodak i840 Document Scanner |
For the past several years there have been incremental improvements in the electronics of image scanners, but this year Kodak has taken the big step. A new line of high-end, duplex scanners was being demonstrated. The i810 and i820 scanners operate at 120 pages per minute; the i830 and i840 operate at 160 pages per minute, comparable to today’s top-of-the-line scanners. However, all support feeders with large stacks (thousands of pages without operator intervention - the input hopper in the picture is only half full) so the throughput per day should be very high. Further, both 820 and 840 are combination color and b/w scanners - producing high quality output in both modes simultaneously. The color can be used for OCR drop out and image enhancement, or the images can be stored in color. The best news is that the top of the line unit - fast, duplex, color and b/w, is expected to sell for $85,000 when it is available 4Q2001. And if you start at the low end, the upgrade can be field installed. The only thing it seemed to be missing was the imprinter often used to record the page and box numbers where the originals are stored.
The mid-range Kodak scanners were still available, and were even the scanners of choice for mixed and damaged documents. The mid-range Kodak 3500 and 4500 seem to be better at feeding and handling marginal documents than most slower units.
Bell+Howell sold most of their imaging business to Kodak, but the U.S. government would not let them sell the scanner business, since the sale would reduce competition. I expected that the remaining unit would be so small that B+H would just drop out of the scanner business. Instead I talked to Roland Simons, their chief engineer, and he introduced me to a rejuvenated management team that is excited about the future - it looks like they will continue to be competitive. There weren't many new B+H products this year (development was in limbo through the sale), but they had a color flatbed scanner with document feeder ($3,495). I expect to see good things in the coming year(s).
Everybody wants color. But what will it take to make color images practical for everyday use? If you are storing color photographs of P&C claims - water damage to a house, a wrecked car, hail damage to a roof - color is important today, and the storage required by a color photo is not excessive. If you want to drop-out the red or blue form to simplify an OCR process, color scanning is far simpler than the tedious reconfiguration of a black and white scanner. If you are faced with complex color documents, where the black and white rendering is hard to interpret, you may be willing to pay the premium for color. But if you have an ordinary letter with some yellow highlighting and red-ink notes, the cost of color in today’s technology is generally prohibitive. Storage is getting cheaper but 10-20 times as much cost to store a color document image is too high for most users.
A new technology is emerging - multilayer image compression. This technology will allow pictures that are part of a document to be compressed with one technology, while the text is compressed with another, and the highlights and annotations on the original document are compressed separately, etc. By using the technology that is most appropriate for each part of the document, a far better quality image can be stored, taking far less space than treating the entire document as a single color picture. Luratech has provided this technology for over a year, with a color document typically about the same size as a traditional black and white document. The newest JPEG standard will provide a similar layer-technology (but not compatible with Luratech). The primary authors on the standard committee are Lou Sharpe, President of Picture Elements Inc., and Luratech. Together they are "writing" the extension to the JPEG 2000 standard for multilayer JPEG color documents. This specification will probably be completed within the year (the file suffix will probably be .jpm).
Assuming that the standard is approved as expected, it will remove the barriers to color imaging, and will probably make color document imaging popular in the next 2-4 years. Therefore, someone investing in a new production scanner today, that should last for 5 years or more, may want to get one that has, or can be upgraded to, color.
The vendor that I have used for conversions in the past was bought by another company, which was bought by another, etc. I have not been impressed with the successor, so I have been in the market for a replacement film conversion vendor. Many conversion vendors were present, but none stood out.
In addition to the traditional vendors, I found the film scanning service bureau that many conversion companies subcontract their work to. According to the film scanner vendors, they do a superb job and handle very high volumes. According to the owner, they consider themselves a wholesaler - they do not want to get involved with project requirements, indexing (key data entry) or interface to the image systems (something the image system vendor or integrator would normally do anyway). Basic film scanning costs a few cents per page, plus minor overhead charges such as $6 to write a CD. (By comparison, one of our clients recently got a quote that was 10 times as high from a local service company).
 Wicks and Wilson 5100 Microfiche Scanner |
There are three dominant vendors of microfiche conversion equipment. Mekel (with an automatic feeder for more efficient operation, but barely acceptable image quality, Sunrise (no automatic feeder, but far better image quality), and a new $65,000 product from Wicks and Wilson Ltd. (a long-time vendor of roll film scanners from the UK). Fiche are manually loaded in a special carrier for the Wicks and Wilson scanner, but carriers with up to 20 fiche can be fed and scanned automatically.
Since fiche are indexed for each contract/policy/customer, but do not have any identification of individual documents, a bulk conversion is not practical. (Treating the entire fiche as a single 50-page document is easy, but will be a productivity disaster for years to come. Therefore special effort is required to identify individual documents - an entire subject in itself.) An on-demand conversion is normally the practical solution, which indirectly requires the user to have on-site microfiche scanners and custom programs to help identify the individual documents.
Roll film scanners are available from multiple vendors, but roll film always has an index, and is cheaper to do a bulk conversion (get all the images while the roll of film is mounted, and all the film has been processed together so is consistent density, etc.). Therefore bulk conversion is preferable, and a service bureau is generally in order.
If the roll-film conversion is to be done on site, there is one piece of software generally provided by service bureaus but not generally available. Each document is identified by a series of "blip" marks on the film. The reader-printer unit reads the blips as it rapidly searches the film, and finds the requested frame (image). In practice, as a user retrieves a document, the reader-printer has missed some blips, so the user must manually search a few adjacent images to find the desired document. When converting an entire roll of film, some blips may also have been missed, and thus the wrong images are identified. However, if we know 10,000 images are on a roll of film, but the conversion only found 9,997, the missing images can be located and that problem can be repaired with the appropriate software.
IImage Retrieval, Inc. in Dallas is one of the primary distributors of Sunrise equipment (the likely film equipment vendor for our customers). They have a programmer, Greg English, working on the problem. Greg expects to have completed the product, which will sell for $1-2,000 per repair station, in the next couple months.
Optical disc storage continues to be an important component (even though many systems are now keeping the working copy of all images on magnetic discs). Sony, who provides the drives for most vendors, recently announced a 9.1 GB 5.25 inch disc, with a migration path to 120 GB. The 12 inch storage continues to be exciting, with today's 30 GB expanding to 250 GB in the future.
Sony is the primary vendor of 5.25 inch optical disc drives, providing the drives to Plasmon, HP, and IBM. They have recently announced the next generation of magneto-optical drives with 9.1 gigabytes per unit, up from 5.2 gigabytes currently. The 5.2 GB drive only records in the grooves; the 9.1 GB drive also records on the land between the grooves. (The new disc capacity leads to jukeboxes with up to 2.2 Terabytes capacity). A 40 GB drive has also been developed, using a new "violet" laser technology, that will be available in late 2002. They expect it will be followed by 80 GB and 120 GB drives.
HP, IBM, and Plasmon all provide a simulated non-rewriteable technology by recording a flag in the disk header that marks it as write once. Since the SEC requires that records be stored on non-rewriteable media for equity based products, there is some question whether this technology will be sufficient. (To my knowledge, it has never been challenged in court, and many companies are comfortable with the risk, but other companies are concerned.)
Sony has a true WORM drive, using ablative technology (the laser blasts a permanent pit in the disc), with 5.2 Gigabytes (but not 9.1 gigabytes) capacity. However, Sony only offered that drive through IBM (in their 3995 jukebox line). Therefore, IBM’s claim that they are the only vendor with 5.25 inch WORM discs is correct. Since the true WORM 5.25-inch technology has not advanced to 9.1 GB, and since the market was small, there is real question if the 5.25 inch WORM technology will be continued.
The Plasmon drives formerly from LMSI currently support 30 gigabytes per platter. They use phase change technology - not ablative, but unquestionably a write once, non-rewriteable technology that meets SEC requirements. The product plan for these drives continues through 60, 120, and 240-Gigabyte capacities over the next few years.
There are several advantages to the 12-inch technology that are often overlooked. The 12-inch drives can read and write both sides of the media at the same time, so the media never has to be flipped over (like all 5.25-inch discs). Flipping a disc is the slowest operation that a jukebox must perform, and causes significant degradation of overall optical disc system performance. The peak data transfer rate of a 5.25-inch disc is comparable to the rated data transfer rate of a 12-inch disc. However, the 12-inch disc maintains that rate over the entire surface, while the 5.25-inch disc drops to roughly half as fast at some positions on the disc. Bottom line: there are some compelling legal and performance reasons to consider 12 inch optical discs rather than 5.25 inch.
Some users of 12-inch optical discs and jukeboxes have been put off by the apparently cranky nature of the systems - running perfectly for months, interspersed with days or weeks of troubles. Plasmon has recognized the problem, and suspects that the problem may be related to maintenance vendors who are are less than perfectly motivated, or (dare I say) might even prefer a switch to another product. Therefore when a problem is identified, they are taking special action, even bringing the maintenance responsibility in-house if necessary. Contact John Drollinger, Director of Large Format Optical Products, john.drollinger@plasmon.lms.com, 719-593-4077 if a customer has persistent problems with the LMSI drives and Cygnet jukeboxes (that are now provided by Plasmon).
This does not attempt to be a comprehensive view of all the image toolkits nor all the image system vendors, integrators, or consultants at the AIIM show. I was specifically looking for ways to view existing multipage TIFF images (Group 4 blank and white documents and JPEG color pictures) through a thin client.
Most browsers only support gif and jpeg image compression, and only support single page documents. Neither of these formats is optimal for document image compression. Some vendors still convert multipage documents to a series of JPEG files, and send them to a browser one page at a time. The alternative is to automatically download an applet from the server, to provide support for multipage documents with optimal compression (as stored), color documents, and additional functions. Several vendors provide applets or toolkits for developing thin client viewers.
Snowbound had a viewer applet but their prices are extraordinarily high. They will customize their thin client viewer for $24,000 per server CPU, plus $12,000 per server CPU per year after the first year. If you don’t need customization, or want to do your own customization, the price is still almost as high. Their native image toolkits start at $2495, but quickly rise to $3495 or more, plus $4600 for modules to handle specialized compressions such as PDF or ABIC (check images), and still have a run-time license.
Pixel Translations has a simpler set of tools that they license either per user or per server, but the price is only $4,000 per server. They will provide a trial version if desired. I have worked with their other toolkits.
TMS has a downloadable plug-in (2.1 megabytes) that is automatically installed in a browser, and performs exceptionally fast and flexibly. However, it is licensed per user, and only works with some browsers. A Java applet is under development, and sounds like it will be excellent when it is available in 6 months or so. The applet will only download the pieces that are required (far less than 2.1 MB), and will store the applets on each workstation as space allows. The applet will be licensed on a per-server basis.
Accusoft has the NetVue software to go with their ImageGear toolkit. NetVue is available as a 200K java applet, or as a faster 900K ActiveX control (that must be installed). Images are converted to a proprietary format to be downloaded. NetVue is licensed on a number of concurrent users basis. I have a demonstration CD, or see their web site at www.accusoft.com
I finally got the "straight story" about this very unusual vendor.
Start with Wang imaging, which really had two image systems - the original solid system that ran on their proprietary Wang hardware, and a later version (not as popular) that ran on Intel hardware. It had a toolkit that was pretty good for building complex workflows.
Sigma imaging was developed for Empire BC/BS (New York), and was bought by Wang in 1995. It’s strength was key from image and high volume simple workflow transactions (health claims).
Kodak bought Wang imaging (with Sigma) when Wang was dying. However, their scanner customers threatened to find another source of scanners, rather than buying from a competitor. Therefore Kodak quickly formed a separate company, Eastman Software, to own the Wang products.
An investment firm, eiStream, bought the stagnant Eastman Software in 2000, and is doing business under the Eastman Software name. Eastman Software has built a modern web-based client (based largely on the Sigma client) for the future.
eiStream then bought ViewStar (briefly known as Mosaix) from Lucent. The new Eastman Software client will also retrieve images from the ViewStar archive, reducing the need for conversions.
eiStream then bought Kofile, a low end image system. This leaves Eastman Software as an image system vendor with an extraordinarily diverse set of products.
Back to the home page
Send e-mail comments to
©2001 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 www.plesums.com.