Paper scanners are critical to document imaging as they form the primary method of converting paper to a digitized representation. The quality of scanned documents can make or break an image management system. However, potential users often just accept the scanner that the selected vendor offers and do not include it as part of the vendor selection criteria. We believe that it's essential to evaluate the scanner you intend to use as an integral part of the overall imaging system prior to purchase.

High speed scanners will convert a letter sized document to digital representation in approximately 1-3 seconds at 200 dpi depending on the model of the scanner. As these scanners assume a level of centralized scanning, a user should consider purchasing when moving from a small individual filing system to a more centralized document imaging system. Typically a high speed scanner can be justified when more than 500 pages a day are to be scanned, but large installations will scan 10,000 or more a day.

The term "Scanner" is used by many different industries to define their product. Even when the term is used for document scanner, it can mean different things: image scanner or OCR scanner, flat bed scanners or high speed scanners with a paper transport, hand-held scanners, film scanners etc..

Most document scanners are however, composed of a few similar components: camera, image processor, illumination, paper transport, interface to host, possibly an automatic feeder and a stacker and sometimes certain additional processes such as compression and Optical Character Recognition. The objective of this paper is to describe these components with emphasis on high speed document scanning.

The Camera:

A sensor analyses the light reflection of a single spot on the document to determine whether it is black or white and sets a switch on or off. A linear CCD is a line of sensors looking simultaneously at a line on the document, converting the signal to series of on and off bits.

To create a digital image of a document, one must convert all black and white dots on a page to on and off bits in the computer memory. The document is moved in front of the camera or in the case of flat-bed scanner, the camera is moved in front of the page. This array is built into a page memory representing the digital image.

Image Resolution:

Horizontal image resolution is determined by the number of "sensors" sampling the line. Vertical resolution is controlled by how fast the paper moves past the CCD. For example, if the linear CCD has 2000 elements, it will convert a line of 20 cm long to an image with a resolution of 10 points per mm. In reality, sensors have 1728, 2048, or 4096 pixels depending on the scanner and will drop the extra pixels when less than the maximum resolution is required. A 4K array (4096 pixels) allows the scanning of a page at 400 dots per inch or 16 points per mm, but also by dropping half of its pixels at 200 DPI or 8ppm. Most scanners offer 8, 10 or 12ppm and Japanese vendors usually also offer 16ppm.

Since resolution affects the image size in both directions (8x8, 12x12), a 8ppm raw image of an A4 page will be almost 0.5 MB and a 12ppm raw image will be almost 1.0 MB. In principle, the more points per mm, the better. When scanning images for viewing only, 8ppm will do in most cases. On the other hand, for OCR applications, scanning at 12ppm is recommended, especially if the text is smaller than 12 points.

Image Enhancement and Gray Scale:

Image cameras are providing more than bitonal (black or white) results. Depending on the digitizer they use, they can convert the CCD's output to a certain level of gray. To describe a level of gray requires more than one bit (2 possible values). A 16 level of gray requires 4 bits and 256 levels requires a full byte. A simplified example would be, a point in the image could be completely black, almost black, dark gray, gray, light gray, almost white or white. The dot has only one value out of a possible 7 choices, but it would take 3 bits to represent its value.

Grayscaled images thus take up a lot more memory and storage space than black and white. Most imaging systems are not directly using grayscaled images. However, the larger number of choices provided allows more processing choices. For example, a pencil mark scanned in black and white may not show up dark enough to be converted to a black pixel, but scanned in 16 shades of gray, it may be dark enough to appear in the image. Filters, image enhancement and thresholding programs then convert the image to bitonal, allowing better separation between background and actual data, light data and noise etc.. Image enhancement can sometimes have adverse effect on certain applications, for example when attempting edge enhancement (to better define actual data from background), the algorithm may tend to fill gaps between bars of a bar-code and make it impossible for the bar-code to be read. Some, but few applications require true gray scale. For publishing purposes it may be needed, even though few publishing software actually use gray scale. Most programs use pseudo-gray or half toned images which represent gray level by using several dots.

Illumination:

For a CCD to see a dot on a page, this page must be illuminated with a bright light source. It is the reflection of that light which is captured and analyzed by the CCD. When paper is too thin, the light may go through the paper and reflect on a white background, creating character bleed-through. Black background scanners do not have this problem. However, a black background creates a black boarder in the image at the papers edge. Most high speed scanners use two fluorescent lamps running at a very high frequency, so as not to create patterns. The second lamp avoids shadows in the indentations of the paper.

Since the color of the light source determines which colors on the paper are visible or not to the scanner, white lamps will be neutral and see all colors on the paper. However, white lamps tend to loose brightness faster than green phosphor lamps which are used in most high speed scanners. When using green light sources, scanners tend to drop light green and light blue colors. When performing forms processing for OCR or OMR purposes, it is often better to remove the background of the form. By using a pastel blue or green color form with green lamps, this can be easily achieved. The traditional drop out color is however red. Many health claim forms, questionnaires, or tax forms use red for the background of the form. Most scanner vendors thus sell special models or offer options for their scanners, such as "red-ink drop out scanners" which use red lamps for illumination.

Interfacing the Scanner:

Scanners need a certain degree of intelligence to perform their basic functions. Usually these functions are commanded by a host computer. A scanning sub-system is composed of a scanner, an interface card (to the host) and a scanning program which communicates the user's requirements to the scanner and generates the appropriate output.

A selection menu, is available on the scanner for most options such as resolution, contrast selection, paper format, whether the image is bitonal or half-toned, or whether an autofeeder is used or not. Most scanning programs override these scanner choices and control the selection from the host.

Two types of data commute between the host computer and the scanner via the interface card: image data and control data. The control data, initiated by the host, selects the scanning options and tells the scanner when to start and stop scanning. Scanned image lines are then sent back to the host. Many high speed scanners use two communication lines such as a serial communication for the controls and a parallel line for the image. This method is usually referred to as serial/Video and requires a special interface card. Since there is no standard protocol for such communication, each manufacturer developed its own protocol and interface cards (or compression cards) and a different flavor of their boards for each scanner manufacturer (Bell&Howell compatible, Fujitsu compatible etc..) Certain scanner vendors such as VisionShape and Photomatrix sell scanners that emulate other scanner vendors rather than having their own standard.

To Compress or not to Compress:

A scanner interface card provides a hand-shake between the scanner and the computer, sending the command from the host and then receiving the image data from the scanner and placing it into the computer memory.

Raw image records are very large but contain a lot of irrelevant data (usually the white space between the text). Compression consists of reducing the space between relevant data to the number of pixels of the same color rather than the pixels themselves. An algorithm calculates the number of consecutive black or white dots and stores that value. The more white space between text, the better the compression. On the other hand, grayish background will compress very poorly.

Compression can be done in the horizontal direction (G3) or in both horizontal and vertical directions (G4). G3/G4 refers to the compression method used and are standards set by the International Telecommunication Commission (CCITT). In addition to the compressed image, an image record usually contains a header that identifies the image parameters such as resolution, height, width and other relevant information. Most vendors use a standard called Tiff, which stands for Tagged Image File Format. A G4 Tiff image is thus an image compressed using bi-directional G4 algorithm with a header that complies with the Tiff convention. Several vendors sell systems that use proprietary header formats but still use a G3 or G4 compression methods. Scanners themselves usually do not compress the image or place a header in the record. This is usually done in the host PC by the scanning software or the compression card.

In the early days of imaging, special chips were required to compress a raw image into a compressed record. With the advent of high speed PCs, in particular the Pentium, these chips are of little value for the compression and decompression of images. Simple interface cards such as Dunord, will transfer image data to the PC memory and a compression software program can compress the image at the same speed if not faster than most compression boards for a fraction of the cost.

A few scanners offer the option to compress the image in the scanner, in reality, a complete compression board is integrated in the scanner. One of the advantages of this method is that it allows the ability to use a more standard communication interface such as SCSI or Ethernet to transfer the data to the host PC. When the scanner does not compress the image but still uses a SCSI interface, the scanner still needs to include a page buffer and tends to be significantly slower than serial/video scanners.

There is little doubt that traditional compression cards are doomed and will be replaced by some high speed interface standards such as SCSI with higher performances within a few years.

Paper Transport:

The fastest way to scan an image is to move the paper quickly in front of the CCD using a paper transport. High-speed scanners have several different types of transports. There is no "best" way to accomplish this, because each kind of transport works better in specific cases. Generally speaking, four methods are used to transport paper: roller transports, belt transports, drum transports and vacuum transports.

The Roller Transport passes the document between two rubber rollers and feeds it past the CCD, much like a fax. Not surprisingly Fujitsu and Ricoh use these types of transports. The advantage of this transport is that it is simple and low cost, but it tends to double feed and is unable to handle a variety of paper thickness as well as other types of transports.

The Belt Transport passes the paper between two sets of opposing belts that transport the paper past the CCD. In some case, like the BancTec 2600, the upper transport is hinged to rest on the lower transport, enabling it to adjust up or down for thicker or thinner paper. The image is projected via mirrors through a lens onto the CCD. Since paper does not pass over glass, this type of transport is good at handling the occasional staple or paper clip that slips through document preparation. However, because of the large number of belts used for this transport, they can occasionally be damaged by staples or the belts may come undone.

Drum transports have the advantage that they can return the paper to the operator at the end of scanning which is very helpful if a document needs re-scanning. Bell & Howell and Kodak use this kind of transport. The disadvantage of drum transports is that they tend to have a difficult time with small documents or mixed batches of paper with very different thickness. Drum scanners tend to jam less than other transports, but when they do jam it tends to be more difficult to clear because you sometimes have to dig into the machine to find the paper. Drum transports also tend to skew more than other scanners.

Vacuum straight-through transports like Photomatrix and BancTec's 4000 series use a vacuum to hold the paper against a belt. These rugged transports, typically on the very high volume scanners, are very tolerant of rips and staples. The disadvantage is high purchase and maintenance cost and sometimes the fact that the paper is deposited in the back of the machine.

Single-side Versus Two-sided Scanners:

The cost of a duplex scanner is usually twice as much as an equivalent simplex scanner. However, the break-even point in deciding for a duplex scanner is somewhere between 20% and 40% (of dual-sided documents). It is feasible to have operators flip the pages and scan the other side, as long as the number of pages is relatively small, otherwise the operator cost, the risk of error and the requirement to handle small batches (or even sheet per sheet) outweighs the cost of a duplex machine.

High speed duplex scanners often have two complete cameras in the paper path. Whether straight-through or drum, the paper moves in front of one side, then the other side before being ejected. Bell & Howell and VisionShape scanners have two complete sets of cameras and control cards which requires two interface/compression cards in the host. On the other hand, the Fujitsu duplex scanner buffers one of the images and transfers the data through a single channel. Some high speed duplex scanners use a single camera but loop the paper back in front of the camera to scan the other side. The cost of a second camera seldom exceeds that of a complex loop mechanism.

Automatic Feeders:

The key to high speed scanning is high speed feeding. There is no such thing as a high speed feeder that can reliably feed all sizes and thickness of paper, like those found in accounts payable applications. Most mixed paper applications are still hand-fed. Bell & Howell, BancTec, Kodak and VisionShape scanners are the most appropriate for hand feeding documents, most Japanese scanners are not practical for this feeding method.

Scanners can have top-feeders or bottom feeders. With a top feeder scanner, like the Bell & Howell, the documents are placed face up (since the camera is on the outside of the drum) and the documents feed from the top of the stack first. The paper is driven into the scanner by a central wheel and separated from the next page by a belt that rotates in the opposite direction, pushing the second page backward, while the top wheel is pulling the top page in. The advantage of a top-feeder is that the stack is not exercising pressure on the document being fed, however the single wheel feeder has a tendency to exaggerate skew when the paper is not perfectly aligned in the feeder. Since the feeder is peeling from the top, the autofeeder can hold a larger number of pages than bottom feeder can.

In the case of bottom feeders, the documents are usually placed face down and the bottom sheet is fed first. When the stack of paper becomes too large (usually over 100 sheets), the pressure on the bottom page prevents the feeding process or increases double feeding. Because the documents are driven into the scanner by pressure rollers that extend on a larger area of the paper or on several areas, skew is less significant than with top-feeders. Bottom feeders also offer the advantage of allowing to add to the stack while scanning, eliminating the pause between stacks which is required by a top feeder.

Both top and bottom feeders are unfriendly to mixed width paper formats. The feeders utilize guides to hold the stack of documents together, so when a large page follows a narrow page, the guides hold the largest of the two and the narrow page goes into the scanner unguided, dramatically increasing the skew. Mixing long and short documents is usually not a significant problem, unless the short document is very short (less than 10cm).

Short documents are traditionally difficult to scan with page scanners. Some vendors like VisionShape offer narrow and short document attachments to help with these situations.

Options, Accessories and Gadgets:

There are many different requirements needed to scan and feed paper, this is why many options can be added to high end scanners. Already mentioned was the possibility of different color lamps (or filters) to drop back-ground colors and special attachments to hold shorter or narrower documents.

In some applications, it is important to be certain that a page has been scanned. Some scanners thus offer an endorser option. Endorsers are usually nothing else than a stamp which notifies that the page was scanned. On the other hand, imprinters usually print actual data on the paper such as the date of scanning and the sequential number of the page. Printing can be done horizontally or vertically, however horizontal printing involves stopping the paper, positioning the print head and printing the data, which tends to slow down the scanning process. Vertical printing is achieved with the paper moving and a fixed print head, but the data appears printed from top to bottom which makes it less legible.

Preventing double feeds is one of the most difficult tasks for a scanner. Most double feeds can be prevented by the operator and by following good document preparation procedures. There are many reasons why pages remain attached to each other and consequently are fed together into the scanner. Static is one of these reasons. A very dry atmosphere tends to create more static which virtually glues pages together. On the other hand, very humid air will create misfeeds (paper does not feed). It is thus recommended to place production scanners in a conditioned environment. Another reason for double feed is when the pages where stapled together and the staple was removed, the pages may remain attached to each other. When document preparation personnel remove staples, they should also fan the stack to create a true separation. Glue, left over from an envelope or a booklet, is obviously a major cause of double feeding. Sometimes this can be alleviated by scanning in another direction or upside down and then rotating the images in the software. No matter how much preparation is done, scanners will double feed.

Scanners that use retard belts to separate documents, such as Bell & Howell or VisionShape, allow users to put more or less pressure on these belts to increase separation (the more pressure, the better the separation). Another possibility is to detect a double feed before it is actually scanned with a double feed detector. One technique used is to beam an infrared light beam through the page, if the light does not go through the page there is a double feed. This method is rather inaccurate and only works with certain types of paper and constant backgrounds. VisionShape is placing multiple sensors in the paperpath and controls the detection with software algorithms.

Selecting a Scanner:

Most vendors of high speed scanners sell through indirect channels (i.e. OEM's, Distributors and Resellers). As a result, the selection of scanners, pricing and service is often dictated by the application vendor. Since scanners can make or break an application, users should carefully review the proposed scanner with their requirements and not necessarily take the most popular scanner, but rather the scanner that will offer them the best price/performance.

Normally scanners carry a one year warranty. With the smaller desktop scanners this is usually based on the "return to factory" policy. In a production environment, it's unacceptable to lose your main input device while it's returned to the factory for repair. High speed scanner users should therefore arrange for on-site service or replacement from the vendor within an 8 hour period. Higher volume users and those using document imaging as a transaction based paper processing system (as opposed to an archival storage and retrieval system) should consider purchasing at least two scanners, a primary high performance production scanner and a secondary lower speed device for repair and occasional scanning that can be used in case of emergency.

When selecting a scanner, it is always better to run some tests with actual documents with more than one vendor. Calculate the average speed of several batches of documents including loading, and clearing jams. Keep an eye for double feeds. Take the list price of the product including interface card and software and divide this number by the real page per minute throughput. Then compare the numbers, the lower the number the greater the productivity. For example, if a scanning system cost $10,000 (scanner, board and software) and it successfully runs 33 pages per minute with your documents, the score of this scanner is 300. If an $18,000 scanning system scans 45 pages per minute, the score is 400 which makes it less productive than the lower priced model. It is only when scanning over 15,000 pages per day, that the faster models start to make a significant difference and even then, a user may be better off with 2 scanners at $10,000 than one at $18,000.

Harvey Spencer is President of Harvey Spencer Associates, a consultancy specializing in Image Systems and Peripherals technical marketing.

Daniel Borrey is Vice President of VisionShape, a manufacturer of scanners and scanning systems. Prior to VisionShape, Daniel Borrey was President of Correlative Systems International. CSI developed the original CopiScan scanner, sold in 1986 to Bell & Howell. Daniel Borrey was also President of Corsys Belgium, a Belgian software company sold to DMR of Canada in 1987.

© VisionShape 1999
VisionShape and VisionScan are registered trademarks of VisionShape.
AutoScan is a trademark of VisionShape.
Windows is a trademark of Microsoft Corp.
All trademarks and product names are property of their respective owners.