Flatbed scanners operate using the basic principle of reflection. The subject to be scanned is placed face-down on the 'bed' of the scanner and as the scanning head moves down the bed, the light generated from a light source is directed onto the subject and sensors measure the amount of light reflected from it.
The reflected light is converted by the sensors to a voltage that is proportional to the light intensity, i.e. the more light reflected the greater the voltage generated. This analogue voltage is then converted to a digital format that can be used by a computer.
Once the computer receives the digital data it is passed to an application program which assembles the data into an image for subsequent editing.
The scanning head
The scanning head contains the light source, the CCD sensor and an arrangement of mirrors and lenses for managing the light reflected from the subject and directing it onto the CCD sensor.
{2}The head is mounted on a belt-driven assembly and moves up and down along the bed of the scanner. Its movement is controlled by a stepper motor. Unlike a conventional
motor which is designed to spin freely, a stepper motor is normally held stationary but can be made to move in steps one way or another by the application of electrical pulses. Its position can therefore be controlled very accurately - exactly how accurately depends on its precise design. Stepper motors are used for controlling the print heads of inkjet and dot matrix printers, which is why leading printer manufacturers often also produce flatbed scanners.
{3}The light source
The light source used in most desktop colour scanners consists of one or more fluorescent lamp tubes. A red, a green and a blue light source are required for colour scanning. Epson scanners use an individual colour tube for each of the three colour sources. Others may use a single white light with coloured filters or a prism system to obtain the red, green and blue sources.
Monochrome images are generally scanned using a white light source. Colour scanners usually mix red, green and blue light to produce the white light required for monochrome scans. Special results can be obtained in monochrome mode by scanning with either red, green or blue light instead of white.
{4}This will cause any areas of the document which are the same colour as the scanning colour to disappear from the resulting image. This process is known as scanning using a 'dropout' colour. It can be useful for scanning documents with coloured backgrounds or monochrome images with additional colour annotations.
The CCD image sensor
The CCD sensor is incorporated into the scanning head. CCD is an abbreviation for Charge Coupled Device and,
in the form used in a scanner, it resembles a long thin computer chip with a glass window in the top. The function of the CCD is to store and transmit the levels of received reflected light from the image as accurately as possible.
The CCD itself consists of thousands of photosensitive elements. The number of these elements is the main determinant of the resolution of the scanner. For example, if the scanner has maximum scan width of 8.5" and the CCD contains 2,540 individual photosensitive elements, the maximum resolution an image can be scanned at 2,540/8.5 = 300 dpi. There are two types of CCDs commonly in use: the linear array and the matrix array (see opposite).
{5}Most scanners have moving light sources and tend to use linear array CCDs. Other types of image-capturing devices such as digital cameras and camcorders tend to use matrix arrays.
The output from the CCD needs to be converted to a digital format before it can be processed by a computer.This conversion process is performed by an analogue-to-digital converter circuit (ADC).
{6}Optical scanner resolution
The usual objective of scanning is to produce as close a reproduction of the original image as possible. One of the main factors in this reproduction process is the physical resolution of the scanner which is usually measured in dots per inch (dpi).
The physical resolution depends on two factors:
the maximum resolution obtainable by the CCD in the main-scan direction (see previous page)
the mechanical accuracy of the stepper motor which moves the scanner head vertically down the image - (the sub-scan resolution)
{7}In most A4 desktop scanners, both the motor and the CCD image sensor have a maximum potential resolution of 300 dpi. In more advanced desktop scanners the physical resolution can be even higher.
Data format
The quality of the scanned image also depends on the data format selected in your scanning software. The data format is usually measured in 'bits per pixel' - the number of binary digits used to represent the level of one scanned pixel, and is often referred to as the pixel depth. The appropriate setting of this format will be determined by the type of image being scanned as shown in the next table.
When line art is scanned, each pixel is returned as either black or white. This means that a data format of only 1 bit per pixel is required to reproduce this image - each pixel being either 0 (black) or 1 (white).
A greyscale image is usually scanned at 8 bits per pixel which allows 256 (2 to the power of 8) possible levels for each pixel. Each pixel can have the value 0 (black), 255 (white) or any of the 254 shades of grey in between.
{8}A full colour image typically uses 8 bits per pixel for each of the three primary colours, red, green and blue which gives 24 bits in all. A maximum of 16.7 million (256 x 256 x 256) different colours can therefore be represented, which is more than the human eye can perceive.
Colour scanning sequence
Colour images need to be scanned in red, green and blue, and the scan data for each colour sent back to the computer. There are two ways of doing this: three pass scanning or single pass scanning
three pass
In this mode, the original document is scanned three times, once each with red, green and blue light. This separates the image into its red, green and blue components and the information for each colour is sent separately back to the computer.
single pass
In single pass mode, the document is scanned in one pass of the scanner head with the red, green and blue lamps lighting alternately. The information for the three colours is sent back to the computer a line at a time.
{9}The scanning head can move faster in three pass mode than in single pass mode. However, since it has to make three passes over the original, returning to the top after each, the total time taken may be longer.
One mode or the other may give slightly improved results depending on the nature of the material being scanned. There are arguments for and against single pass and three pass scanning.
For example, some argue that you will get colours interfering with each other in a single pass device, or that you will have registration problems with a three pass device. In practice, the current devices on the market generally offer good quality scans. Results will be influenced to a large extent by the nature and quality of the original image being scanned. Some manufacturers avoid the single versus three pass issue by offering their scanners with the ability to do both.
{10}Scanner pre-processing
Many desktop colour scanners are intelligent devices which contain the processing abilities to allow them to manipulate images within the scanner before sending them to the computer. Most image-editing applications can perform similar functions on the image after scanning, but used properly, pre-processing facilities can reduce the amount of work required after scanning has taken place. Only practical experience can help you decide whether to use the scanner's own processing capabilities or whether to bring the data in unprocessed and let your application or printer driver do the work.
{11}Interpolation
Many scanners incorporate an interpolation function to increase the output resolution. Interpolation may effectively double the resolution of the scanner, for example 400 dpi physical resolution can be increased to 800 dpi output resolution.
Interpolation works by reading the values for adjacent pixels and calculating an average intermediate - in effect, filling in the gaps. This can provide a useful increase in resolution without affecting accuracy significantly.
{12}Halftoning
Some scanners are able to carry out a process called 'halftoning'. Halftoning is used to convert continuous tone images into patterns of dots. This process is necessary because most printing devices print using colour or black and white dots - they cannot print continuous tones. The best place to find an example of a halftone is in a newspaper - you will not need to look too closely to see that images are composed of patterns of small dots.
Although it is more usual to carry out halftoning using your application software after scanning, since some scanners can halftone at the scanning stage, this would seem the appropriate point to introduce the basic concepts.
{13}Creating halftones
The need to reproduce continuous tone images in magazines and books has existed for a lot longer than scanners and image-technology.
Traditional halftoning
The traditional way to produce a halftone is to re-photograph an image through a screen made up of tiny rows of holes. Large dots are used to fill dark areas and smaller dots fill the lighter areas. The density of the dots (i.e. their resolution) is called the 'screen frequency' which is measured in lines per inch (lpi).
As well as having different screen frequencies, halftones can also have different screen angles and dot shapes. Screen angle refers to the angle at which the screen was placed when the halftone was created. Turning the screen diagonally to 45o is the most successful technique, since it does not interfere with vertical and horizontal lines in the image. The shape of the holes in the halftone screens can be round, elliptical or square which produces corresponding shaped dots in the halftone. Particular shaped dots are better for reproducing different types of original image.
{14}For colour images, four halftones need to be created - one for each of the cyan, yellow, magenta and black components of the colour image (refer to the Colour theory section for more details). In addition, the halftone screen needs to be rotated to a slightly different angle for each colour. This ensures that on printing, there are no clashes where various colour dots overlap, which would interfere with colour-blending and produce 'MoirΘ' interference patterns.
Creating halftones using a scanner
Some scanners have pre-processing functions which mimic the traditional method of producing halftones. They usually provide a choice of screen frequencies, screen angles and dot shapes. However, it is more usual to halftone images using your applications software at the post-scanning stage. Although it can cut down image file sizes, using the scanner's built-in halftoning functions may limit the extent to which you can manipulate your image later on.
{15}Working with halftones
If you are trying to print halftones, the final quality of the image will depend to a large extent on the characteristics of the output device you use - refer to the Output devices section for more details.
Finally on halftones, a word of warning - avoid scanning images which have already been halftoned (such as in magazines and other printed materials). The image may look as though it will work until you try to apply your own halftone: the additional set of dots will produce noticeable interference effects in the printed image. However, if it is essential to scan a pre-printed image, try using different scanning resolutions to avoid these interference patterns, or use the facilities like 'blur' and 'sharpen' in your image-editing package to modify your image to an acceptable quality.
{16}Dithering
Dithering is another technique sometimes carried out at the pre-processing stage by scanners which do not have any greyscale capability. It is more usually carried out at the output end by your application software or printer driver.
Dithering is designed to compensate for the fact that dots produced by a printer are not the same as halftone dots: halftone dots can give the illusion of greyscale (or colour tones) by varying in size, but printer dots are fixed at the resolution of the printer. Dithering introduces a third type of dot sometimes called a 'halftone cell'. This increases the number of greyscales which can be reproduced, but at the expense of resolution (see Output devices for more details). If you have a choice, always leave dithering until the output stage. However, if your scanner has no greyscale capability, you may have no choice.
{17}Gamma correction
Another function available in most scanners is Gamma correction. This function uses a mapping table to adjust the brightness of pixels in the image relative to their original brightness, typically by highlighting the dark-to-middle tones while having less of an effect on the lighter tones. This enables tones to be rendered accurately on various types of devices with different characteristics, for example different kinds of display monitor or printer.
Colour correction
If you are printing a colour image, you may need to adjust the image to compensate for the process used by your output device to represent colour images. This is achieved by colour correction functions.
{18}These are generally only available when scanning in single pass mode, because in three pass mode the scanner does not have the red, green and blue values for a particular pixel available to it all at once. By the time the 'blue' information is scanned the 'red' information has long since been sent to the computer, therefore it cannot assess the correction required.
Scanner interfaces
Because of the quantity of data that is transmitted when scanning at high resolutions - 25 MB for a 300 dpi A4 colour image - one of the primary requirements of the interface is high speed. This eliminates RS-232C serial interfacing on practical grounds. Even if the RS-232C interface circuit on a typical PC-compatible could be reliably run at its highest possible speed, 25 MB would take half an hour to transmit.
A low cost solution is to use a bi-directional version of the parallel printer interface, as fitted to many modern PCs. However, the 25 MB of data would still take between five and ten minutes to send.
The best solution, however, is to use a SCSI interface. SCSI interfaces can support high data transfer rates - several megabytes per second, much faster than scanners can scan. Another advantage is that because several SCSI devices, such as a scanner, CD-ROM drive, hard disk and tape backup unit can be 'daisy chained' together, only one connection is needed to the host PC.
{19}Options
Two options are commonly available for desktop scanners: an automatic document feeder and a transparency adaptor.
Automatic Document Feeder
Automatic document feeders are special sheet feeders which provide 'hands-off' operation for multiple sheet work.They work either by moving the paper past the normal scanning head while it is parked in a special position, or by feeding and positioning consecutive sheets so that the scanning head can move across them.
Page hopper sizes can vary from 10-50 sheets but specialist devices may hold even more. Speedwise, typical general purpose scanners can achieve between 1 to 8 pages per minute. Specialist archiving or DIP scanners can reach speeds in excess of 50 pages per minute.
{20}Transparency Adaptor
Transparencies cannot be scanned in the same way as reflective materials since they reflect only a small amount of the light directed at them. As a result they appear extremely dark when scanned. Placing a white reflective background behind the transparency may help a little, but a far better solution is to position a second light source behind the transparency to shine light through the transparency onto the CCD. This, in essence, is how a transparency adaptor works.
When using a transparency adaptor, the overall resolution of the scanner normally remains unchanged even though the transparency itself may be quite small, particularly if it is a 35 mm slide. As a result, high scanning resolutions will usually be needed to achieve sufficient magnification, and the best results will be achieved with large-format (5" x 4") professional transparencies.
