Scor­pio News


April–June 1988, Volume 2, Issue 2

Page 30 of 35

A computer, say when used as a word processor, normally stores text as the actual letters that compose the text but if it has to store pictures this is done completely differently, as a pattern of dots. An imaging system can’t take chances, nor is it intelligent enough to make the decision between text and pictures, and anyway, even the best character reading software in not infallible and makes mistakes, so we can assume a computer can’t read well. (I always thought the name Palantir for the supposed best OCR system around was either chosen by mistake or as a wry joke perpetrated by the designers, on the users. No I shan’t explain that, for a definition of Palantir see ‘Lord or the Rings’ book 2, ‘The Two Towers’ by Tolkien.) To play safe the computer is sure of one thing, whether the image is text or a picture it can always be treated as a picture. That’s what it does – hence imaging system – it always treats anything it sees as a picture,

The scanner.

Enough descriptions of what goes on, starting at the beginning, how do images get into system ? I’ve already said that the image is stored as a pattern of dots. The binary system is ideal for dealing with dots, it means that for a monochrome image, we only have to think in terms of black and white – ‘0’s can represent black and ‘1’s can represent white. The device used to convert the image on a piece of paper into a digital representation is called a scanner. These can be flatbed, like most photocopiers, or rotary, where the paper is popped in a slot and re-appears out of another. In a flatbed scanner, the paper stays still and the scanner sensor moves (or at least a mirror reflecting into the scanner sensor moves), in a rotary scanner the scanner sensor stays still and the paper moves. There are two common sizes of scanner, A3 and A4. A3 scanners are always to be flatbed, at least, I haven’t seen a rotary A3 scanner.

The way a scanner works is simple. I guess you’ve an idea how.a television picture is built up on a TV screen, it’s drawn as horizontal lines, 625 of them in Europe, from the top to the bottom of the screen. The same in reverse for the scanner. As the image moves past the scanner optics, the image is projected on to the scanning sensor, the sensor is usually a device known as a CCD (Charge Coupled Device). An A4 size CCD consists typically of 3,500 minute discrete light sensors side by side ina strip. The ‘Charge Coupled’ bit describes the way they work and is of no concern at the moment. As the image falls on them, each sensor will either see light reflected from the paper, the white bits, or no light reflected from the paper, the black bits. They can also make some decisions about colour and halftones. Some scanners don’t use a CCD but mechanically scan the document with a single laser using a system of rotating mirrors. These are much faster, the resolution is very much better, but they have real problems with halftones and colours and cost the earth.

Want to try it ? Take a piece of paper with a picture on it, a newspaper picture will do nicely, lay a ruler graduated in millimetres horizontally down on the paper so the

This is an OCR’d version of the scanned page and likely contains recognition errors.

Page 30 of 35