Input and output devices

A standard keyboard has about 104 keys. Note that Backspace deletes to the left of the cursor and Delete to the right — a distinction questions sometimes ask for.

• Pixel — picture element: the smallest individually addressable dot on a display. Every image is built from pixels.

• Resolution — the total number of pixels on the screen, written $\text{width}\times \text{height}$, for example $1920\times 1080$ (Full HD). Higher resolution means finer detail and more content visible at once.

• Refresh rate — the number of times per second the whole image is redrawn, measured in hertz (Hz). $60$ Hz is standard; higher rates give smoother motion. On old CRT monitors a low refresh rate produced visible flicker.

• Dot pitch — the distance between the centres of two adjacent pixels, in millimetres. Smaller is better: a smaller dot pitch packs pixels closer, giving a sharper image.

Note that resolution alone does not fix sharpness — $1920\times 1080$ looks razor sharp on a phone and coarse on a large television, because the same pixel count is spread over a larger area, giving a larger dot pitch.

Impact — two examples.

1. Dot matrix printer. The print head carries a vertical column of pins; selected pins are driven forward to strike the ribbon, so each character is formed from a matrix of dots. 9-pin heads give draft quality, 24-pin heads near-letter quality. A character printer. Still used where carbon copies are needed — railway tickets, invoices, bank passbooks.

2. Line printer (drum or chain). Prints an entire line at a time rather than a character, making it very fast. A drum printer carries the character set on a rotating drum; a chain printer on a revolving chain. Used with mainframes for very high volume listings.

Non-impact — two examples.

1. Inkjet printer. Sprays extremely fine droplets of liquid ink through nozzles onto the paper. Inexpensive to buy, excellent colour and photographic reproduction, but high cost per page and ink that can smudge or dry in the nozzles.

2. Laser printer. A laser beam writes the page image as a pattern of electrostatic charge on a rotating drum; dry toner powder clings to the charged regions, is transferred to the paper and fused by heat. A page printer — it assembles and prints a complete page at a time. Fast, very sharp, and cheap per page, making it the office standard.

The decisive difference to state in an answer: only an impact printer can produce carbon copies, because only it applies physical pressure to the paper.

All four read data from a printed document, but each recognises something different.

1. OCR — Optical Character Recognition. Software that examines a scanned image and identifies the shapes in it as characters, converting a picture of text into editable, searchable text. Works on printed and, less reliably, handwritten text.
Application: digitising printed books, records and forms; reading vehicle number plates.

2. OMR — Optical Mark Recognition. Detects only the presence or absence of a mark at a fixed, predetermined position on a specially printed sheet. It does not identify characters at all, which is exactly why it is so fast and so accurate.
Application: marking multiple-choice examination answer sheets; ballot papers; survey forms.

3. MICR — Magnetic Ink Character Recognition. Reads characters printed in ink containing iron oxide, in the standard E-13B font, by magnetising the ink and sensing the pattern. Because the reading is magnetic rather than optical, it succeeds even when the characters are overprinted, stamped, folded or smudged, and it is very hard to forge.
Application: the code line at the foot of bank cheques, giving cheque number, bank and branch code and account number.

4. Barcode reader. Scans a printed pattern of parallel bars of varying thickness that encodes a number, and passes that number to a program which looks up the corresponding record. A QR code reader does the same for a two-dimensional code, which holds much more data and can be read at any angle.
Application: retail billing and stock control; library book issue; parcel tracking.

How to keep them apart: OCR reads characters, OMR reads marks, MICR reads magnetic characters, and a barcode reader reads a bar pattern.

A plotter is a hard-copy output device that draws with one or more pens moved mechanically across the paper under program control. Rather than depositing dots to build an image, it draws continuous lines from point to point.

Differences from a printer.

Two types.

1. Drum plotter. The paper is wrapped round a rotating drum. The drum turns the paper along one axis while the pen carriage moves along the other, so the two motions together trace any line. Because the paper can keep rolling, a drum plotter can produce a drawing of almost unlimited length while occupying little floor space.

2. Flatbed plotter. The paper lies flat and stationary on a bed, and the pen moves over it in both directions. Accuracy is higher, since the paper never moves, but the machine must be at least as large as the biggest sheet it can take.

Applications. Engineering and architectural drawings, circuit layouts, maps and survey plans, and graph plotting — all cases where the correctness of a line matters more than the appearance of text.

The two files look similar on screen but contain fundamentally different things.

A typed page is stored as character codes. The letter A is held as the number $65$ (in ASCII), and the file records a sequence of such codes together with formatting instructions. Because the file knows which characters are present, the program can search for a word, change a font, correct a spelling, count words, or read the text aloud.

A scanned page is stored as an image — a grid of pixels, each with a colour value. The file records that a certain pixel is black and its neighbour is white. It contains no characters at all. The shapes only look like letters to a human reader; to the program they are indistinguishable from a photograph of a tree. Nothing can be searched, corrected or restyled, because there is nothing there to search.

$$\text{typed page}=\text{sequence of character codes}\text{scanned page}=\text{grid of pixel values}$$

What OCR does. Optical Character Recognition bridges the two. It:

1. pre-processes the image — straightens it, removes speckles, and reduces it to black and white
2. segments it, locating blocks of text, then lines, then words, then individual characters
3. recognises each isolated shape, by comparing it against stored patterns or by measuring its features, and decides which character it most probably is
4. post-processes using a dictionary and language rules, so that a shape read ambiguously as rn can be corrected to m when the surrounding letters make a real word

The output is a file of character codes — now genuinely editable, searchable and reformattable.

Why OCR is imperfect. The recognition step is a decision under uncertainty. Poor scan quality, unusual fonts, handwriting, small print, tables, multiple columns and mixed languages all raise the error rate, and the classic confusions ($0$/O, $1$/l/I, rn/m) are between shapes that genuinely differ by very few pixels. This is why OCR output always needs proofreading, and why a scanned archive is usually stored as image and OCR text together — the image as the faithful record, the text so it can be searched.