Two steps : ... to the digital world
Digitisation (Sampling) : Discretisation in space or time
Sampling pattern : image space is tessellated into discrete, local, compact, regions (regular rectangular planar grid)
Sampling process : Point in neighbourhood, Average over neighbourhood
Nyquist Criterion : Sample take at least twice highest frequency contained in the signal of interest
 |
Sound : Rate (44kHz) and Size (8 or 16 bits)Image : Resolution (300 dps) and number of colours |
Quantisation : Discretisation in Value
The function can take on only finitely many values
E.g. for images 3 general domains :
-
Large: 32-64 bits per channel : a billion or more intensity values, "real number" representation : computational purposes
-
Full Web colors : 16*16 * 16*16 * 16*16 = 2^8 * 2^8 * 2^8 = 2^24
-
Typical JPEG images provide 24 bits per pixel, but that's 8-bits per channel (red, green, and blue)
-
Moderate: 8-12-16 bits per pixel : 256-4096-65536 intensity values, "integer" representation : Good enough for almost all display purposes 8 : 256 colors / 12 = 3 * 4 : 16*16*16
-
Small: 1 or 2 bits : black and white
See http://www.devx.com/projectcool/Article/19997/0/page/7 for more on Bit Depth