Multimedia - School of Computer Science

School of Computer Science &
Information Technology
G6DPMM - Lecture 6
Colour Science & Colour Models
Colour Representation


Colour is represented as a number
Indexed Colour (<24 bit)

Each number is an index into a lookup table (palette)
For example (a 2-bit palette):
0 – black
 1 – white
 2 – red
 3 – green





Implemented in hardware (obsolete)
True colour definition
Colourmapping
True Colour (25 bit)

Each number represents a colour using a mathematical model
known as a “colour model”
The Physics of Colour

Light is an EM wave in the “visible” part of the
spectrum (400-700nm)
0.00001nm
0.001nm
1nm
10nm
Cosmic
Rays
Gamma
Rays
X-Rays
UV
0.00003m
Visible
Light
IR
400nm

Frequency represents colour
 Amplitude represents brightness
Microwaves
0.03m 0.3m 30m
TV
700nm
Radio
5,500km
Electricity
The Biology of Colour

The human eye



Rods



Very sensitive – can detect amplitude of light
No mechanism for colour detection
Cones




Light is focused onto the retina
The retina contains rods and cones
Less sensitive
Three types – Red, Green & Blue
Each can detect the amplitude of one “primary” colour
Additive colours


Cognition not physics!
Approximately 10-20 million additive colours can be distinguished
Primary / Additive Colours
Primary
Additive
Red
Green
Blue
 Red
 Green
 Blue
Red & Green
Red & Blue
Green & Blue
Red, Green & Blue
None





Yellow
Purple
Cyan
White
Black
RGB Colour Model

3 Colour Channels





RGB - 1 byte each - 0-255
Encodes 256 x 256 x 256 = 16,777,216 Colours
True colour (24 bit colour)
Notation is 3 integers (often written as hex)
Examples




255, 255, 255 (FF FF FF) - White
255, 0, 0 (FF 00 00) - Red
255, 0, 255 (FF 00 FF) - Magena
100, 100, 50 (64 64 32) - Olive
HSB Colour Model

Hue, Saturation & Brightness

Hue is an angle (0-360)
specifying the position on a
colour wheel.

Saturation is a percentage representing
the difference from a neutral grey.

Brightness is a
percentage representing
the continuum from
black to white.
0%
100%
0%
100%
Other Colour Models

CMYK


YIQ & YUV


Used for broadcast TV, an analogue system based upon
luminance, chrominance of wave phases.
YCC


Used mostly for printing, based on cyan, magenta,
yellow and black inks used for colour separation.
Developed by Kodak for Photo CD
Pantone

Colour “catalogue” used by printing industry.
Colour Management
“Perceived” colour is hard to keep exactly constant.
 Many factors affect this


Colour model
 Monitor
 Ambient lighting
 Platform (eg Macintosh is typically “brighter” than Windows)

Major problem in multimedia
Image Processing

Processing techniques can create an illusion of
colour and detail that is not really present.

Dithering


When reducing colour depth each pixel must be
replaced with a corresponding pixel in the target
palette.
Antialiasing

When resizing each pixel may be replaced by
intermediate colours to avoid “pixelation”
Dithering

Each pixel must be replaced with a corresponding
pixel in the target palette.
 Adjacent pixels are examined and intermediate
colours may be used
 Dithering software is built into most bitmap
editing/processing software
 Algorithms:




Random
Average
Ordered
Floyd-Steinberg
Random Dither

Generate a random (0-255) number
for each pixel
 If greater than the number
pixel=white otherwise black
 Crude and “noisy”.
 Almost never used
Random Dither
Average Dither

Calculate an average pixel value
 If each pixel is above this then white,
else black.
 Crude and “contrasty”.
 Almost never used.
Average Dither
Ordered Dither (pattern)

Divide the image into ordered cells - ie
matrices.
 Uses matrix arithmetic to compare each pixel
with the average “threshold”.
 Generate a block of pixels to represent each
cell.
 Widely used by the printing industry - rare in
multimedia.
Ordered Dither
Floyd-Steinberg Dither

Error diffusion, diffusion, dispersion.
 For each pixel the closest colour is found
 The difference between this and the original is the error for
that pixel.
 The error is then “diffused” over adjacent pixels that have not
yet been processed.
 When these pixels are processed, the error is added to the
newly calculated colour.
 Widely used in multimedia.
 Many minor variants.
Floyd-Steinberg Dither
Original (24-bit) Image
8 Colour - No Dither
8 Colour - Ordered Dither
8 Colour - Diffusion Dither
Antialiasing



Resampling
Avoids pixellation (“jaggies”) on resizing.
Creates intermediate colour pixels around
edges.
Original Image
Scaled Image
Scaled Image (detail)
Scaled Image (antialiased)
Antialiased image - detail