Digital Images

1
Digital Images

• An image is a visual output of data stored in
  terms of a numeric, recordable elements.
• Most digital images we use today are output in
  terms of a regular grid of pixels, referred to as
  a raster.
• Intermediately the image can be stored in the
  form of lines, curves and filled areas, referred
  to as vector primitives.
• All of these are eventually represented in a
  numerical way.

2
Overview

• How are images stored and displayed
• Hardware
• Colour representation colour formats
• Palettes and CLUTs
• Rasterization, resolution, compression

3
Display Hardware

• Hard Copy devices
• Printers, plotters
• Transient displays
• LCD Monitors, CRT Monitors, projectors
• 3D Displays

4
Cathode ray tube

• (Still ?) Most common is Cathode Ray Tube (CRT)
  monitor
• Horizontal and vertical deflectors focus an
  electron beam on any spot on a phosphor coated
  screen

Electrons hit the screen phosphor molecules and
excite them
Karl Ferdinand Braun 1897 Cathode Ray
Oscilloscope
5
Colour Systems

• Phosphors have a colour. Colour systems have
  groups of 3 different phosphors, for red, green
  and blue.
• Each pixel consists of 3 dots of phosphor,
  arranged as triangle. 3 Electron guns used, for R
  G and B
• Combining different intensities of phosphors can
  generate different colours

Hitachi EDP
Standard Dot-trio
SONY Trinitron CRT
NEC Hybrid Mask
6
Shadow Masks
7
Vector Display Devices

• A.K.A. Vector Scan Displays, Random Scan Devices,
  Line Plotters
• The electron beam directly draws the picture e.g.

DrawLine(A, B) Turn beam off, move to A. Turn
beam on, move to B.
8
Raster Scan Devices

• Scans the screen from top to bottom in a regular
  pattern (common TV technology)
• A raster is a row of pixels (picture elements)
  covering the width of the screen
• The electron beam is turned on/off (high/low) so
  the image is a collection of dots painted on
  screen one row (scan line) at a time.

9
Raster Image

• An image made up of many small regularly placed
  cells called pixels (picture elements)
• Stored as an array of numerical values commonly
  called a pixelmap, pixmap or bitmap

10
Raster Image Storage
11
Computer memory is usually one dimensional (or
sequential) The 2D image is unravelled in order
to store this in memory
12
Colour Rasters
Blue 0,0,255 Red 255,0,0
13
Unravelling this leads to a list thats 3 times
bigger
14
Sprites
15
Sprites
16
2. Colour

• A sensation which occurs when light energy,
  incident on the retina is interpreted by the
  brain.

Colour is a psychological interpretation, not a
physical property of light itself. Source
Psychology Themes and Variations - Wayne
Weiten
17
(No Transcript)
18
(No Transcript)
19
Half-toning exploits perceptual defects to
create the appearance of more shades/colours than
can physically be created by the output medium.
20
Colour in Computer Graphics

• How we see colour Colour Vision
• The response of the visual system to colour or
  Colour Appearance
• How to measure colour Colorimetry
• Computer Graphics is interested in
• Colour Vision (perception)
• Colour Reproduction
• Colour Synthesis

21
The Process of Seeing Colour

• Light enters the eye as a spectrum of colours,
  distributed by wavelength can be represented by
  a Spectral density function
• Spectral Density impinges on retina and is
  picked up by light-sensitive cells
• The retina of the eye is its light-sensitive
  membrane, containing two types of cells
• cones colour sensitive cells
• rods cannot distinguish colour, nor fine detail,
  but sensitive to low levels of light (night
  vision)

22
Visible Light

• LIGHT is an electromagnetic phenomenon, like
  television waves, infrared radiation, and x-rays

23
Pure Spectral light

• The human eye responds to light with wavelengths
  between approximately 400 and 700 nm
• Some light sources emit light of essentially a
  single wave-length (e.g. lasers)
• Some spectral densities for pure spectral light,
  and how we perceive them

Spectral Density (power per unit wavelength)
Green
Orange
Violet
Blue
Yellow
Red
400
500
600
700
wavelength
24
Spectral Density Function

• In most cases, light consists of a range of
  wavelengths
• These mixture of wavelengths can be represented
  by a Spectral Density function
• Some spectral functions densities are shown to
  the right

25
Sample Spectral Distributions
26
Cones

• Human beings have three types of cones responding
  to different wavelengths of light
• Short, Medium or Long wavelengths
• A range of wavelengths not actually Red/ Green/
  Blue
• but sensitivity determined by the degree to which
  each cone-type is stimulated
• Cones absorb light and send signals to the brain
• The Spectrum is encoded into three values that
  correspond to each type of cone trichromacy
  model

27
Chromatic Vision

• Humans and some primates are trichromats
• Colour blindness results when one cone system is
  compromised
• Mammals are mostly dichromats, many birds and
  marsupials are tetrachromats
• Some human females are said to be tetrachromatic!
  -)
• Rods are lower intensity light receptors and only
  kick in in low light situations.

rods
400
700
Rod and Cone Response N.B. Sensitivity is NOT TO
SCALE!
28
Color Vision Deficiency
As seen by a person with red/green color deficit
(deuteranopia)
As seen by a person with a blue/yellow color
deficit (tritanopia).
http//www.vischeck.com/examples/
29
Colour Perception
Brain sees due to cone response
Different Spectra sometimes appear the same
30
Perceived Colour

• The light from most sources usually consists of a
  distribution of wavelengths
• An enormous variety of spectra is perceived by
  the eye as having the same colour
• Conversely two objects of the same colour may
  appear different to different observers or when
  viewed in different conditions - Metamerism

31
Representing Colour

• How can a particular colour described
• Precisely?
• Unambiguously?
• Ideally, every colour should be described
  uniquely in some numeric way
• What coding scheme can be used to map colours
  into numbers, and vice versa?

32
Dominant Wavelength Theory

• Capitalizes on the variety of spectra that
  produce the same perceived colour
• Specifies a spectrum having this simple shape

B
D
A
W
400
700
620
33
Dominant Wavelength Theory

• Luminance, the total power in the light
• L (D - A)B AW
• Hue the location of the dominant wavelength, i.e.
  the colour of the main pure light present
• Saturation the purity of the light, i.e. the
  percentage of luminance that resides in the
  dominant component

34

• Dominant wavelength, luminance and saturation
  fully define a colour

D A, saturation is 0
A0, pure light
Unsaturated pastel colour
35
Hue saturation and lightness
36
HSL Doublecone
37
Visible Hues and Saturations

• The eye can distinguish about 128 different hues,
  and about 20 different saturations (depending on
  the hue).

256 shades
32 shades
16 shades
38
The RGB Model

• An Additive colour model based on the primitives
  Red Green and Blue
• Most commonly used in Computer Graphics
• Approximates human visual response fairly well

39
Additive Colours

• CRT Systems employ additive colour mixing
• Red Green Yellow
• Red Blue Magenta
• Green Blue Cyan
• Red Green Blue White
• We get intermediate colour mixes by varying the
  intensity of each of the three primaries.
• Light Orange
• 100R 50G 20B
• Solid spectral colours are optically mixed
• by being placed closely together or
• by being presented in rapid succession
• Two or more colours perceived as being one

40
Digital Colour

• RGB colour
• 24 bit colour 8 bits for each primary
• 0-255 in unsigned integers
• Many graphics packages however would use real
  numbers ranging from 0 to 1 e.g. float (1.0, 0,
  0) equiv. to (255, 0, 0)
• 32 bit colour 8 bits per colour 8 bits for
  transparency
• 48 bit color 16bits for each primary. For
  editing high-color resolution is important

41
(No Transcript)
42
RGB Colour Space
B
BLUE (0,0,1)
Cyan (0,1,1)
Magenta (1,0,1)
White (1,1,1)
GREEN (0, 1, 0)
G
RED (1,0,0)
Yellow (1,1,0)
R
43
Device Dependency

• Unfortunately the choice of red, green and blue
  can be quite different in different
  implementations.
• If the phosphors colours are slightly different
  the vector space is different so we cannot use
  RGB to universally define a unique colour.
• We require a device independent colour space.

44
CIE Standard

• Standard developed by Commission Internationale
  de L'Eclairage (1931)
• Defines colours based on the three standardised
  primaries X, Y and Z which are able to produce
  ALL visible colours.
• Provides a standard reference for comparing other
  colour systems

45
CMY Colour

• Subtractive scheme based on primitives Cyan,
  Magenta and Yellow
• Commonly used in colour print production

46
Subtractive Colours

• Subtractive colour mixing results from selective
  absorption of light wavelengths
• Overlapping layers of varying percentages of
  transparent primaries are laid down.

Layers of CMY ink subtract inverse percentages
from the reflected light so that we see a
particular colour.
47
CMYK

• 'black' generated by mixing the subtractive
  primaries is not as dense as that of a genuine
  black ink
• four-color printing uses black ink in addition to
  the subtractive primaries
• Ccyan, Mmagenta, Yyellow, Kkey (black)
• Better for contrast

48
(No Transcript)
49
Relevant Issues
50
Rasterization

• Line drawing algorithms render a mathematical
  ideal of a line onto raster storage
• Similar techniques for curve-drawing and shape
  filling

51
Aliasing

• Anti-aliasing processes effectively blur edges to
  soften the boundaries

52
Resolution

• Rasterizing objects effectively binds them to a
  certain raster level of detail.

53
Screen Resolution and Colour

• An image is stored in a special graphics memory
  area called a frame buffer
• A display processor scans this memory and
  controls the electron beam at each pixel
  accordingly
• Monochrome system each pixel is either on or
  off 1 bit per pixel
• Grayscale images 8 bits per pixel gives 256
  different intensities of gray
• True color 8 bits/pixel giving a total of 24
  bits per pixel and about 16,000,000 displayable
  colours.

54
Memory Usage Example
Black-and-white
Greyscale
True-colour
8x8x1 64 bits
8x8x8 512 bits
8x8x24 1536 bits
55
Palettes
56
Colour Lookup table (CLUT)

• In many colour raster systems we use a colour
  lookup table or palette.
• The value of a pixel in the frame buffer is not
  used to directly control the beam, but is an
  index into the LUT
• The entry in the LUT is used to directly control
  the colour of the pixel
• Eg. If we use 1 byte (8 bits) per pixel in frame
  buffer, 6 bits for each of R,G and B in LUT, then
  an application can choose 256 (28) colours out of
  262,144 (26x26x26) available colours.

8x8x8 512 bits But a large range of colours
57
Direct Color
Indirect Color with Colour Lookup Table
256 shades
DC 24 bits per pixel required 7.2M CLUT256
entries 3.6M
32 shades
DC 18 bits per pixel required 5.4M CLUT64
entries 1.8M
16 shades
DC 12 bits per pixel required 3.6M CLUT16
entries 1.2M
58
Scanning Colours

• The CCD used in scanners and digital cameras, is
  a sensor containing an array of capacitors.
• Incident light hits the sensors and generates a
  small electronic current which can be digitized
  and stored
• Ratio of incident light to recorded colour it is
  much more effective than real film (Often used in
  astronomy)
• CCDs are typically sensitive to infrared can
  be used for night-vision

Fuji "Super CCD"
59
Bayer Colour Filter Array

• CCDs just record intensity of incoming light
  (photons)
• To record colour, we filter light for individual
  sub-pixels
• Bayer array uses regular pattarn of colours
  arranged in a 2 x 2 squares
• Demosaicing techniques are used to reconstruct
  the image
• N.B. There are various alternative patterns, in
  different implementations

Bayer filters overlayed over sensor array
60
Demosaicing

• We have a mosaic of Bayer squares. Each pixel is
  either entirely green entirely blue or entirely
  red.
• But we need RGB colours for each pixel!
• One naive solution
• Assign one colour C for each of the four pixels
• Rredintensity, Bblueintensity, G 0.5 (g1g2)
• However we effectively have pixels of half the
  original resolution

(0, 0.1, 0)
(0.5, 0, 0)
(0.5, 0.2, 0.3)
(0, 0, 0.3)
(0, 0.3, 0)
61
Demosaicing
Bayer Pixels
Demosaiced With Linear Interpolation
Other strategies nearest neighbour replication,
bilinear interpolation, bicubic interpolation,
spline interpolation,laplacian interpolaion