Visual Data Reproduction

1
Imaging and Special Effects CE00376-2
Fac. of Comp., Eng. Tech. Staffordshire
University
Visual Data Reproduction
Dr. Claude C. Chibelushi
2
Outline

• Introduction
• Image capture
• Light sensor
• Capture card
• Colour spaces
• Image file formats
• Image quality issues
• Summary

3
Introduction

• Image quality is affected by each stage of
  imaging chain
• image capture hardware, software tools, display
  device, image storage, human operator
• Achieving and maintaining high image quality is
  difficult
• requires adjustment and calibration of devices /
  processes, and skill of human operator
• e.g. proper setting of computer monitor, camera
  or scanner controls careful choice of image
  capture hardware and software image correction
  or enhancement using image processing software
  careful choice of image file format ...

4
Image Capture

• Image / video capture usually requires
• light sensing device
• e.g. analogue / digital camera, or scanner
• capture card
• high capacity storage
• high throughput hardware (e.g. CPU computer bus
  memory I/O)

5
Image Capture
Block diagram of image / video acquisition system
6
Image Capture

• Colour mixing
• Mixing colours produces other colours
• in theory, any colour can be produced as mixture
  of 3 primary colours
• primary colours independent colours
• independence one primary cannot be produced
  through linear combination of others
• pixel colour can be expressed as 3 numbers
• each number represents required amount of primary
  colour
• additive / subtractive combination RGB / CMY

7
Light Sensor

• Camera or scanner contains light sensor
• nature of sensor material determines response
  range visible light, infra-red light, ...
• Light sensor may be
• vacuum-tube device e.g. Vidicon
• solid-state device e.g. charge-coupled device
  (CCD), Complementary Metal-Oxide Semiconductor
  (CMOS)
• more common than vacuum tube
• due to long life, robustness, low power
  consumption, geometric stability, ...

8
Light Sensor

• Sensor is made up of light sensitive cells
  (photo-sites)
• Cell converts photons into electric charge
  (electrons)
• Sensor structure
• two-dimensional array of cells
• one-dimensional array of cells
• e.g. line-scan camera
• (for cameras) 1D array less common than 2D array
  
• for 2D imaging second dimension obtained via
  motion of object or sensor array

9
Light Sensor

• Colour cameras
• Single-chip camera
• one light-sensing chip, with stripe or mosaic
  colour filter for R, G, B
• poor colour fidelity
• Three-chip camera
• one light-sensing and filter chip for each colour
  plane
• high colour fidelity (true colour)

10
Light Sensor

• Colour cameras

11
Capture Card

• Two main types analogue or digital input
• Common components
• camera / computer interface
• digitiser analogue-to-digital conversion
• frame buffer a.k.a frame store
• I/O-mapped or memory-mapped to host computer
• codec compressor / decompressor

12
Capture Card

• Simplified block diagram of analogue frame grabber

13
Colour Spaces

• Colour space / model coordinate system for
  quantifying colour
• tristimulus colour theory
• 3-dimensional space is required
• Many possible colour spaces
• RGB, CMY, HSV, YUV, CIE Luv, CIE Lab, ...
• Chosen colour space must meet quality
  requirements of given imaging task

14
Colour Spaces

• Humans perceive colour as comprising two main
  components
• luminance (brightness), and chrominance (hue and
  saturation)
• brightness relates to amount of reflected /
  emitted light
• hue determines colour (e.g. red, purple, green,
  )
• relates to wavelength of light
• saturation relates to colour purity

15
Colour Spaces

• Colour spaces built around luminance and
  chrominance are often used in image processing
• Examples
• HSV (chrominance Hue, Saturation luminance
  Value)
• HSI (chrominance Hue, Saturation luminance
  Intensity)
• YUV (chrominance and some luminance U and V
  luminance Y)

16
Colour Spaces

• RGB colour model

17
Colour Spaces

• HSV colour model

HSV or HSB (Hue, Saturation, Value / Brightness)
18
Colour Spaces

• Colour-space transformation is often needed prior
  to further processing, e.g.
• RGB normalisation
• RGB to YUV transformation
• RGB to HSV transformation

19
Colour Spaces

• Colour space transformation
• RGB to rgb RGB normalisation (against uniform
  luminance scaling)

20
Colour Spaces

• Colour space transformation
• RGB / YUV conversion

Y 0.299 R 0.587 G 0.114 B U 0.493 (B - Y)
- 0.147 R - 0.289 G 0.436 B V 0.877 (R - Y)
0.615 R - 0.515 G - 0.100 B R Y 0.000 U
1.140 V G Y - 0.396 U - 0.581 V B Y 2.029
U 0.000 V
21
Image File Formats

• File format description of arrangement of file
  content
• image size (height, width, depth)
• colour table
• data compression
• error detection
• encryption
• interlaced/non-interlaced
• user-specific information, ...
• Chosen file format must meet quality requirements
  of given imaging task

22
Image File Formats
Example BMP file layout
23
Image Quality Issues

• Some imaging factors that govern image quality
• spatial resolution
• pixel depth
• colour channel registration
• colour balance
• electronic noise
• aspect ratio
• compression
• lens
• ...

24
Image Quality Issues

• Spatial resolution
• Spatial resolution represents ability of device
  to distinguish fine spatial detail (e.g. resolve
  lines)
• relates to number of pixels per unit area of
  surface being imaged
• terms used to express resolution dots-per-inch
  (dpi) or pixels-per-inch (ppi), number of pixels
  along rows and columns of sensed image
• Image resolution determines sharpness or detail
  of images

25
Image Quality Issues
Spatial resolution
Digital camera resolution chart used for testing
camera resolution
http//www.sinepatterns.com/i_Stdrds.htm
26
Image Quality Issues

• Spatial resolution
• Artefacts associated with resolution aliasing
• in still images jagged (jaggies) blocky /
  pixelated, or Moiré patterns
• in moving images wagon-wheel effect, pixel
  popping (or crawlies)

http//www.photocanvasprinting.com/images/face1.jp
g http//www.photocanvasprinting.com/images/face2
.jpg
http//www.library.cornell.edu/preservation/tutori
al/conversion/conversion-04.html
(also see lecture on Lens Distortion Effects
for other illustrations of aliasing)
27
Image Quality Issues

• Spatial resolution
• Two types of resolution optical and interpolated
• optical resolution of camera or scanner absolute
  number of image photo-sites (pixels) on sensor
• interpolated resolution resolution achieved
  through interpolation
• additional pixels introduced into image in order
  to exceed optical resolution
• colour of new pixel estimated (typically by
  weighted averaging) from surrounding pixels
• NOTE interpolation does not increase amount of
  information (details) present in image, it merely
  makes image larger

28
Image Quality Issues

• Pixel depth
• Pixel depth determines number of tones
  (gray-scale or colour) that can be distinguished
  in image
• Maximum number of tones that can be represented
  with x bits per pixel is 2x tones
• dynamic range expresses full range of tonal
  variations from lightest to darkest

29
Image Quality Issues

• Pixel depth
• Artefact associated with pixel depth visible
  tonal steps (banding or posterization)
• most noticeable for smooth tonal gradients
• histogram reveals banding / bunching through
  appearance of more peaks (modes)

30
Image Quality Issues

• Colour channel registration
• pixel positions misaligned across separate colour
  components of image introduce colour artefacts
  e.g. colour fringing at object edges
• Colour balance
• colour digital images produced through
  combination of correct proportions of primary
  colours or other components
• incorrect proportions introduce unexpected
  colours in image

31
Image Quality Issues

• Electronic noise typically visible as speckles
• due to electronic devices used in image capture
  hardware
• e.g. dark current noise
• Aspect ratio ratio of image width to height
• different image sensors / output devices may have
  different aspect ratios
• aspect ratio may affect shapes and relative
  height and width (dimensions) of details shown in
  image
• Compression
• high lossy compression often introduces
  artefacts
• e.g. reduced sharpness, unnatural colour
  patterns, blockiness

32
Image Quality Issues

• Compression artefacts

33
Image Quality Issues

• Lens distortion
• Geometric distortion may occur when using wide
  angle, telephoto or zoom lenses

34
Image Quality Issues

• Some display device and file format issues
• Colour gamut and colour palette impose limited
  set of colours
• may result in change of some pixels to colours
  available in gamut or palette
• Colour gamut discrete set of colours that output
  device (such as computer monitor or printer) can
  reproduce
• Colour palette discrete set of defined colours
  used , for example, by 8-bit or 256-colour image
  files, or by some graphics adapter display modes
• Chosen file format must meet requirements of
  given imaging task
• e.g. adequate spatial resolution, bit depth,
  compression, metadata, ...

35
Image Quality Issues
Colour palette for 8-bit colour graphics mode
36
Summary

• Image quality is affected by each stage of
  imaging chain
• Nature and quality of image sensor may introduce
  image artefacts
• Chosen colour space and file format must meet
  quality requirements of given imaging task
• Some imaging factors that govern image quality
  spatial resolution, pixel depth, colour channel
  registration, colour balance, electronic noise,
  aspect ratio, compression, lens, ...