1. The Digital Image
Dr. John Ryan

3. What would this look like in grayscale?

4. grayscale

13. 50 43 48
134
124
100
250
234
187

14. Is this the lowest level we can go?

15. 15

17. Why process images?
Source: Hornegger & Paulus, Erlangen University

18. Line Defect Interpolation
Source: Hornegger & Paulus, Erlangen University

19. Contrast Enhancement
Source: Hornegger & Paulus, Erlangen University

20. Noise Reduction
Source: Hornegger & Paulus, Erlangen University

21. Edge Enhancement
The filter works by identifying sharp edge boundaries in the image, such as the edge between a subject and a background of a contrasting color, and increasing the image contrast in the area immediately around the edge
Source: Hornegger & Paulus, Erlangen University

22. The Difference?
Original
Result

23. What is a Digital Image?
The digital image is sampled and mapped as a grid of dots or picture elements (pixels). Each pixel is assigned a tonal value (black, white, shades of gray or color), which is represented in binary code (zeros and ones).

24. Digital radiography
is a form of X-ray imaging, where digital X-ray sensors are used instead of traditional photographic film. Advantages include time efficiency through bypassing chemical processing and the ability to digitally transfer and enhance images. Also less radiation can be used to produce an image of similar contrast to conventional radiography

25. What is a Pixel?
A pixel is the smallest element within an image that has a single intensity value
The pixel value varies depending on intensity resolution (range, depth)

26. Pixels

27. What is a Voxel? A voxel is like the 3D version of a pixel
It is the smallest unit within a volume

28. A Fly on the Ceiling Y X

29. Image Coordinate System
Top left corner is the origin.
Bottom right corner is the final pixel or (width, height).
Although this is the most common image coordinate system, it may vary.
There are many definitions for “pixel”, but in the context of medical imaging, a pixel is the smallest intensity or colour component of an image.

30. Bits and Bytes

32. How Computers Store Images
Uncompressed images are stored as a sequence of pixel values
From left to right, then down to next row
For 8 bit, 1 byte per pixel
Deep down it’s stored as binary information:

33. Bits and Bytes 8 bits = 1 byte (Total is 256)
A bit can have two values: 1 or 0, on or off 1 2 4 8 16 32 64
128
8 bits = 1 byte (Total is 256)
Computer data is stored in binary
Binary digits (bits)
Looks like ……

34. Getting a Value 128 64 32 16 8 4 2 1 Example 1:
= = 129
Example 2:
= = 18

35. Resolution Intensity resolution Spatial resolution Temporal resolution
The range from totally white to totally black.
Intensity – relating to brightness value
Spatial resolution
The width and height of the image.
Temporal resolution
The rate of frames per second of animation or video.
Temporal – relating to time

36. contrast Contrast is a measure of the
magnitude of the measured signal differences between physically different regions of the imaged object.

37. contrast
Two versions of a wrist MR image. Image a plainly has higher contrast than image
b – the bones are much brighter relative to the surrounding tissue background, even though in
image b the average brightness of the bones is greater than in image a. Measures of contrast are
based on the relative or absolute difference in average intensity of an object and its background

38. Artifacts
Any intensity or color fluctuations that make it difficult to see what you want to see that occur due to specific properties of the imaging method

39. Intensity Resolution Bits per pixel Or, number of gray levels
8 bpp [256; (256 colours)],
16 bpp [65536; (65,536 colours, known as Highcolour)],
24 bpp [ ; (16,777,216 colours, known as Truecolour)].
48 bpp [ ; (281,474,976,710,656 colors, used in many flatbed scanners and for professional work)

40. Intensity Resolution
3 bit, 8 gray levels
8 bit, 256 gray levels

41. Lets Build a Chest X-Ray From scratch, 0 bits – 0 levels

42. 1 bit, 2 levels

43. 2 bits, 4 levels

44. 3 bits, 8 levels

45. 4 bits, 16 levels

46. 5 bits, 32 levels

47. 6 bits, 64 levels

48. 7 bits, 128 levels

49. 8 bits, 256 levels

50. Intensity Resolution
3 bit, 8 gray levels
8 bit, 256 gray levels

51. Quantization
Quantization involves the downsizing of the number of gray levels
This allows us to compress the image (less number of bits)
However, there are pitfalls:
An effect called posterisation can be produced
Vital information may be omitted

52. Original Grayscale Image
Full range of intensity values.

53. Quantized to 8 Intensities / Shades
Intensity values limited to 8 shades.
Note the contours.

54. Quantized to Black and White
Intensity values limited to just 2 shades: black and white

55. Posterisation
Affects areas of low spatial frequency the most

56. Spatial Resolution The number of pixels in an image
WIDTH
The number of pixels in an image
Can be expressed by WIDTHxHEIGHT or actual value
Eg. 256x256 pixels
Or, pixels
HEIGHT

57. Spatial Resolution
800x800
50x50, Scaled Up

58. Temporal Resolution Frames per second For video or animation
Normally fps

59. Temporal Aliasing Not enough frames per second
This causes flickering or strobing of the video
Solution: sample at a higher rate or apply some image pre-processing techniques
An example of temporal aliasing would be the “wagon wheel” effect, where a wheel or helicopter rotor-blade appears to be slowing down or in reverse, due to a different sampling rate.

60. Histogram A Histogram is a distribution of intensity values.
pixels
Pixel Value
A Histogram is a distribution of intensity values.
Usually expressed as a graph.
X – axis: Pixel value
Y – axis: Number of pixels

61. What does this image look like?
Mean: Median: 207
Standard Deviation: 23
Pixels:
Depth: 256 (8bits)

62. Example 1

63. How about this? Mean: 39 Median: 35 Standard Deviation: 33
Pixels:
Depth: 256 (8bits)

64. Example 2

65. And this? Mean: 122 Median: 125 Standard Deviation: 71 Pixels: 640000
Depth: 256 (8bits)

66. Example 3

67. More Histograms A Histogram is a distribution of intensity values.
Sky and clouds
Shrubbery

68. Even More Histograms
Air
Brain
Bone

69. Luminance-based segmentation
By knowing about image statistics, we can do interesting things like segmentation of bone for 3D reconstruction.

71. 3D Reconstruction of Child’s Skull

72. Brightness (Level)
Original
Level Adjusted

73. Brightness (Level)
“Level” or “Brightness” is adjusted by adding or subtracting to the current pixel value.
This is applied evenly throughout the image.
If the limit of the intensity values are reached (i.e., totally black or totally white), the current pixel is assigned the same value as the limit.

74. Contrast
Original
Contrast Adjusted

75. Contrast AKA “Histogram stretching”
“Contrast” involves the stretching of an original narrow range of values to a wider grayscale range or vice-versa
i.e., Expanding from a “clumped” histogram to a more well spaced distribution.

76. References http://grail.cs.washington.edu/projects/dance-symmetry/
Previous image analysis lectures by Dr. Hamish Carr
Digital Image Processing by Gregory A Baxes.
Hornegger & Paulus, Erlangen University