IP 001

Image Processing Operations

Point Operation

$b[m, n] = f(a[m, n])$

• It only depends on the value of the pixel itself, not on the values of its neighbors.
• e.g. current pixel + 20.
• to increase the brightness of an image, adjust contrast, or apply a threshold to create a binary image.

Local Operation

$b[m, n] = f(a[m - 1, n - 1], a[m - 1, n], a[m - 1, n + 1], a[m, n - 1], a[m, n], a[m, n + 1], a[m + 1, n - 1], a[m + 1, n], a[m + 1, n + 1])$

• It depends on the values of the pixel and its neighbors.
• e.g. current pixel + average of 8 neighbors.
• to blur an image, sharpen an image, detect edges, or convolution with a kernel.
• The most common type of neighborhoods are:
  • 4-neighbors:
    • top, bottom, left, right.
  • 8-neighbors:
    • top, bottom, left, right, and the 4 diagonal neighbors.

Global Operation

$b[m, n] = f(a[0, 0], a[0, 1], ..., a[M - 1, N - 1])$

• It depends on the values of all pixels in the image.
• e.g. current pixel + average of all pixels in the image.
• to compute the histogram equalization, apply a global threshold, or perform a Fourier transform.

Image Histogram

• It is a graph showing how many pixels in an image have each possible intensity value.
  • Intensity value: the brightness of a pixel.
• e.g. 8-bit grayscale image has 256 possible intensity values (0-255).
  • The histogram will graphically display 256 numbers showing the distribution of pixels among those gray-scale values.

Histogram Equalization

0                     255
|------████████--------|
       80~140에 몰림

0                     255
|--██--██--██--██--██--|

• It spreads out the intensity values that are concentrated in a narrow range, increasing the contrast of the image.
• It is useful when the images have been acquired under poor lighting conditions or have low contrast (different circumstances).

Noise

• Any undesired information that contaminatest the image.
• During the analog-to-digital conversion process, it is a side effect of the physical conversion of patterns of light energy into electrical patterns.
• The shape of distribution of noise types used to describe many of them and is related closely to the histogram.

frequency
 ^
 |             █
 |          █████
 |       █████████
 |    █████████████
 +----------------------> noise gray level
        -20  0  +20

• Gaussian noise: the most common type of noise, with a bell-shaped distribution.

frequency
 ^
 | █                       █
 | █                       █
 | █                       █
 +----------------------------> gray level
   0                         255

• Salt-and-pepper noise: the most common type of impulse noise, with a distribution that has two spikes at the minimum and maximum intensity values.