1. Digital Image Processing
ECE 448: Spring 2014
Lab 5
Digital Image Processing

2. Agenda for today Part 1: Midterm Exam Grades & Discussion
Part 2: Introduction to Lab5 2

3. Digital Image
The digital image is picture information in digital form consisting of pixels.
The position of each pixel is specified in terms of an index for the number of columns and another for the number of rows.
p(2, 8) = 86
m = 2
n = 8

4. Image Quality More is better!
The image quality mostly depends on
1. Spatial Resolution: The number of pixels in the representation of a digital image.
Typically represented as
width x height
As high as 1600 x 1200, or
As low as 320 x 200
100 x 200
25 x 50
Digital: Computer manipulation:
Image as a data matrix
More is better!

5. Image Quality Color Depth:
The color depth determines how many different colors each pixel can display
Bits Per pixel
Possible Colors
Common Name 1 2
Monochrome 4
CGA 16
EGA 8
256
VGA
65536
XGA, High Color 24
SVGA, True Color

6. 1-bit
2-bit
4-bit
8-bit
24-bit

7. 8-bit Gray Level Images
Each pixel is encoded on a gray scale from 0 – 255, where 0 is black and 255 is white.
All numbers in between represent levels of the grayscale

8. 24-bit Color Images Each pixel has three components
Red
Green
Blue
Each component is encoded in 8-bits
24 bit Color
8-bits
8-bits
8-bits
24 bit Image

9. 8-bit Indexed Color Images
Each pixel value is a color index that points to a color lookup table.
Color Map
8-bit indexed color Image

10. 8-bit Indexed Color Images

11. RGB to Grayscale Conversion
Conversion from RGB space to YIQ space required to generate the histogram.
Y : luminance channel (represents the light intensity)
I : In-space channel
Q: Quadrature chrominance channel
Both I and Q channels represent color details.
Y(m, n) carries grayscale information with most of the signal energy

12. Grayscale to RGB Conversion
The following transformation is used to convert back to RGB space.

13. Grayscale Histogram A graph to show how many pixels are at each level.
The histogram contains information needed for image equalization, where the image pixels are stretched to give a reasonable contrast.
Image encoded with 8 bits per pixel, the pixel value ranges from 0 to 255

14. Histogram Generation for different Image Types
For grayscale image:
Use the grayscale values to generate the histogram
For RGB color image:
Convert RGB to YIQ
Use Y channel to generate the histogram
For indexed color image:
Convert the color map from RGB to YIQ

15. Equalization of the Histogram data
Image pixels are stretched to give a reasonable contrast in the equalized image
The accumulative counts can range up to the total number of pixles, so the equalized pixel value can vary from 0 to maximum scale level.

16. RGB Image Equalization Flow

17. Original & Equalized Image