Topic 4 - Image Mapping - I

1
Topic 4 - Image Mapping - I
Department of Physics and Astronomy
DIGITAL IMAGING Course 3624
Professor Bob Warwick
2
Typical Image Processing Steps
ORIGINAL IMAGE
PRE-PROCESSING STEPS
ENHANCEMENT RESTORATION
IMPROVED IMAGE
IMAGE ANALYSIS
3
Image Mapping Processes
Image Mapping encompasses a range of enhancement
methods which adjust the way the image data are
displayed (ie how the data are "mapped" onto the
display device).
4
Histograms of a Colour Image
5
The Form of the Image Histogram
The form of the image histogram P(f) provides
useful information on the content/quality of the
image
P(f) ?
P(f) ?
P(f) ?
f ?
f ?
f ?
Good contrast Poor contrast
Saturated?
Image histogram modification techniques aim to
improve the gray level distribution in the
displayed image so as to make as much use as
possible of the rather limited ability of the eye
to discern gray shades.
6
Discriminating between Gray Levels - I
I Intensity of Scene
7
Discriminating between Gray Levels - II
Typically we are able to discern 32 25 gray
levels in any particular image
8
Discriminating between Gray Levels - III
Small squares have different intensity but same
apparent brightness.
Small squares have same intensity but different
apparent brightness.
9
Image Enhancement by Histogram Modification
Original Image
New" image
(Inefficient) Implementation Method Once fout
T(fin) has been defined, we compute a new image
by fin ? fout on a pixel-by-pixel basis 15 20
12 25 30 16 15 22
? 25 32

10
Forms of T(f) A Linear Contrast Stretch

• The parameters of the process f1 f2 might be
  determined
• Interactively
• Automatically

11
Example of Contrast Stretching
12
Improved Contrast?
13
Forms of T(f) Increased Gamma
14
Forms of T(f) Decreased Gamma
15
4.2 Image Enhancement by Histogram Matching
The objective is to set up the displayed image so
that its histogram has a specified form.
16
Histogram Equalisation Problem
Note that the result is only a crude
approximation to the target uniform distribution
due to the very coarse digitization of the
input image data
17
Comments on Implementation
Highly Efficient Method Load the look-up table
of the display device with the required
transformation
18
Histogram Equalisation in Action
Original Image
Original Histogram
Final Image
Equalised Histogram
19
Histogram Equalisation in Action
Original Image
Final Image
Equalised Histogram
Original Histogram
20
The General Case
The general formula above can be applied to give
any form for the output image histogram. The
procedure to apply this formula is
Equalization General
f
f

• A practical implementation might involve
• For each fin calculate C1(fin)
• Compute a look-up table of fout versus C2(fout)
• For each fin find the nearest C2 value to C1(fin)
• Determine the fout value the C2 value
• Load the resulting mapping fin ? fout into the
  display device look-up table

21
Image Enhancement by Histogram Specification
22
Example Histogram Specification
Image P(f)
f
23
Histogram to be matched taken from a second image
Target P(f)
f
24
Histogram Matching Example
Image CDF
Target CDF