Scribd
Upload
44 views13 pages

Histogram Equalisation The Algorithm

The document describes the steps to equalize the histogram of an image. It involves: 1) Calculating the cumulative frequency distribution of pixel intensities in the original image. 2) Designing an intensity mapping that makes the cumulative frequency distribution a straight ramp, to match an ideal equalized histogram. 3) Mapping each pixel intensity in the original image to a new intensity in the transformed image to equalize the histogram.

Uploaded by

Amr Yassin
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
44 views13 pages

Histogram Equalisation The Algorithm

The document describes the steps to equalize the histogram of an image. It involves: 1) Calculating the cumulative frequency distribution of pixel intensities in the original image. 2) Designing an intensity mapping that makes the cumulative frequency distribution a straight ramp, to match an ideal equalized histogram. 3) Mapping each pixel intensity in the original image to a new intensity in the transformed image to equalize the histogram.

Uploaded by

Amr Yassin
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 13

Histogram Equalisation

The algorithm
• Given an image as below, derive the intensity mapping that will as best
as possible equalise the image histogram. The image histogram is as
shown on the right. There are 8 possible grey scale levels from 0 to 7.

4 4 4 4 4
3 4 5 4 3 I 0 1 2 3 4 5 6 7
3 5 5 5 3 f(I) 0 0 0 6 14 5 0 0
3 4 5 4 3
4 4 4 4 4
Step 1: calculate the cumulative
frequency distribution

I 0 1 2 3 4 5 6 7
f(I) 0 0 0 6 14 5 0 0
Cuf 0 0 0 6 20 25 25 25

Idea is to derive an intensity mapping that will make the CuF


turn into
a straight ramp.
Step 2: Compare with the CuF of
an equalised histogram

I 0 1 2 3 4 5 6 7
f(I) 0 0 0 6 14 5 0 0
Cuf 0 0 0 6 20 25 25 25
Feq 3 3 3 4 3 3 3 3

In this case the Equalised (ideal) histogram needs to have 25/8


pels in each bin = 3.125 but only integer numbers of pels are
possible (its frequency after all). So I’ve added in one to make
it ok.
Step 2: Compare with the CuF of
an equalised histogram

I 0 1 2 3 4 5 6 7
f(I) 0 0 0 6 14 5 0 0
Cuf 0 0 0 6 20 25 25 25
Feq 3 3 3 4 3 3 3 3
CuFeq 3 6 9 13 16 19 22 25
Step 2: Design the mapping

I 0 1 2 3 4 5 6 7
f(I) 0 0 0 6 14 5 0 0
Cuf 0 0 0 6 20 25 25 25
Feq 3 3 3 4 3 3 3 3
CuFeq 3 6 9 13 16 19 22 25
For each intensity in the original image, find an intensity in the
transformed image that has as close as possible, the same amount
of Cumulative frequency.
Step 2: Design the mapping

I 0 1 2 3 4 5 6 7
f(I) 0 0 0 6 14 5 0 0
Cuf 0 0 0 6 20 25 0 0
Feq 3 3 3 4 3 3 3 3
CuFeq 3 6 9 13 16 19 22 25
So for intensity 0, this has Cuf 0, and the closest in the xformed image
Is a CuFreq of 3. So Intensity 0 in input maps to intensity 0 in output
Step 2: Design the mapping

I 0 1 2 3 4 5 6 7
f(I) 0 0 0 6 14 5 0 0
Cuf 0 0 0 6 20 25 25 25
Feq 3 3 3 4 3 3 3 3
CuFeq 3 6 9 13 16 19 22 25
Intensity 1 in input maps to Intensity 0 in output.
[Sometimes you can decide not to map into previously used
intensities]
Step 2: Design the mapping
Input I Output I
I 0 1 2 3 4 5 6 7 0 0
f(I) 1 0
0 0 0 6 14 5 0 0
2 0
Cuf 0 0 0 6 20 25 25 25
3 1
Feq 3 3 3 4 3 3 3 3

CuFe 3 6 9 13 16 19 22 25
q

For each intensity in the original image, find an intensity in the


transformed image that has as close as possible, the same amount
of Cumulative frequency.
Step 2: Design the mapping
Input I Output I
I 0 1 2 3 4 5 6 7 0 0
f(I) 1 0
0 0 0 6 14 5 0 0
2 0
Cuf 0 0 0 6 20 25 25 25
3 1
Feq 3 3 3 4 3 3 3 3 4 5
CuFe 3 6 9 13 16 19 22 25
q

4 -> 5
Step 2: Design the mapping
Input I Output I
I 0 1 2 3 4 5 6 7 0 0
f(I) 1 0
0 0 0 6 14 5 0 0
2 0
Cuf 0 0 0 6 20 25 25 25
3 1
Feq 3 3 3 4 3 3 3 3 4 5
CuFe 3 6 9 13 16 19 22 25 5 7
q

5 ->7
Step 2: Design the mapping
Input I Output I
I 0 1 2 3 4 5 6 7 0 0
f(I) 1 0
0 0 0 6 14 5 0 0
2 0
Cuf 0 0 0 6 20 25 25 25
3 1
Feq 3 3 3 4 3 3 3 3 4 5
CuFe 3 6 9 13 16 19 22 25 5 7
q
6 7
7 7

And so on …
The new histogram
Input I Output I
I 0 1 2 3 4 5 6 7 0 0
f(I) 1 0
0 0 0 6 14 5 0 0
2 0
Cuf 0 0 0 6 20 25 25 25
3 1
Feq 3 3 3 4 3 3 3 3 4 5
CuFeq 3 6 9 13 16 19 22 25 5 7
6 7
F(EQ) 0 6 0 0 0 14 0 5
7 7
Cu(EQ) 0 6 6 6 6 20 20 25
Output histogram only uses
4 bins!
The result
Input I Output I
4 4 4 4 4 0 0 5 5 5 5 5
3 4 5 4 3 1 0
1 5 7 5 1
2 0
3 5 5 5 3 1 7 7 7 1
3 1
3 4 5 4 3 4 5 1 5 7 5 1
4 4 4 4 4 5 7 5 5 5 5 5
6 7
Original 7 7 Equalised

You might also like