EE5623 Lecture on Basic Image Processing Concepts

Advanced Digital Image Processing EE5623
Lecture 2:
Basic relationships between pixels

and pixel operations
2/16/2016 Capital University of Science and Technology Advanced Digital Image Processing EE5623 1

Arrangement of pixels: 0 1 1
0 1 0
0 0 1
4 neighbours N4(p): 1
0 1 0
0
Diagonal neighbours ND(p): 0 1

1
0 1
8 neighbours N8 (p) = ND(p) U N4(p) : 0 1 1

0 1 0
1
Basic relationships between Pixels
 Connectivity between pixels:
An important concept used in establishing boundaries of objects
and components of regions
Two pixels p and q are connected if
– They are adjacent in some sense
– If their gray levels satisfy a specified criterion of similarity
V: Set of gray level values used to define the criterion of

similarity

Path
Let coordinates of pixel p: (x, y), and of pixel q: (s, t)
A path from p to q is a sequence of distinct pixels with
coordinates: (x0, y0), (x1, y1), ......, (xn, yn) where
(x0, y0) = (x, y) & (xn, yn) = (s, t),
and (xi, yi) is adjacent to (xi-1, yi-1) 1 i  n
Regions
A set of pixels in an image where all component pixels are
connected
Boundary of a region
A set of pixels of a region R that have one of more neighbors
that are not in R
2
Distance Measures
Given coordinates of pixels p, q, and z: (x,y), (s,t), and (u,v)
Euclidean distance between p and q:
De ( p, q)  ( x  s) 2  ( y  t ) 2
– De distance  r from (x,y) define a disk of radius r centered at (x,y)
City-block distance between p and q:
D4 ( p, q)  x  s  y  t
– The pixels with D4 distance  r from (x,y) form a diamond centered at (x,y)
– the pixels with D4=1 are the 4-neighbors of (x,y)
Chessboard distance between p and q:
D8 ( p, q)  max(| x  s |, | y  t |)
– The pixels with D8 distance  r from (x,y) form a square centered at (x,y)
– The pixels with D8=1 are the 8-neighbors of (x,y)
Image Enhancement
Process an image to make the result more suitable than the original
image for a specific application
– Image enhancement is subjective (problem/application
oriented)
Image enhancement methods
Spatial domain: Direct manipulation of pixel in an image(on the
image plane)
Frequency domain: Processing the image based on modifying the
Fourier transform of an image
Many techniques are based on various combinations of methods from

these two categories
3
Image Enhancement
Types of image enhancement operations
Point/pixel operations Output value at specific coordinates (x,y) is
dependent only on the input value at (x,y)
Local operations The output value at (x,y) is dependent on the
input values in the neighborhood of (x,y)
Global operations The output value at (x,y) is dependent on all
the values in the input image
Basic concepts
Spatial domain enhancement methods can be generalized as
g(x,y)=T[f(x,y)]
f(x,y) : input image
g(x,y): processed (output) image
T[*] : an operator on f (or a set of input images),
defined over neighborhood of (x,y)
Neighborhood about (x,y): a square or rectangular sub-

image area centered at (x,y)
4
Basic Concepts
3x3 neighborhood about (x,y)
Basic concepts
Pixel/point operation:
Neighborhood of size 1x1: g depends only on f at (x,y)
T: a gray-level/intensity transformation/mapping function
Let r = f(x,y) s = g(x,y)
r and s represent gray levels of f and g at (x,y)
Then s = T(r)
Local operations:
g depends on the predefined number of neighbors of f at (x,y)
Implemented by using mask processing or filtering
Masks (filters, windows, kernels, templates) :
a small (e.g. 3×3) 2-D array, in which the values of the
coefficients determine the nature of the process
5
Common pixel operations
 Image negatives
 Log transformations
 Power-law
transformations
Image negatives
 Reverses the gray level order
 For L gray levels the transformation function is
s =T(r) = (L-1)-r
Input image (X-ray image) Output image (negative)

6
Image negatives
Application: To enhance the visibility for images with more
dark portion
Original digital mammogram Output image

Image scaling
s =T(r) = a.r (a is a constant)
7
Log transformations
Function of s = c Log(1+r)
Log transformations
Properties of log transformations
– For lower amplitudes of input image the range of gray levels is
expanded
– For higher amplitudes of input image the range of gray levels is
compressed
Application:
– Dynamic range of a processed image far exceeds the capability
of the display device
• (e.g. display of the Fourier spectrum of an image)
– Also called “dynamic-range compression / expansion”
8
Log transformations
Fourier spectrum with values of The result applying log transformation,

range 0 to 1.5 x 106 scaled linearly c=1
Power-law Transformation
Basic form:
s = crg ,
where c & g
are positive
Plots of equation
s = crg,
For various values of g
(c  1)
9
For γ < 1: Expands values of dark pixels, compress values of
brighter pixels
For γ > 1: Compresses values of dark pixels, expand values of
brighter pixels
If γ=1 & c=1: Identity transformation (s = r)
A variety of devices (image capture, printing, display) respond

according to a power law and need to be corrected;
Gamma (γ) correction

The process used to correct the power-law response phenomena
 Example of gamma correction
 To linearize the CRT response a pre-distortion circuit is

needed s = cr1/g
10
Gamma correction
Linear Response of
wedge gray CRT to Linear
scale image wedge
Gamma Output of
corrected monitor
wedge
Power-law Transformation: Example
MRI image of Result of applying Result of applying Result of applying

fractured human power-law power-law power-law
spine transformation transformation transformation
c = 1, g  0.6 c = 1, g  0.4 c = 1, g  0.3
11
Power-law Transformation: Example
Original Result of applying

satellite power-law
image transformation
c = 1, g  3.0
Result of Result of applying

applying power-law
power-law transformation
transformation
c = 1, g  5.0
c = 1, g  4.0
Piecewise-linear transformation
Contrast stretching
Goal:
Increase the dynamic range of the gray levels for low contrast
images
Low-contrast images can result from

– poor illumination
– lack of dynamic range in the imaging sensor
– wrong setting of a lens aperture during image acquisition
12
Piecewise-linear transformation: contrast stretching
Method
where a1, a2, and a3 control the result of contrast stretching

if a1 = a2 = a3 = 1 no change in gray levels
if a1 = a3 = 0 and r1 = r2, T(*) is a thresholding function,
the result is a binary image
Contrast Stretching Example
Form of
Transformation Original low-
function contrast image
Result of Result of
contrast thresholding
stretching
13
Contrast stretching through histogram
If rmax and rmin are the maximum and minimum gray level of the input
image and L is the total gray levels of output image
The transformation function for contrast stretching will be
 
s  T (r )  (r  rmin )
L

 rmax  rmin 
rmin rmax
0 L-1
Histogram equalization: continuous case

 Idea: To find a non-linear transformation
s = T (r)
to be applied to each pixel of the input image f(x,y), such that a
uniform distribution of gray levels in the entire range results for
the output image g(x,y).
 Assuming ideal, continuous case, with normalized histograms

– that 0  r  1 and 0  s 1
– T(r) is single valued i.e., there exists r= T-1(r)
– T(r) is monotonically increasing
14
Single valued monotonically increasing function
A function T(r) is monotonically increasing
if T(r1) < T(r2) for r1 < r2,
and monotonically decreasing
if T(r1) > T(r2) for r1 < r2.
Example of a transformation function

which is both single valued and
monotonically increasing
Background (probability distribution)

Assume continuous random variables
The cumulative probability distribution function or cumulative
distribution function (cdf)
The probability that the random variable is less than or equal to a
specified constant a. We write this as
for all values of a (i.e.,   < a < ),
The probability density function (pdf) or density function of random

variable x is defined as the derivative of the cdf:
15
Histogram equalization: continuous case
 Fr(r) and Fs(s) : cdfs of original and transformed gray levels r and s.
 pr(r) and ps(s) : pdfs of original and transformed gray levels r and s.
For strictly monotonically increasing transformation function
Fs(s) = Fr(r) or ps(s) ds = pr(r) dr
Goal of histogram equalization:
Gray levels are uniformly distributed
i.e. pdf ps(s) = 1 over the range 0  s  1
 dr  ds dT (r )
ps ( s)  pr (r )   1 or pr (r )  
 ds  dr dr
r
 s  T (r )   pr ( )d
0
Histogram equalization
If the following transformation function is used
r
s  T (r )   pr ( )d for 0  r 1
0
Then the pdf ps(s) = 1 over the range 0  s  1
In words
If we select T(r) as the cumulative distribution of r
Then the output image will have a uniform pdf of gray levels
Now Consider
1. a digital (gray level) case
2. the gray levels 0  r  L  1
16
Histogram equalization: Discrete case
The discrete approximation of the transformation
function for histogram equalization is:
k
sk  T (rk )   pr (rj ) for 0  k  L  1
j 0
nj L 1
pr ( r j )  , j  0,........, L  1 and n  nj
where n j 0
nj : number of pixels with gray level rj

n : total number of pixels
Note: For digital images, gray-level pdf cannot be
exactly uniform after histogram equalization
Histogram equalization examples

Input image Output image
Input histogram and cdf Output histogram and cdf

17
Low contrast image Output image
Equalized histogram
Equalized histogram
high contrast image Output image

Dark input image Output image
Equalized histogram
Equalized histogram
Bright input image Output image
18
Transformation functions for

histogram equalization
19
20
21

EE5623 Lecture on Basic Image Processing Concepts

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

EE5623 Lecture on Basic Image Processing Concepts

Uploaded by

Copyright:

Available Formats

Advanced Digital Image Processing EE5623

Basic relationships between pixels

Basic relationships between pixels

Diagonal neighbours ND(p): 0 1

8 neighbours N8 (p) = ND(p) U N4(p) : 0 1 1

V: Set of gray level values used to define the criterion of

Basic relationships between pixels

Many techniques are based on various combinations of methods from

Neighborhood about (x,y): a square or rectangular sub-

3x3 neighborhood about (x,y)

Input image (X-ray image) Output image (negative)

Original digital mammogram Output image

Fourier spectrum with values of The result applying log transformation,

A variety of devices (image capture, printing, display) respond

Gamma (γ) correction

 To linearize the CRT response a pre-distortion circuit is

Power-law Transformation: Example

MRI image of Result of applying Result of applying Result of applying

Original Result of applying

Result of Result of applying

Low-contrast images can result from

where a1, a2, and a3 control the result of contrast stretching

Contrast Stretching Example

Histogram equalization: continuous case

 Assuming ideal, continuous case, with normalized histograms

Example of a transformation function

Background (probability distribution)

for all values of a (i.e.,   < a < ),

The probability density function (pdf) or density function of random

nj : number of pixels with gray level rj

Histogram equalization examples

Input histogram and cdf Output histogram and cdf

Histogram equalization examples

Transformation functions for

Histogram equalization examples

Histogram equalization examples

You might also like