Thi Giac May Tinh Vo Quang Hoang Khang Dip03 Image Enhancement Historgram Equalization (Cuuduongthancong - Com)

Lecture 03
om
.c
Image Enhancement
ng
co
Histogram Equalization
an
th
o ng
du
u
cu
CuuDuongThanCong.com https://fb.com/tailieudientucntt
Contents
In this lecture we will look at image enhancement point
om
processing techniques:
.c
 What is histogram?
ng
 Histogram equalization
co
 Histogram matching
an
th
o ng
du
u
cu
Image Enhancement-Histogram Equalization 2

What is the histogram of a digital image?
 The histogram of a digital image with gray values
om
r0 , r1 ,, rL1 is the discrete function
.c
ng
nk
p(rk ) 
co
n
an
nk: Number of pixels with gray value rk
th
ng
n: total Number of pixels in the image
o
du
u
 The function p(rk) represents the fraction of the total

cu
number of pixels with gray value rk.

What is the histogram of a digital image?
 Histogram provides a global description of the
om
appearance of the image.
.c
ng
 If we consider the gray values in the image as
co
realizations of a random variable R, with some
an
probability density, histogram provides an
th
approximation to this probability density. In other
ng
words,
o
du
Pr( R  rk )  p(rk )
u
cu

Some typical histograms
 The shape of a histogram provides useful information
om
for contrast enhancement
.c
ng
co
an
th
ng
Dark image
o
du
u
cu

om
Bright image
.c
ng
co
an
th
o ng
du
Low contrast
u
image
cu

om
.c
ng
High contrast image
co
an
th
o ng
du
u
cu

Histogram equalization
 The histogram equalization is an approach to enhance
om
a given image. The approach is to design a
.c
transformation T(.) such that the gray values in the
output is uniformly distributed in [0, 1].
ng
co
 Let us assume for the moment that the input image to
an
be enhanced has continuous gray values, with r = 0
th
ng
representing black and r = 1 representing white.
o
du
 We need to design a gray value transformation s =

u
cu
T(r), based on the histogram of the input image, which

will enhance the image.

we assume that:
om
 T(r) is a monotonically
.c
increasing function for
0 r 1 (preserves
ng
co
order from black to
an
white).
th
 T(r) maps [0,1] into ng
[0,1] (preserves the
o
du
range of allowed Gray

u
values).
cu

 Let us denote the inverse transformation by r T -1(s) .
om
We assume that the inverse transformation also
.c
satisfies the above two conditions.
ng
 We consider the gray values in the input image and
co
output image as random variables in the interval [0, 1].
an
th
ng
 Let pin(r) and pout(s) denote the probability density of
o
the Gray values in the input and output images.
du
u
cu

 If pin(r) and T(r) are known, and r T -1(s) satisfies
om
condition 1, we can write (result from probability
.c
theory): dr  
pout ( s)   pin (r ) 
ng
 ds  r T 1 ( s )
co
 One way to enhance the image is to design a
an
transformation T(.) such that the gray values in the
th
output is uniformly distributed in [0, 1], i.e. pout (s)  1,
ng
0  s 1
o
du
 In terms of histograms, the output image will have all

u
cu
gray values in “equal proportion” .

 This technique is called histogram equalization.

 Next we derive the gray values in the output is
om
uniformly distributed in [0, 1].
 Consider the transformation
.c
r
s  T (r )   pin ( w)dw ,
ng
0  r 1
co
0
 Note that this is the cumulative distribution function
an
(CDF) of pin (r) and satisfies the previous two
th
ng
conditions.
o
 From the previous equation and using the fundamental
du
theorem of calculus,
u
cu
ds
 pin (r )
dr

 Therefore, the output histogram is given by
om
 1 
pout ( s )   pin (r )   1r T 1 ( s )  1, 0  s 1
.c

 pin (r )  r T 1 ( s )
ng
co
 The output probability density function is uniform,
an
regardless of the input.
th
 Thus, using a transformation function equal to the CDF
ng
of input gray values r, we can obtain an image with
o
du
uniform gray values.

u
 This usually results in an enhanced image, with an

cu
increase in the dynamic range of pixel values.

How to implement histogram equalization
 Step 1:For images with discrete gray values, compute:
om
nk
pin (rk )  0  rk  1 0  k  L 1
.c
n
ng
L: Total number of gray levels
co
an
nk: Number of pixels with gray value rk
th
n: Total number of pixels in the image
ng
 Step 2: Based on CDF, compute the discrete version
o
du
of the previous transformation :

u
cu
k
s k  T (rk )   pin (r j ) 0  k  L 1
j 0

Example
 Consider an 8-level 64 x 64 image with gray values (0,
om
1, …, 7). The normalized gray values are (0, 1/7, 2/7,
…, 1). The normalized histogram is given below:
.c
ng
co
an
th
o ng
du
u
cu
 NB: The gray values in output are also (0, 1/7, 2/7, …, 1).

Example
om
.c
# pixels
ng
Fraction
co
of # pixels
an
th
o ng
du
Gray value Normalized gray value

u
cu

Example
k
 Applying the transformation,s k  T (rk )   pin (r j ) we have
j 0
om
.c
ng
co
an
th
o ng
du
u
cu

Example
 Notice that there are only five distinct gray levels ---
om
(1/7, 3/7, 5/7, 6/7, 1) in the output image. We will
relabel them as (s0, s1, …, s4 ).
.c
 With this transformation, the output image will have
ng
co
histogram
an
th
o ng
du
u
cu

Example
Histogram of output image
om
.c
# pixels
ng
co
an
th
o ng
du
Gray values
u
 Note that the histogram of output image is only

cu
approximately, and not exactly, uniform. This should not be

surprising, since there is no result that claims uniformity in
the discrete case.
Example Original image and its histogram
om
.c
ng
co
an
th
o ng
du
u
cu

Example
Histogram equalized image and its histogram
om
.c
ng
co
an
th
o ng
du
u
cu

Example
om
.c
ng
co
an
th
o ng
du
u
cu

Example
om
.c
ng
co
an
th
o ng
du
u
cu

Example
om
.c
ng
co
an
th
o ng
du
u
cu

Comments
 Histogram equalization may not always produce
om
desirable results, particularly if the given histogram is
.c
very narrow. It can produce false edges and regions. It
can also increase image “graininess” and “patchiness.”
ng
co
an
th
o ng
du
u
cu

Comments
om
.c
ng
co
an
th
o ng
du
u
cu

Histogram matching
 Histogram equalization yields an image whose pixels
om
are (in theory) uniformly distributed among all gray
.c
levels.
 Sometimes, this may not be desirable. Instead, we
ng
co
may want a transformation that yields an output image
an
with a pre-specified histogram. This technique is called
th
histogram specification. ng
o
z=H(r)
du
u
cu
Input Uniform Output

image s=T(r) image v=G(z) image

Histogram matching
 Suppose, the input image has probability density in p(r). We
want to find a transformation z H (r) , such that the
om
probability density of the new image obtained by this
.c
transformation is pout(z) , which is not necessarily uniform.
ng
 First apply the transformation
co
r
an
s  T (r )   pin ( w)dw , 0  r 1 (*)
th
0
ng
This gives an image with a uniform probability density.
o
 If the desired output image were available, then the

du
following transformation would generate an image with

u
cu
uniform density:
z
v  G ( z )   pout ( w)dw, 0  z 1 (**)
0
Histogram matching
 From the gray values nwe can obtain the gray values
z by using the inverse transformation, z  G-1(v)
om
.c
ng
 If instead of using the gray values nobtained from (**),
co
we use the gray values s obtained from (*) above (both
an
are uniformly distributed ! ), then the point
th
transformation ng
o
z=H(r)= G-1[ v=s =T(r)]
du
u
cu
will generate an image with the specified density out

p(z), from an input image with density in p(r) !

Histogram matching
 For discrete gray levels, we have
om
k
s k  T (rk )   pin (r j ) 0  k  L 1
.c
j 0
ng
k
vk  G ( z k )   pout ( z j )  sk 0  k  L 1
co
j 0
an
 If the transformation zk G(zk) is one-to-one, the
th
inverse transformation sk G-1 (sk) , can be easily
o ng
determined, since we are dealing with a small set of
du
discrete gray values.

u
cu
 In practice, this is not usually the case (i.e., ) zk 

G(zk) is not one-to-one) and we assign gray values to
match the given histogram, as closely as possible.
Algorithm for histogram matching
 Step 1: Equalize input image to get an image with
om
uniform gray values, using the discrete equation:
.c
k
s k  T (rk )   pin (r j ) 0  k  L 1
ng
j 0
co
 Step 2: Based on desired histogram to get an image
an
with uniform gray values, using the discrete equation:
th
k
vk  G ( z k )   pout ( z j )  sk
ng
0  k  L 1
o
j 0
du
 Step 3:
u
cu
1 1
  
z G ( v=s ) z G [T ( r )]

Example
 Consider an 8-level 64 x 64 previous image.
om
.c
ng
co
# pixels
an
th
o ng
du
u
cu
Gray value

Example
 It is desired to transform this image into a new image,
om
using a transformation Z=H(r)= G-1[T(r)], with
.c
histogram as specified below:
ng
co
an
th
# pixels
ng
o
du
u
cu
Gray values

Example
 Step 1: The transformation T(r) was obtained earlier
om
(reproduced below):
.c
ng
co
an
th
o ng
du
u
cu

Example
 Step 2: we compute the transformation G as before.
om
.c
ng
co
an
th
o ng
du
u
cu

Example
 Step 3: Computer z=G-1 (s=v),Notice that G is not
om
invertible.
.c
G-1(0) = ?
ng
G-1(1/7) = 3/7
co
G-1(2/7) = 4/7
an
G-1(3/7) = ?
th
ng
G-1(4/7) = ?
o
G-1(5/7) = 5/7
du
G-1(6/7) = 6/7
u
cu
G-1(1) = 1

Example
 Combining the two transformation T and G-1 , compute
om
z=H(r)= G-1[v=s=T(r)]
.c
ng
co
an
th
o ng
du
u
cu

Example
 Applying the transformation H to the original image
om
yields an image with histogram as below:
.c
ng
co
an
th
o ng
du
 Again, the actual histogram of the output image does

u
cu
not exactly but only approximately matches with the

specified histogram. This is because we are dealing
with discrete histograms.
Example
om
.c
ng
co
an
Original image and its histogram
th
o ng
du
u
cu
Histogram specified image and its histogram

Example
om
.c
Desired histogram
ng
co
an
th
o ng
du
u
cu

Example
om
.c
ng
co
an
th
o ng
du
u
cu

Example
om
.c
ng
co
an
th
o ng
du
u
cu

Example
om
.c
ng
co
an
th
o ng
du
u
cu

Example
om
.c
ng
co
an
th
o ng
du
u
cu

Example
om
.c
ng
co
an
th
o ng
du
u
cu

Summary
 We have looked at histogram equalization and
om
histogram matching
.c
ng
 Next time we will start to look at neighbourhood
co
operations – in particular filtering and convolution
an
th
o ng
du
u
cu


Thi Giac May Tinh Vo Quang Hoang Khang Dip03 Image Enhancement Historgram Equalization (Cuuduongthancong - Com)

Uploaded by

Copyright:

Available Formats

You might also like

Thi Giac May Tinh Vo Quang Hoang Khang Dip03 Image Enhancement Historgram Equalization (Cuuduongthancong - Com)

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

Thi Giac May Tinh Vo Quang Hoang Khang Dip03 Image Enhancement Historgram Equalization (Cuuduongthancong - Com)

Uploaded by

Copyright:

Available Formats

Lecture 03

In this lecture we will look at image enhancement point

Image Enhancement-Histogram Equalization 2

 The histogram of a digital image with gray values

 The function p(rk) represents the fraction of the total

number of pixels with gray value rk.

Image Enhancement-Histogram Equalization 3

 Histogram provides a global description of the

Image Enhancement-Histogram Equalization 4

 The shape of a histogram provides useful information

Image Enhancement-Histogram Equalization 5

Image Enhancement-Histogram Equalization 6

Image Enhancement-Histogram Equalization 7

 The histogram equalization is an approach to enhance

 We need to design a gray value transformation s =

T(r), based on the histogram of the input image, which

Image Enhancement-Histogram Equalization 8

range of allowed Gray

Image Enhancement-Histogram Equalization 9

 Let us denote the inverse transformation by r T -1(s) .

Image Enhancement-Histogram Equalization 10

 If pin(r) and T(r) are known, and r T -1(s) satisfies

 In terms of histograms, the output image will have all

gray values in “equal proportion” .

Image Enhancement-Histogram Equalization 11

 Next we derive the gray values in the output is

 Note that this is the cumulative distribution function

Image Enhancement-Histogram Equalization 12

 Therefore, the output histogram is given by

uniform gray values.

 This usually results in an enhanced image, with an

increase in the dynamic range of pixel values.

Image Enhancement-Histogram Equalization 13

 Step 1:For images with discrete gray values, compute:

of the previous transformation :

Image Enhancement-Histogram Equalization 14

 Consider an 8-level 64 x 64 image with gray values (0,

Image Enhancement-Histogram Equalization 15

Gray value Normalized gray value

Image Enhancement-Histogram Equalization 16

Image Enhancement-Histogram Equalization 17

Image Enhancement-Histogram Equalization 18

 Note that the histogram of output image is only

approximately, and not exactly, uniform. This should not be

Image Enhancement-Histogram Equalization 20

Image Enhancement-Histogram Equalization 21

Image Enhancement-Histogram Equalization 22

Image Enhancement-Histogram Equalization 23

Image Enhancement-Histogram Equalization 24

 Histogram equalization may not always produce

Image Enhancement-Histogram Equalization 25

Image Enhancement-Histogram Equalization 26

Input Uniform Output

Image Enhancement-Histogram Equalization 27

 If the desired output image were available, then the

following transformation would generate an image with

will generate an image with the specified density out