Name : Asrat W/Mariam IDNo:PGR/48777/13

Department : Information Technology.

Digital Image processing Assignment Report II
Problem Four:

Problem Four

This problem targets on different types of Noise those are basically..

% UNIFORM % GAMMA
%GAUSSIAN % EXPONENETIAL
%RAYLIGH % IMPULSE(SALT AND PEPPER) and reduction methods
those are ordered in the following way :
Gaussian noise is also called additive noise(shot noise/photon) as each pixel is modified such
that certain distribution is added to each pixel. Noise caused in Gaussian due to poor
illumination/high temperature and also caused due to present in electronic circuits. So , its
models are used frequently in practice and Gaussian models being used in situations in which
they are marginally applicable at best. It follows a Poisson distribution.
Mathematical traceability of Gaussian models:

Therefor the following image compares by taking the input image of circuit from
the user to add Gaussian noise and the result of noise.

Digital image Processing Assignment- II Page 1

Digital image Processing Assignment- II Page 2
  Parameter a must be positive for exponential type.

Digital image Processing Assignment- II Page 3

 The following investigation of the salat and pepper noise shows impulse noise can be
characterized as sparse disturbances in the image leading to discoloration of a few
pixels(black and white) in the image and caused due to sudden disturbances like dust or a
faulty charge coupled device during the capturing of the image ,so the image comparisons
of the original image and the results of salt and pepper noise those was the noise
effects only a small number of pixels leaving the rest of the picture.
In the below Image corrupted by pepper & salt noise with a probability of 0.1 .

Digital image Processing Assignment- II Page 4

 Removing Gaussian noise are described in the following order of investigations :
1. 5x5 Arithmetic mean filter is the simplest mean filters let Sxy represents the set of
coordinates in a rectangular sub image window(neighborhood) of size mxn, centered at
point (x,y).
 Computes the average value of the corrupted image g(x,y) in the area defined by Sxy .
 Pixels in the regions defined by Sxy.
 Operations implemented using a spatial filter of size mxn in coefficients have value
1/mn .
 Smooth’s local variations in an image and noise is reduced as result of blurring.

2. Geometric mean filer :achieves smoothing comparable to the arithmetic mean filter, but
it tends to lose less image detail in the process and also it was ineffective for pepper type
of noise.
The following image compares the original image ,result of Gaussian noise eliminated
by Geometric mean filter and result of 5x5 Geometric mean filter.

The image of original circuit & Reconstructed image using Geometric mean filter

3. Harmonic mean filter:works well for salt noise, but fails for pepper noise.It does well
also with other types of noise like Gaussian noise.

4. Contra harmonic mean filter:

- yields a restored image based on the expressions

Digital image Processing Assignment- II Page 5

 well suited for reducing or virtually eliminating the effects of salt-and-pepper noise.For
positive values of Q(Q>0), the filter eliminates pepper noise. For negative values of Q
(Q<0) it eliminates salt noise and Q is the order of the filter. It can not do both
simultaneously. Note that contra harmonic filter reduces to the arithmetic filter if Q=0,
and to the harmonic filter if Q=-1
In general ,the arithmetic and geometric mean are well suited for random noise like
Gaussian or uniform noise. The contra harmonic filter is well suited for impulse noise,
but it has disadvantage that it must be known whether the noise is dark or light in
order to select the proper sign for Q. The result of choosing the wrong sign for Q can be
disastrous.
The below image compares whether the noise is dark or light

Median filter: the best known order-statistic filter which implies, replaces the values of a
pixel by the median of the intensity levels in the neighborhood of that pixel and Quite
popular because for certain type of random noise ,they provide excellent noise-reduction
capabilities, with considerably less blurring than linear smoothing filters of similar size.
 Are effective in the presence of both bipolar and unipolar impulse
noise and particularly good when salt pepper noise is present

Max and Min filters:

i) Max filter:useful for finding the brightest points in an image. Also b/c pepper noise has
very low values,it is reduced by this filter as result of the max selection process in the sub image
area Sxy. The following image comparisons shows taking the original image of the circuit and
out puts of the result of Maximum of 0s filter to find the maximum value in the area
encompassed by the filter and enhances bright areas of the image not only finding but also
reducing of the pepper noise as a result of the max operation.

Digital image Processing Assignment- II Page 6

The following image operations Max filter shows how to remove pepper noise.

ii) Min filter:0th percentile filter and useful for finding the darkest points in an
image .Also reduces salt noise as a result of the min operation.
- Used for removing salt noise.

The following image operations Min filter shows how to remove salt noise.

Midpoint filter:computes the mid points b/n the maximum and minimum values in the above area
of images encompassed by the filter and also combines order statistics and averaging .It works
best for randomly distributed noise like Gaussian or uniform noise.

Alpha-trimmed mean filter:

Digital image Processing Assignment- II Page 7

Digital image Processing Assignment- II Page 8
Problem Five

This problem is accomplished in 4 different parts and the solution deals with Implement a
blurring filter using the equation above & Blur the test image book_cover.jpg using parameters
a=b=0.1, T=1, Add Gaussian noise of 0 mean and variance of 650 to the blurred image, Restore
the blurred image and the blurred noisy image using the Inverse Filter and Wiener
Deconvolution Filter respectively and Add Gaussian noise of 0 and different variances to the
blurred image and repeat the above step (i.e.3) . All of which are implemented in Matlab.
The first part of this problem is Implement a blurring filter using the equation above and Blur
the test image book_cover.jpg using parameters a=b=0.1, T=1.
If the image is moving towards x and y separately, X0(t) = at/T, y0 (t) = bt/T then we can use the
following degrading function to process the image.

Digital image Processing Assignment- II Page 9

The figure below (on the right) depicts the result of implementing blurring filter on the image on
the left (original image)

a)

Blurred image less clear or distinct. It is done with the help of various low pass filter
kernels.

Blurred image is :hasin smoothing the image Low intensity edges are removed.
 helps in hiding the details when necessary. For e.g. in many cases
policedeliberately want to hide the face of the victim, in such cases blurring is
required.
b) When the size of parameters increase the blurred image should be more dark.
( H = motionDegradation(height, width, 0.1, 0.1, 1);
and again add Gaussian noise of 0 mean and variance of 650 to the blurred image on which is

performed at the first part. For the motion blurred image, we consider adding the Gaussian
Noise.
The figure below refers Gaussian noise of 0 mean and variance of 650 is added on the image on
the left image or on the blurred image.

Digital image Processing Assignment- II Page 10

C) In this phase of filtering the blurred image and the blurred noisy image using the Inverse
Filter and Wiener Deconvolution Filter .
The figure below (on the right) shows applying inverse filter in order to restore the blurred and
the blurred noisy image .

The formula for the Wiener Filtering:-

(2)
Here k is a special value. We tried different k values for test while I have used 0.009 in my
implementation

The following image shows the effect of applying wiener filter in order to restore the blurred and
the blurred noisy image by.

Digital image Processing Assignment- II Page 11

The above two filters wiener Deconvolution filtering with special value of k has better result
than inverse filtering
d) In this phase by adding Gaussian noise of 0 and different variances to the blurred image and
investigate the performance of the Wiener filter by using different k values and also investigation
of variable parameters are given as follows gaussian_ variances = [90, 350, 650], Wiener
parameters = [0.0009, 0.009, 0.09];

The figure below show that the effect of Adding Gaussian noise of 0 mean and variance of 90 to
the blurred image and applying wiener filter with wiener parameter(k value =0.0009,0.009 and
0.9 respectively) in order to restore the blurred and the blurred noisy image

The following image effect of Adding Gaussian noise of 0 mean and variance of 350 to the
blurred image and applying wiener filter with wiener parameter(k value =0.0009,0.009 and 0.9
respectively) in order to restore the blurred and the blurred noisy image

Digital image Processing Assignment- II Page 12

The effect of Adding Gaussian noise of 0 mean and variance of 650 to the blurred image and
applying wiener filter with wiener parameter(k value =0.0009,0.009 and 0.9 respectively) in
order to restore the blurred and the blurred noisy image

Problem Six discussions are the following based on practical result takes the
input image of ray_trace_bottle from the user out puts the following images by
rotating, translating and scaling specified amounts using the nearest neighbor and
bilinear interpolation methods as the following…

A) Image Rotation: transformation in which a figure turned about a point.90 degree

rotation turns 1 quadrant 180 degree rotation turns 2 quadrants 270 degree rotations
turns 3 quadrants.

a rotations all points in an image along a circular path in the plane centered at the
pivot point. First ,we will assume the pivot is at the origin. Positive values for the

Digital image Processing Assignment- II Page 13

 rotation angle define counterclockwise rotation and negative defines clockwise
direction.

Note :By Image Rotation, the image is rotated about its center by a specified
number of degrees. The rotation of an image is a geometric transformation. It can
be done either by Forward Transformation (or) Inverse Transformation

Here, in the following Image Processing uses Inverse Transformation and then some
Pixel Values Beyond Image Boundaries i.e Pixel values lying outside the Dimension of
the image. Such Values will not be displayed in the output image. 

There for ,the following image is rotated 45 Degree using the input of the original image
of ray_trace_bottle.

.
Scaling changes the size of an image and involves two scale factors X and Y
coordinates.
There are three scaling factors:
i) A scaling constant> 1 indicates expansion of length ie. Magnification.
ii) A scaling constant<1 indicates compression of length ie reduction.
iii) A scaling constant=1 leaves the size of image unchanged.

Digital image Processing Assignment- II Page 14

Digital image Processing Assignment- II Page 15
Digital image Processing Assignment- II Page 16
Digital image Processing Assignment- II Page 17