Image Image

Processing Processing
Image Restoration
8/14/2010
1
Contents
Speckle noise.
Gaussian noise.
Salt & pepper noise.
Cleaning Gaussian noise
Image Restoration
8/14/2010
2
Periodic noise.
Cleaning salt and pepper noise
Cleaning Periodic noise
Image Restoration
Image restoration concerns the removal or
reduction of degradations which have occurred
during the acquisition of the image
Such degradations may include noise, which are
errors in the pixel values, or optical effects such as
out of focus blurring, or blurring due to camera
motion.
8/14/2010
3
Image Restoration cont.
Assume we have an image of f(x,y) and a spatial
filter h(x,y) if h(x,y) consists of a single line of ones,
the result of the convolution will be a motion blur in
the direction of the line. Thus we may write
g(x,y) =f(x,y)*h(x,Y)
We may use additive function to generate noise
g(x,y) =f(x,y)*h(x,y)+n(x,y)
If we know h,n we may acquire the original image
f(x,y)= (g(x,y)-n(x,y))/h(x,y)
Noise
We may define noise to be any degradation in the
image signal, caused by external disturbance. If an
image is being sent electronically from one place to
another, via satellite or wireless transmission, or
through networked cable, we may expect errors to
occur in the image signal
Cleaning an image corrupted by noise is thus an
important area of image restoration.
Salt and Pepper noise
Its appearance is randomly scattered white or
black pixles over the image
tw=imread('twins.tif');
t=rgb2gray(tw);
To add noise, we use the Matlab function imnoise which
takes a number of different parameters.
To add salt and pepper noise
t_sp=imnoise(t,'salt & pepper'); or
imnoise(t,'salt & pepper',0.2);
would produce an image with 20% of its pixels corrupted
by salt and pepper noise
Salt and Pepper noise-Cont
Original image
With added salt &
pepper noise
Gaussian noise-Cont.
Gaussian noise is an idealized form of white noise .
We can observe white noise by watching a television
which is slightly mistuned to a particular channel.
If the image is represented as I. and the Gaussian
noise by
N , then we can model a noisy image by simply adding
the two: I+N
t_spk=imnoise(t, Gaussian ');
Speckle noise
Whereas Gaussian noise can be modeled by random
values added to an image . speckle noise can be
modeled by random values multiplied by pixel values
hence it is also called multiplicative noise.
t_spk=imnoise(t,'speckle');
peckle noise is implemented as
I(1+N)
Speckle noise-Cont.
Gaussian noise
Speckle noise
Periodic noise
The function imnoise does not have a periodic option,
but it is quite easy to create our own, by adding a
periodic matrix
>> s=size(t);
>> [x,y]=meshgrid(1:s(1),1:s(2));
>> p=sin(x/3+y/5)+1;
>> t_pn=(im2double(t)+p/2)/2;
Periodic noise-Cont.
Periodic noise
Cleaning salt and pepper noise
Low pass filter
So we might try filtering with an average filter:
a3=fspecial('average');
t_sp_a3=filter2(a3,t_sp);
Median filtering
Median filtering seems almost tailor-made for
removal of salt and pepper noise.
The median of a set is the middle value when they
are sorted
Thus the median will in general replace a noisy value
with one closer to its surroundings
A median filter is an example of a non-linear
spatial filter
Median filtering-Cont.
In Matlab, median filtering is implemented by the
medfilt2 function
t_sp_m3=medfilt2(t_sp) ;
Median filtering
Rank-order filtering
Median filtering is a special case of a more general
process called rank-order filtering Rather than take
the median of a set, we order the set and take the
n_th th value, for some predetermined value of n
ordfilt2(t_sp,3,[0 1 0;1 1 1;0 1 0]);
Rank-order filtering-Cont.
Applying the median filter can in general be a slow
operation: each pixel requires the sorting of at least
nine values .The use of cleaning salt and
pepper noise by treating noisy pixels known as
outliers.
This leads to the following approach for noise
cleaning.
1. Choose a threshold value D
2. For a given pixel, compare its value P with the
mean M of the values of its eight neighbors
3. If(P-M)>D then classify the pixel as noisy,
otherwise not
4. If the pixel is noisy, replace its value with M
otherwise leave its value unchanged.
Rank-order filtering-Cont.
Rank-order filtering-Cont.
D=.2
D=.4
Cleaning Gaussian noise
It may sometimes happen that instead of just one
image corrupted with Gaussian noise, we have many
different copies of it. An example is satellite imaging; if
a satellite passes over the same spot many times, we
will obtain many different images of the same place
Simple approach to cleaning Gaussian noise is to
simply take the average of all the images .
To see why this works, suppose we have 100 copies of
our image, each with noise then the Ith noisy image
will be:M+NI
M is the matrix of original values and NI a matrix of
normally distributed random values with mean 0
We can define the mean M of these images by the
usual add and divide method
Cleaning Gaussian noise-Cont.
Removal of periodic noise
Periodic noise may occur if the
imaging equipment is subject to
electronic disturbance of a repeating
nature.
We can easily create periodic noise by
overlaying an image with a
trigonometric function:
>> [x,y]=meshgrid(1:256,1:256);
>> p=1+sin(x+y/1.5);
>> cm = imread('cameraman.tif');
>> tp=(double(cm)/128+p)/4;
>> imshow(tp)
>>fcm= fftshift(fft(tp));
>>figure,imshow(fcm)
Where cm is cameraman image.
Removal of periodic noise-Cont
The second line simply creates a sine
function, and adjusts its output to be
in the range 0-2.
The third line first adjusts the
cameraman image to be in the same
range; adds the sine function to it,
and divides by 4 to produce a matrix
of type double with all elements in
the range 0.0 1.0.
Removal of periodic noise-Cont
y This can be viewed directly with imshow,
and it is shown in gure 6.1(a). We can
produce its shifted DFT and this is shown
in gure 6.1(b). The extra two spikes
away from the centre correspond to the
noise just added.
The extra two spikes away from the centre
correspond to the noise just added.
Removal of periodic noise-Cont
Removal of periodic noise-Cont
There are two methods we can use to
eliminate the spikes :
1. Band reject filtering.
2. Notch filtering.
Removal of periodic noise-Cont
Band reject filtering
y We create a filter consisting of ones with a ring
of zeroes; the zeroes lying at a radius of 49
from the centre:
>> z=sqrt((x-129).^2+(y-129).^2);
>> br=(z < 47 | z > 51);
y where z is the matrix consisting of distances
from the origin. This particular ring will have a
thickness large enough to cover the spikes. Then
as before, we multiply this by the transform:
>> tbr=fcm.*br;
>> figure,imshow(tbr,[])
Removal of periodic noise-Cont
and this is shown in figure 6.2(a).
The result is that the spikes have
been blocked out by this filter.
Taking the inverse transform
produces the image shown in gure
6.2(b).
Note that not all the noise has gone,
but a significant amount has,
especially in the centre of the image.
Removal of periodic noise-Cont
Removal of periodic noise-Cont
Notch filtering
y With a notch Filter, we simply make
the rows and columns of the spikes
zero:
>> tf(156,:)=0;
>> tf(102,:)=0;
>> tf(:,170)=0;
>> tf(:,88)=0;
and the result is shown in gure 6.3(a).
The image after inversion is shown in
gure 6.3(b).
Removal of periodic noise-Cont
Removal of periodic noise-Cont
LOGO
Any questions