Scribd
Upload
48 views3 pages

Premcode 444

The document describes adding salt and pepper noise to an image and then applying various filters to reduce the noise, including a 3x3 median filter, 5x5 median filter, and Wiener filter. It calculates the peak signal-to-noise ratio (PSNR) between the original image and filtered images to evaluate the effectiveness of the different filtering approaches at reducing noise. It also implements a decision-based algorithm to filter salt and pepper noise that replaces corrupted pixels with the median value of neighboring pixels if it satisfies certain criteria, otherwise replaces it with the value of a neighboring pixel.

Uploaded by

hayumbe08
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOC, PDF, TXT or read online on Scribd
48 views3 pages

Premcode 444

The document describes adding salt and pepper noise to an image and then applying various filters to reduce the noise, including a 3x3 median filter, 5x5 median filter, and Wiener filter. It calculates the peak signal-to-noise ratio (PSNR) between the original image and filtered images to evaluate the effectiveness of the different filtering approaches at reducing noise. It also implements a decision-based algorithm to filter salt and pepper noise that replaces corrupted pixels with the median value of neighboring pixels if it satisfies certain criteria, otherwise replaces it with the value of a neighboring pixel.

Uploaded by

hayumbe08
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOC, PDF, TXT or read online on Scribd
You are on page 1/ 3

a=imread('coins.

png');
subplot(3,3,1);
imshow(a);
title('Original Image');
b=imnoise(a ,'Salt & Pepper',0.4);
subplot(3,3,2);
imshow(b);
title('Salt & Pepper Noise');
h1=1/9*ones(3,3);
h2=1/25*ones(5,5);
c1=conv2(b,h1,'same');
c2=conv2(b,h2,'same');
subplot(3,3,4);
imshow(uint8(c1));
title('3x3 Median Filter');
subplot(3,3,5);
imshow(uint8(c2));
title('5x5 adap Median Filter');
c3=medfilt2(b,[3 3]);
subplot(3,3,6);
imshow(c3);
title('3x3 MFI');
c4=medfilt2(b,[5 5]);
subplot(3,3,7);
imshow(c4);
title('adaptive');
c5=wiener2(b,[3 3]);
subplot(3,3,3);
imshow(c5);
title('3x3 median filter');
c6=imfilter(b,h2);
subplot(3,3,8);
imshow(c6);
title('5x5 Filter');
max2_A
max2_B
min2_A
min2_B

=
=
=
=

max(max(b));
max(max(c5));
min(min(b));
min(min(c5));

if max2_A > 1 || max2_B > 1 || min2_A < 0 || min2_B < 0


disp('input matrices must have values in the interval [0,1]')
end
error_diff = max2_A - max2_B;
decibels = 100*(1/(sqrt(mean(mean(error_diff^2)))));
disp(sprintf('PSNR adaptive median = %5.2f ',decibels));
max2_n
max2_h
min2_n
min2_h

=
=
=
=

max(max(b));
max(max(c6));
min(min(b));
min(min(c6));

if max2_n > 1 || max2_h > 1 || min2_n < 0 || min2_h < 0


disp('input matrices must have values in the interval [0,1]')
end
error_diff2 = max2_n - max2_h;
decibels2 = 100*(1/(sqrt(mean(mean(error_diff2^2)))));

iii

disp(sprintf('PSNR median= %5.2f ',decibels2));


max2_d
max2_e
min2_d
min2_e

=
=
=
=

max(max(b));
max(max(c2));
min(min(b));
min(min(c2));

if max2_d > 1 || max2_e > 1 || min2_d < 0 || min2_e < 0


disp('input matrices must have values in the interval [0,1]')
end
error_diff1 = max2_d - max2_e;
decibels1 = 100*(1/(sqrt(mean(mean(error_diff1^2)))));
disp(sprintf('PSNR normal filter= %5.2f ',decibels1))

x= imread('coins.png');
%% Demo
disp('Noise density lies between 0 and 1');
%% Noise Generation
disp(' ');
ND = input('Enter Noise Density [0.5] : ');
if isempty(ND)
ND = 0.5;
end
y = imnoise(x,'salt & pepper',ND);
y = double(y);
Y = y;
[R C] = size(x);
%% Implementation
for i = 2:R-1
for j = 2:C-1
clear tmp;
tmp = Y(i-1:i+1,j-1:j+1);
flg = 0;
S = sort(tmp(:));
if Y(i,j) == 0 || Y(i,j) == 255
flg = 1;
end
if flg == 0 % if the Pixel is uncorrupted.
if S(1) < Y(i,j) && Y(i,j) < S(5) && 0 < S(1) && S(9) < 255 %
The
% P(X,Y) is an uncorrupted pixel if Pmin < P(X,Y) < Pmax,
% Pmin > 0 and Pmax < 255; the pixel being processed is
% left unchanged. Otherwise, P(X,Y) is a corrupted pixel.
Y(i,j) = Y(i,j);
end
else % if the Pixel is corrupted.
if S(1) < S(5) && S(5) < S(9) && 0 < S(5) && S(5) < 255 % If
P(X,Y)
% is a corrupted pixel, it is replaced by its median value
% if Pmin < Pmed < Pmax and 0 < Pmed < 255.
Y(i,j) = S(5);

iv

end
if S(1) >= S(5) || S(5) >= S(9) || S(5) == 255 && S(5) == 0
% If Pmin < Pmed < Pmax is not satisfied or 255 < Pmed =

0,

% then is a noisy pixel. In this case, the P(X,Y) is

replaced

% by the value of neighborhood pixel value.


Y(i,j) = Y(i,j-1);
end

end

end
end
%% Border Correction
Y(1,:) = Y(2,:);
Y(R,:) = Y(R-1,:);
Y(:,1) = Y(:,2);
Y(:,C) = Y(:,C-1);
f = medfilt2(y,[3 3]);
figure;subplot(2,2,1);imshow(x,[]);title('Given Image');
subplot(2,2,2);imshow(y,[]);title(strcat('Noisy : Noise Density ',num2str(ND)));xlabel('PSNR(x,y)');
subplot(2,2,3);imshow(f,[]);title('Traditional
Median');xlabel('PSNR(x,f)');
subplot(2,2,4);imshow(Y,[]);title('DBA');xlabel('PSNR(x,Y)');
%% PSNR
% function P = PSNR(A,B)
% A = double(A);B = double(B);
% error = abs(A - B);
% P = 20*log10(255/(sqrt(mean(mean(error.^2)))));
%
% )

You might also like