Scribd Logo
Upload

Welcome to Scribd!

  • 42 views

    Text Figure (4.2) 1D Discrete Fourier Transform

    Uploaded by

    Mohammad Bilal Mirza
    This document contains MATLAB code and figures demonstrating various digital image filtering techniques using the discrete Fourier transform (DFT), including 1D and 2D DFT, notch filtering, high pass/low pass filtering, ideal low pass filtering, Butterworth filtering, and Gaussian filtering. The code generates filter kernels in the Fourier domain, applies them to test images by multiplying the image Fourier transform, and transforms back to the spatial domain to show the filtered results.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PDF, TXT or read online from Scribd

    Text Figure (4.2) 1D Discrete Fourier Transform

    Uploaded by

    Mohammad Bilal Mirza
    42 views20 pages
    This document contains MATLAB code and figures demonstrating various digital image filtering techniques using the discrete Fourier transform (DFT), including 1D and 2D DFT, notch filtering, high pass/low pass filtering, ideal low pass filtering, Butterworth filtering, and Gaussian filtering. The code generates filter kernels in the Fourier domain, applies them to test images by multiplying the image Fourier transform, and transforms back to the spatial domain to show the filtered results.

    Original Description:

    Enhancement_set2

    Original Title

    Enhancement Set2

    Copyright

    © Attribution Non-Commercial (BY-NC)

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    This document contains MATLAB code and figures demonstrating various digital image filtering techniques using the discrete Fourier transform (DFT), including 1D and 2D DFT, notch filtering, high pass/low pass filtering, ideal low pass filtering, Butterworth filtering, and Gaussian filtering. The code generates filter kernels in the Fourier domain, applies them to test images by multiplying the image Fourier transform, and transforms back to the spatial domain to show the filtered results.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    42 views20 pages

    Text Figure (4.2) 1D Discrete Fourier Transform

    Uploaded by

    Mohammad Bilal Mirza
    This document contains MATLAB code and figures demonstrating various digital image filtering techniques using the discrete Fourier transform (DFT), including 1D and 2D DFT, notch filtering, high pass/low pass filtering, ideal low pass filtering, Butterworth filtering, and Gaussian filtering. The code generates filter kernels in the Fourier domain, applies them to test images by multiplying the image Fourier transform, and transforms back to the spatial domain to show the filtered results.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 20

    Text figure (4.

    2) 1D Discrete Fourier Transform


    %Matlab Code
    A = 1; K = 10; M = 100;
    f1 = zeros(1,M); f2 = zeros(1,M);
    f1(1:K) = A; f2(1:2*K) = A;
    fft1 = abs(fftshift(fft(f1)))/M;
    fft2 = abs(fftshift(fft(f2)))/M;
    figure;
    subplot(2,2,1);
    plot(f1); title('Filter for M=100, K=10, A=1');
    ylabel('f(x)'); axis([0 M 0 1.1*A]);
    subplot(2,2,2);
    plot(fft1); title('Fourier Transform for Filter
    ylabel('|F(u)|'); axis([0 M 0 1.1*[A*K]/M]);
    subplot(2,2,3);
    plot(f2); title('Filter for M=100, K=10, A=1');
    ylabel('f(x)'); axis([0 M 0 1.1*A]);
    subplot(2,2,4);
    plot(fft2); title('Fourier Transform for Filter
    ylabel('|F(u)|'); axis([0 M 0 1.1*[2*A*K]/M]);

    xlabel('x');

    in (a)'); xlabel('u');

    xlabel('x');

    in (c)'); xlabel('u');

    Fourier Transform for Filter in (a)

    0.1

    0.8

    0.08

    0.6

    0.06

    |F(u)|

    f(x)

    Filter for M=100, K=10, A=1

    0.4

    0.04

    0.2

    0.02

    20

    40

    60

    80

    100

    20

    40

    60

    80

    Filter for M=100, K=10, A=1

    Fourier Transform for Filter in (c)

    100

    0.2

    0.8

    0.15

    0.6

    |F(u)|

    f(x)

    0.1

    0.4
    0.05

    0.2
    0

    20

    40

    60
    x

    80

    100

    20

    40

    60
    u

    Text figure (4.3) 2D Discrete Fourier Transform

    80

    100

    4.4)
    %Matlab Code
    close all; clear all;
    img = imread('Fig4.03(a).jpg');
    imgfft = abs(fftshift(fft2(img)));
    figure;
    subplot(1,2,1);
    imshow(img);
    subplot(1,2,2);
    imshow(imscale01(log10(imgfft+1)));

    Text figure (4.6) 2D Fourier Notch Filtering

    %Matlab Code
    close all; clear all;
    img = imread('Fig4.04(a).jpg');
    imgfft = fftshift(fft2(img));
    imgfft(size(img,1)/2+1, size(img,2)/2+1) = 0;
    imgfilt = abs((ifft2(imgfft)));
    figure;
    imshow(imscale01(imgfilt));

    Text figure (4.7) High Pass / Low Pass 2D Filtering

    %Matlab Code
    close all; clear all;
    img = imread('Fig4.04(a).jpg');
    [x,y] = size(img);
    lp = zeros(size(img));
    cx = x/2; cy = y/2; std = 30;
    for i=1:x; for j=1:y;
    lp(i,j) = exp(-[(i-cx)^2 + (j-cy)^2]/std^2);
    end end
    hp = 1-lp;
    lpplot = lp(1:8:x, 1:8:y); hpplot = hp(1:8:x, 1:8:y);
    imglp = fftshift(fft2(img)); imglp = imglp .* lp;
    imghp = imglp; imghp = imghp .* hp;
    imglp = abs((ifft2(imglp))); imghp = abs((ifft2(imghp)));
    figure;
    subplot(2,2,1);
    subplot(2,2,2);
    subplot(2,2,3);
    subplot(2,2,4);

    mesh(lpplot); axis tight; colormap(gray(1));


    imshow(imscale01(imglp));
    mesh(hpplot); axis tight; colormap(gray(1));
    imshow(imscale01(imghp));

    Text figure (4.8) Modified 2D High Pass Filter

    %Matlab Code
    close all; clear all;
    img = imread('Fig4.04(a).jpg');
    [x,y] = size(img);
    lp = zeros(size(img));
    cx = x/2; cy = y/2; std = 30;
    for i=1:x;
    for j=1:y;
    lp(i,j) = exp(-[(i-cx)^2 + (j-cy)^2]/std^2);
    end
    end
    hp = 1-lp; hp = hp + 0.5*max(hp(:));
    imghp = fftshift(fft2(img)); imghp = imghp .* hp;
    imghp = abs((ifft2(imghp)));
    figure; imshow(imscale01(imghp));

    Text figure (4.12) 2D Ideal Low-pass Filtering


    %Matlab Code

    close all; clear all;


    img = double(imread('Fig4.11(a).jpg')); [x,y] = size(img);
    lp5
    = zeros(size(img)); lp15 = zeros(size(img));
    lp30 = zeros(size(img)); lp80 = zeros(size(img));
    lp230 = zeros(size(img));
    cx = x/2; cy = y/2; std = 30;
    for i=1:x;
    for j=1:y;
    if(sqrt([i-cx]^2 + [j-cy]^2) < 5)
    lp5(i,j) = 1;
    end
    if(sqrt([i-cx]^2 + [j-cy]^2) < 15)
    lp15(i,j) = 1;
    end
    if(sqrt([i-cx]^2 + [j-cy]^2) < 30)
    lp30(i,j) = 1;
    end
    if(sqrt([i-cx]^2 + [j-cy]^2) < 80)
    lp80(i,j) = 1;
    end
    if(sqrt([i-cx]^2 + [j-cy]^2) < 230)
    lp230(i,j) = 1;
    end
    end
    end
    imgfft = fftshift(fft2(img));
    lp5
    = abs(ifft2(imgfft .* lp5)); lp15 = abs(ifft2(imgfft .* lp15));
    lp30 = abs(ifft2(imgfft .* lp30)); lp80 = abs(ifft2(imgfft .* lp80));
    lp230 = abs(ifft2(imgfft .* lp230));
    figure;
    subplot(3,2,1); imshow(imscale01(img)); subplot(3,2,2); imshow(imscale01(lp5));
    subplot(3,2,3); imshow(imscale01(lp15)); subplot(3,2,4); imshow(imscale01(lp30));
    subplot(3,2,5); imshow(imscale01(lp80)); subplot(3,2,6); imshow(imscale01(lp230));

    Text figure (4.13) Ideal Low-Pass Filter Response

    %Matlab Code
    close all; clear all;
    img = zeros(500,500);
    x = 500; y = 500;
    cx = 250; cy = 250;
    for i=1:x;
    for j=1:y;
    if(sqrt([i-cx]^2 + [j-cy]^2) < 5)
    img(i,j) = 1;
    end
    end
    end
    imgfft = (fftshift(ifft2(img)));
    a = ones(1,500);
    a(1:2:500) = -1;
    plota = real(imgfft(250,:)).*a;
    img2
    = zeros(size(img));
    img2(128,128) = 1; img2(128, 500-128) = 1; img2(250, 250) = 1; img2(500-128, 128) = 1;
    img2(500-128, 500-128) = 1; img3 = conv2(img2, imgfft,'same');
    idx = 1:size(img,2)+1:length(img(:));
    plotb = real(img3(idx));
    figure;
    subplot(3,2,1);
    subplot(3,2,2);
    subplot(3,2,3);
    subplot(3,2,4);
    subplot(3,2,5);
    subplot(3,2,6);

    imshow(imscale01(img));
    plot((plota)); axis tight;
    imshow(imscale01(0.5*log10(abs(imgfft)+1)));
    imshow(imscale01(img2));
    plot((plotb)); axis tight;
    imshow(imscale01(0.5*log10(abs(img3)+1)));
    -4

    x 10
    2
    1
    0

    100

    200

    -4

    x 10

    2
    1

    0
    100

    200

    300

    400

    500

    Text figure (4.15) Butterworth Low-pass Filtering


    %Matlab Code

    300

    400

    500

    close all; clear all;


    img
    = im2double(imread('Fig4.11(a).jpg'));
    imgfft = fftshift(fft2(img));
    H5
    = zeros(500,500); H15 = H5; H30 = H5; H80 = H5;
    H230 = H5; x = 500; y = 500; cx = x/2; cy = y/2;
    for i=1:x;
    for j=1:y;
    D(i,j)
    = sqrt([i-cx]^2 + [j-cy]^2);
    H5(i,j)
    = 1 / (1 + [D(i,j)/5]^(4));
    H15(i,j) = 1 / (1 + [D(i,j)/15]^(4));
    H30(i,j) = 1 / (1 + [D(i,j)/30]^(4));
    H80(i,j) = 1 / (1 + [D(i,j)/80]^(4));
    H230(i,j) = 1 / (1 + [D(i,j)/230]^(4));
    end
    end
    img5
    = abs(ifft2(imgfft .* H5)); img15 = abs(ifft2(imgfft .* H15));
    img30 = abs(ifft2(imgfft .* H30)); img80 = abs(ifft2(imgfft .* H80));
    img230 = abs(ifft2(imgfft .* H230));
    figure;
    subplot(3,2,1); imshow(imscale01(img)); subplot(3,2,2); imshow(imscale01(img5));
    subplot(3,2,3); imshow(imscale01(img15)); subplot(3,2,4); imshow(imscale01(img30));
    subplot(3,2,5); imshow(imscale01(img80)); subplot(3,2,6); imshow(imscale01(img230));

    Text figure (4.16) Butterworth Low-Pass Filter Response


    %Matlab Code

    close all; clear all;


    img
    = im2double(imread('Fig4.11(a).jpg'));
    imgfft = fftshift(fft2(img));
    H1
    = zeros(500,500); H2 = H1; H5 = H1; H20 = H1;
    H230 = H5; x = 500; y = 500; cx = x/2; cy = y/2;
    for i=1:x;
    for j=1:y;
    D(i,j)
    = sqrt([i-cx]^2 + [j-cy]^2);
    H1(i,j) = 1 / (1 + [D(i,j)/5]^(2));
    H2(i,j) = 1 / (1 + [D(i,j)/5]^(4));
    H5(i,j) = 1 / (1 + [D(i,j)/5]^(10));
    H20(i,j) = 1 / (1 + [D(i,j)/5]^(40));
    end
    end
    h1 = abs(fftshift(ifft2(H1))); h2 = abs(fftshift(ifft2(H2)));
    h5 = abs(fftshift(ifft2(H5))); h20 = abs(fftshift(ifft2(H20)));
    figure;
    subplot(2,4,1);
    subplot(2,4,2);
    subplot(2,4,3);
    subplot(2,4,4);
    subplot(2,4,5);
    subplot(2,4,6);
    subplot(2,4,7);
    subplot(2,4,8);

    imshow(imscale01(20*log10(1+h1)));
    imshow(imscale01(20*log10(1+h2)));
    imshow(imscale01(20*log10(1+h5)));
    imshow(imscale01(20*log10(1+h20)));
    plot(h1(250,:)); axis tight;
    plot(h2(250,:)); axis tight;
    plot(h5(250,:)); axis tight;
    plot(h20(250,:)); axis tight;

    -4

    -4

    x 10

    -4

    -4

    x 10

    x 10

    x 10
    3

    3
    15

    2.5
    2.5

    5
    1

    400

    1.5

    1.5

    200

    10

    200

    400

    0.5

    0.5
    200

    400

    Text figure (4.18) Gaussian Low-Pass Filtering


    %Matlab Code

    200

    400

    close all; clear all;


    img
    = im2double(imread('Fig4.11(a).jpg'));
    imgfft = fftshift(fft2(img));
    H5
    = zeros(500,500); H15 = H5; H30 = H5; H80 = H5; H230
    x = 500; y = 500; cx = x/2; cy = y/2;
    for i=1:x;
    for j=1:y;
    D(i,j)
    = sqrt([i-cx]^2 + [j-cy]^2);
    end
    end
    H5
    = exp(-.5*[D/5].^2); H15 = exp(-.5*[D/15].^2);
    H30 = exp(-.5*[D/30].^2); H80 = exp(-.5*[D/80].^2);
    H230 = exp(-.5*[D/230].^2);
    h5
    = abs(ifft2(H5.*imgfft)); h15
    h30 = abs(ifft2(H30.*imgfft)); h80
    h230 = abs(ifft2(H230.*imgfft));

    = H5;

    = abs(ifft2(H15.*imgfft));
    = abs(ifft2(H80.*imgfft));

    figure;
    subplot(3,2,1); imshow(imscale01(img)); subplot(3,2,2); imshow(imscale01(h5));
    subplot(3,2,3); imshow(imscale01(h15)); subplot(3,2,4); imshow(imscale01(h30));
    subplot(3,2,5); imshow(imscale01(h80)); subplot(3,2,6); imshow(imscale01(h230));

    Text figure (4.20) Gaussian Low-Pass Filtering


    %Matlab Code

    close all; clear all;


    img
    = im2double(imread('Fig4.20(a).jpg'));
    imgfft = fftshift(fft2(img));
    [x,y] = size(img); cx = x/2; cy = y/2;
    for i=1:x;
    for j=1:y;
    D(i,j) = sqrt([i-cx]^2 + [j-cy]^2);
    end
    end
    H100 = exp(-.5*[D/100].^2); H80 = exp(-.5*[D/80].^2);
    h100 = abs(ifft2(H100.*imgfft)); h80 = abs(ifft2(H80.*imgfft));
    figure;
    subplot(2,3,1);
    subplot(2,3,2);
    subplot(2,3,3);
    subplot(2,3,4);
    subplot(2,3,5);
    subplot(2,3,6);

    imshow(imscale01(img(380:460,380:460)));
    imshow(imscale01(h100(380:460,380:460)));
    imshow(imscale01(h80(380:460,380:460)));
    imshow(imscale01(img));
    imshow(imscale01(h100));
    imshow(imscale01(h80));

    Text figure (4.21) Gaussian Low-Pass Filtering


    %Matlab Code

    close all; clear all;


    img
    = im2double(imread('Fig4.21(a).jpg'));
    imgfft = fftshift(fft2(img));
    [x,y] = size(img); cx = x/2; cy = y/2;
    for i=1:x;
    for j=1:y;
    D(i,j) = sqrt([i-cx]^2 + [j-cy]^2);
    end
    end
    H30 = exp(-.5*[D/30].^2); H10 = exp(-.5*[D/10].^2);
    h30 = abs(ifft2(H30.*imgfft)); h10 = abs(ifft2(H10.*imgfft));
    figure;
    subplot(1,3,1); imshow(imscale01(img));
    subplot(1,3,2); imshow(imscale01(h30));
    subplot(1,3,3); imshow(imscale01(h10));

    Text figure (4.24) Ideal High-pass Filtering


    %Matlab Code

    close all; clear all;


    img = double(imread('Fig4.11(a).jpg')); [x,y] = size(img);
    hp15 = zeros(size(img)); hp30 = zeros(size(img));
    hp80 = zeros(size(img));
    cx = x/2; cy = y/2; std = 30;
    for i=1:x;
    for j=1:y;
    if(sqrt([i-cx]^2 + [j-cy]^2) > 15)
    hp15(i,j) = 1;
    end
    if(sqrt([i-cx]^2 + [j-cy]^2) > 30)
    hp30(i,j) = 1;
    end
    if(sqrt([i-cx]^2 + [j-cy]^2) > 80)
    hp80(i,j) = 1;
    end
    end
    end
    imgfft = fftshift(fft2(img));
    hp15 = abs(ifft2(imgfft .* hp15));
    hp30 = abs(ifft2(imgfft .* hp30));
    hp80 = abs(ifft2(imgfft .* hp80));
    figure;
    subplot(1,3,1); imshow(imscale01(hp15));
    subplot(1,3,2); imshow(imscale01(hp30));
    subplot(1,3,3); imshow(imscale01(hp80));

    Text figure (4.25) Butterworth High-Pass Filtering


    %Matlab Code

    close all; clear all;


    img
    = im2double(imread('Fig4.11(a).jpg'));
    imgfft = fftshift(fft2(img));
    H15 = zeros(500,500); H30 = H15; H80 = H15;
    x = 500; y = 500; cx = x/2; cy = y/2;
    for i=1:x;
    for j=1:y;
    D(i,j)
    = sqrt([i-cx]^2 + [j-cy]^2);
    H15(i,j) = 1 / (1 + [15/D(i,j)]^(4));
    H30(i,j) = 1 / (1 + [30/D(i,j)]^(4));
    H80(i,j) = 1 / (1 + [80/D(i,j)]^(4));
    end
    end
    img15 = abs(ifft2(imgfft .* H15));
    img30 = abs(ifft2(imgfft .* H30));
    img80 = abs(ifft2(imgfft .* H80));
    figure;
    subplot(1,3,1); imshow(imscale01(img15));
    subplot(1,3,2); imshow(imscale01(img30));
    subplot(1,3,3); imshow(imscale01(img80));

    Text figure (4.26) Gaussian High-Pass Filtering


    %Matlab Code

    close all; clear all;


    img
    = im2double(imread('Fig4.11(a).jpg'));
    imgfft = fftshift(fft2(img));
    H15 = zeros(500,500); H30 = H15; H80 = H15;
    x = 500; y = 500; cx = x/2; cy = y/2;
    for i=1:x;
    for j=1:y;
    D(i,j)
    = sqrt([i-cx]^2 + [j-cy]^2);
    end
    end
    H15 = 1-exp(-.5*[D/15].^2);
    H30 = 1-exp(-.5*[D/30].^2);
    H80 = 1-exp(-.5*[D/80].^2);
    h15 = abs(ifft2(H15.*imgfft));
    h30 = abs(ifft2(H30.*imgfft));
    h80 = abs(ifft2(H80.*imgfft));
    figure;
    subplot(1,3,1); imshow(imscale01(h15));
    subplot(1,3,2); imshow(imscale01(h30));
    subplot(1,3,3); imshow(imscale01(h80));

    Text figure (4.28) Laplacian Filtering


    %Matlab Code

    close all; clear all;


    img
    = im2double(imread('Fig4.28(a).jpg'));
    imgfft = fftshift(fft2(img));
    [x,y] = size(img); cx = x/2; cy = y/2;
    for i=1:x;
    for j=1:y;
    D(i,j)
    = -((i-cx)^2 + (j-cy)^2);
    end
    end
    lapimg = ifft2(D.*imgfft);
    figure;
    subplot(2,2,1);
    subplot(2,2,2);
    subplot(2,2,3);
    subplot(2,2,4);

    imshow(imscale01(img));
    imshow(imscale01(abs(lapimg)));
    imshow(imscale01((real(lapimg))));
    imshow(imscale01(real(img - abs(lapimg))));

    Text figure (4.29) High-boost Filtering

    %Matlab Code Note: I used a Gaussian HP Filter since I dont trust my Laplacian
    close all; clear all;
    img
    = im2double(imread('Fig4.29(a).jpg'));
    imgfft = fftshift(fft2(img));
    [x,y] = size(img); cx = x/2; cy = y/2;
    k = zeros(x,y);
    for i=1:x;
    for j=1:y;
    D(i,j)
    = sqrt([i-cx]^2 + [j-cy]^2);
    k(i,j) = (-1)^(i+j);
    end
    end
    H
    = 1-exp(-.5*[D/50].^2);
    himg = real(ifft2(H.*imgfft)).*k;
    img1
    = img + himg;
    img17
    = 1.7*img + himg;
    figure;
    subplot(2,2,1);
    subplot(2,2,2);
    subplot(2,2,3);
    subplot(2,2,4);

    imshow(imscale01(img));
    imshow(imscale01(himg));
    imshow(imscale01(img1));
    imshow(imscale01(img17));

    Text figure (4.30) Multiple-operation Filtering

    %Matlab Code
    close all; clear all;
    img
    = im2double(imread('Fig4.30(a).jpg'));
    imgfft = fftshift(fft2(img));
    BHPF
    = zeros(size(imgfft));
    x = size(BHPF,1); y = size(BHPF,2); cx = x/2; cy = y/2;
    Do = floor(0.05*x);
    for i=1:x;
    for j=1:y;
    D(i,j)
    = sqrt([i-cx]^2 + [j-cy]^2);
    if(D(i,j) == 0) D(i,j)=0.01; end
    BHPF(i,j) = 1 / (1 + [Do/D(i,j)]^(4));
    end
    end
    imgBHPF = abs(ifft2(imgfft .* (BHPF)));
    imgHFE
    = abs(ifft2(imgfft .* (0.5 + 2*BHPF)));
    imgHFEeq = floor(255*imscale01(imgHFE));
    [hist1, bins1] = hist(double(imgHFEeq(:)),0:255);
    hist1 = hist1./length(imgHFEeq(:));
    CDF1 = cumsum(hist1);
    img1eq = zeros(size(imgHFEeq));
    for i=0:255
    img1eq(find(imgHFEeq==i)) = CDF1(i+1);
    end
    figure;
    subplot(2,2,1);
    subplot(2,2,2);
    subplot(2,2,3);
    subplot(2,2,4);

    imshow(imscale01(img));
    imshow(imgBHPF+.4);
    imshow(imscale01(imgHFE));
    imshow(imscale01(img1eq));

    Text figure (4.33) Homomorphic Filtering


    %Matlab Code

    close all; clear all;


    img
    = im2double(imread('Fig4.29(a).jpg'));
    img2 = log(img+1);
    imgfft = fftshift(fft2(img2));
    H = zeros(size(img));
    [x,y]=size(H); cx = x/2; cy = y/2;
    Gl = .5; Gh = 2; Gdiff = Gh - Gl; c=1; Do = 10;
    k = zeros(x,y);
    for i=1:x
    for j=1:y
    D(i,j)
    = sqrt([i-cx]^2 + [j-cy]^2);
    H(i,j) = Gdiff*[1 - exp(-c*(D(i,j)/Do)^2)] + Gl;
    k(i,j) = (-1)^(i+j);
    end
    end
    imgen = real(exp(ifft2(H.*imgfft))-1).*k;
    figure;
    subplot(1,2,1); imshow(imscale01(img));
    title('Original Image');
    subplot(1,2,2); imshow(imscale01(imgen));
    title('Homomorphic Image: c=1, D_o=10, G_l=.5, G_h=2');

    Original Image

    Homomorphic Image: c=1, Do=10, Gl=.5, Gh=2

    Acknowledge: I would like to thank my former student Brad Ratliff for providing these
    images.

    You might also like