wie.rar_CLS MATLAB_CLS restoration_filter_image restoration_res

%%Initialisation clear all; close all; clc; T=1; a=0.01; b=0.01; f=imread('lena.bmp'); [row_num col_num] = size(f) [U,V]=meshgrid(1:row_num, 1:col_num); U = U - floor(row_num/2); V = V - floor(col_num/2); K=0.0025; p_matrix = [0 -1 0; -1 4 -1; 0 -1 0]; gamma = 1e-4; pause %% %Generating Motion Blur Filter in Freq Domain %H_degrade=(T./(pi*(U*a + V*b+eps))) .* sin((pi*(U*a + V*b+eps))) .* exp(-1i*pi*(U*a + V*b+eps)); H_degrade = psf2otf(fspecial('motion',30,135), size(f)); imwrite(H_degrade,'Degrade_Filter.bmp'); figure; imshow(fftshift(H_degrade),[ ]); title('Motion Blur Filter'); hold on; pause H_wiener = (1./H_degrade) .* (abs(H_degrade).^2) ./ ((abs(H_degrade).^2) + K); imwrite(H_wiener,'Wiener_Filter.bmp'); figure; imshow(fftshift(H_wiener),[ ]); title('Wiener Filter'); hold on; P = fft2(p_matrix, row_num, col_num); H_cls = conj(H_degrade) ./ ((abs(H_degrade).^2) + (gamma* abs(P).^2)); imwrite(H_cls,'CLS_Filter.bmp'); figure; imshow(fftshift(H_cls),[ ]); title('CLS Filter'); hold on; %% %% %Processing Image for getting blurry image F=fft2(double(f)); G=H_degrade.*F;%G is the fft of blurred image g = real(ifft2(G)); % Crop to original size. %g =g(1:size(f, 1), 1:size(f, 2)); %Normalize the image %g=g/255; %g=im2uint8(g); imwrite(g,'Blurred.bmp'); figure; imshow(g,[]); title('Blurred Picture'); hold on; %% %Restoring image without noise g_new=g; G_new=fft2(g_new); F_restored=G_new./H_degrade; f_restored=real(ifft2(F_restored)); f_restored=f_restored(1:size(f, 1), 1:size(f, 2)); imwrite(f_restored,'Restored.bmp'); figure; imshow(f_restored,[]); title('Restored Picture'); hold on; %% %Adding image to noise g_normalized=g/max(max(g)); n_zeros = zeros(size(g)); n_high = imnoise(n_zeros,'gaussian',0,650/255^2); n_mid = imnoise(n_zeros,'gaussian',0,6.5/255^2); n_low = imnoise(n_zeros,'gaussian',0,0.0065/255^2); g_high=im2double(g_normalized)+n_high; g_mid=im2double(g_normalized)+n_mid; g_low=im2double(g_normalized)+n_low; figure; subplot(4,4,1); imshow(g,[]); title('no noise'); subplot(4,4,5); imshow(g_high,[]); title('high noise'); subplot(4,4,9); imshow(g_mid,[]); title('mid noise'); subplot(4,4,13); imshow(g_low,[]); title('low noise'); %% G_High= fft2(double(g_high)); G_Mid= fft2(double(g_mid)); G_Low= fft2(double(g_low)); F_Restored_High = G_High./H_degrade; F_Restored_Mid = G_Mid./H_degrade; F_Restored_Low = G_Low./H_degrade; f_restored_high=real(ifft2(F_Restored_High)); f_restored_mid=real(ifft2(F_Restored_Mid)); f_restored_low=real(ifft2(F_Restored_Low)); subplot(4,4,2); imshow(f_restored,[]); title('inverse'); subplot(4,4,6); imshow(f_restored_high,[]); title('inverse'); subplot(4,4,10); imshow(f_restored_mid,[]); title('inverse'); subplot(4,4,14); imshow(f_restored_low,[]); title('inverse'); %% G_High= fft2(double(g_high)); G_Mid= fft2(double(g_mid)); G_Low= fft2(double(g_low)); F_Wiener = G./H_wiener; F_Wiener_High = G_High./H_wiener; F_Wiener_Mid = G_Mid./H_wiener; F_Wiener_Low = G_Low./H_wiener; f_wiener=real(ifft2(F_Wiener)); f_wiener_high=real(ifft2(F_Wiener_High)); f_wiener_mid=real(ifft2(F_Wiener_Mid)); f_wiener_low=real(ifft2(F_Wiener_Low)); subplot(4,4,3); imshow(f_wiener,[]); title('wiener'); subplot(4,4,7); imshow(f_wiener_high,[]); title('wiener'); subplot(4,4,11); imshow(f_wiener_mid,[]); title('wiener'); subplot(4,4,15); imshow(f_wiener_low,[]); title('wiener'); %% %cls F_Cls = G./H_cls; F_Cls_High = G_High./H_cls; F_Cls_Mid = G_Mid./H_cls; F_Cls_Low = G_Low./H_cls; f_cls=real(ifft2(F_Cls)); f_cls_high=real(ifft2(F_Cls_High)); f_cls_mid=real(ifft2(F_Cls_Mid)); f_cls_low=real(ifft2(F_Cls_Low)); subplot(4,4,4); imshow(f_cls,[]); title('cls'); subplot(4,4,8); imshow(f_cls_high,[]); title('cls'); subplot(4,4,12); imshow(f_cls_mid,[]); title('cls'); subplot(4,4,16); imshow(f_cls_low,[]); title('cls');