基于黑暗通道算法的图像去雾matlab仿真通过对话框选择测试场景-源码

clc; clear; close all; warning off; addpath(genpath(pwd)); %load test_picture [fName pName]=uigetfile({'*.jpg'},'Open'); if fName original_im=imread([pName fName]); %读入图片 end %formatting im to float original_im = double(original_im); % separate different channels - RGB R_channel = original_im(:, :, 1); G_channel = original_im(:, :, 2); B_channel = original_im(:, :, 3); %create original_dark_channel image with same size [m, n] = size(R_channel); dark_channel_image = zeros(m,n); %extract the minimum value of each point in RGB for dark_channel_image for i=1:m for j=1:n local_pixels =[R_channel(i,j), G_channel(i,j), B_channel(i,j)]; dark_channel_image(i,j) = min(local_pixels ); end end %image erode, minimum filtering kernel = ones(30); final_im = imerode(dark_channel_image, kernel); %transform dark_channel_imaege into Picture Format final_im = uint8(final_im); figure,subplot(1,2,1), imshow(final_im); title("Dark Channel Prior"); %definne the size of filter_window wid_x = 2; wid_y = 2; %augment image by pixels(value=128) augmented_im = ones(m+2*wid_x, n+2*wid_y, 3) * 128; augmented_im(wid_x+1:m+wid_x, wid_y+1:n+wid_y, :) = original_im; %find the value of Atmospheric light, which is the mean of top 0.1% value in %dark_channel_prior v_ac = reshape(dark_channel_image,1,m*n); v_ac = sort(v_ac, 'descend'); Ac = mean(v_ac(1:uint8(0.001*m*n))); %define and calculate minimum_matrix minimum = zeros(m+2*wid_x, n+2*wid_y); for i= wid_x+1 : m+wid_x for j=wid_y+1: n+wid_y %extract local_window local_window = augmented_im(i-wid_x:i+wid_x, j-wid_y:j+wid_y, :); %separate RGB channels local_r = local_window(:, :, 1); local_g = local_window(:, :, 2); local_b = local_window(:, :, 3); %normalize this current pixel in 3 channels channel_values = [ R_channel(i-wid_x,j-wid_y) / Ac, G_channel(i-wid_x,j-wid_y) / Ac, B_channel(i-wid_x,j-wid_y) / Ac ]; %find the min values in 3 channels minimum(i,j) = min(channel_values); end end %recover the augmented marix to previous size original_minimum = ones(m,n); original_minimum = minimum(wid_x+1:wid_x+m, wid_y+1:wid_y+n); %assign the local minimum for each point(Image Erode) kernel_erode = ones(2*wid_x+1,2*wid_y+1); min_minimum = imerode(original_minimum, kernel_erode); %define w as a parameter for adjustment w=0.95; %define and calculate transmittance matrix pre_t = ones(m,n); true_t = pre_t - w*min_minimum; subplot(1,2,2), imshow(uint8(true_t*255)); title("Transmittance map"); %set up a threshold for light transmittance t0=0.1; for i=1:m for j=1:n true_t(i,j) = max(t0, true_t(i,j)); end end %recover the final image by haze function final_image = (original_im - Ac) ./ true_t +Ac; final_image = uint8(final_image); %show the result figure; subplot(1,2,1), imshow(uint8(original_im)); title("Original image"); subplot(1,2,2),imshow(final_image); title("After processing");