mainPso.m

function mainPso()
%% MAINPSO 
% Program to enhance image quality based on 
% Particle Swarm Optimization Alghoritm.
% 
% * Syntax 
%
% [] = MAINPSO()  
% 
% * Examples: 
% 
% Provide sample usage code here
% 
% * See also: 
% 
% <enhanceGsclImage.html ENHANCEGSCLIMAGE>, 
% <fitnessFunction.html FITNESSFUNCTION>, 
% <getImageSharpness.html GETIMAGESHARPNESS>
% 
% * Author: *Dmitrii Leliuhin*
% * Email: dleliuhin@mail.ru 
% * Date: 07/12/2018 11:04:25 
% * Version: 2.0 $ 
% * Requirements: PCWIN64, MatLab R2016a 
% 
% * Warning: 
% 
% # Warnings list. 
% 
% * TODO: 
% 
% # TODO list. 
% 

%% Code 

% Clear command window, workspace, plots.
clc;
clear all;
close all;

% Add subfolder with images.
addpath(strcat(pwd,'\images'), '-end');
addpath(strcat(pwd,'\sources'), '-end');

%% PSO Parameters

wmax=0.9; % Maximal inertia weight.
wmin=0.4; % Minimal inertia weight.

c1=2.4; % Cognitive acceleration coefficient.
c2=1.7; % Social acceleration coefficient.

swarmSize = 24; % Population size.
localSize = 3; % Local window size.
maxIterValue = 50; % Maximum number of iterations.

% Uniformly distributed generated numbers within range [0,1].
r1 = rand;
r2 = rand;

fit_val=[]; % Vector for storing fitness values.
P_best=[]; % Vector for storing pbest values.
G_best=[]; % Vector for storing gbest values.

pbest = 0; % Local best position of the group.
gbest = 0; % Global best position of the swarm.

% Reading specific image from subfolder.
I = imread('lena.jpg');

% Converting original image, I, to gray scale, G.
G = rgb2gray(I);

% Plot images.
figure;
imshow(I);
title('Original image', 'fontsize', 10);

figure;
imshow(G);
title('Grayscaled image', 'fontsize', 10);

% N = imnoise(G, 'salt & pepper', 0.02);
% figure;
% imshow(N);
% title('Noised image', 'fontsize', 10);

% Get image size m*n.
[n, m] = size(I);

%% Particle initialization

for i = 1:swarmSize
    % Updating optimum vales for particle position.
    a(i) = (1.5).*rand(1,1); % [0, 1.5]
    b(i) = (0.5).*rand(1,1); % [0, 0.5]
    c(i) = rand(1,1); % [0, 1]
    k(i) = 0.5 + 1.*rand(1,1); % [0.5, 1.5]
    % Updating particle velocity.
    v(i) = rand; % [0, 1]
    % Zeros vector of current positions of i'th particle.
    x(i) = 0;
end

%% Enhancing
for i = 1:maxIterValue
    % Calculating inertia weight.
    w = wmax -(wmax - wmin) * i / maxIterValue;
    
    for j = 1:swarmSize
        % Generating enhanced image by transformation function.
        E = enhanceGsclImage(G, localSize, a(j), b(j), c(j), k(j));
        
        % Calculating fitness value.
        fitness = fitnessFunction(E, m, n);
        
        % Add calculated fitness value to specific vector.
        fit_val = [fit_val, fitness];
        
        % Get max value in the vector of fitvalues.
        maxFit = max(fit_val);
        
        if (fitness < x(j))
            % Calculate pbest position.
            pbest = x(j);
            % Add calculated pbest value to specific vector.
            P_best = [P_best, pbest];
        else
            % Calculate pbest position.
            pbest = fitness;
            % Add calculated pbest value to specific vector.
            P_best = [P_best, pbest];
        end
    end
    
    gbest = max(P_best);
    
    for j = 1:swarmSize
        % Updating velocity.
        v(j) = w*v(j) + c1.*r1.*(P_best(j) - x(j)) + c2.*r2.*(gbest - x(j));
        
        % Updating particle position.
        x(j) = x(j) + v(j);
    end
    
    P_best = [];
end

% Plot enhanced image
figure;
imshow(E);
title('Enhanced image', 'fontsize', 10);

%% Comparing images

fprintf('Sharpness of Original image: %5.3f \n', ...
        mean(getImageSharpness(I)));
fprintf('Brightness of Original image: %5.3f \n', mean2(I));
fprintf('Contrast of Original image: %5.3f \n\n', max(I(:)) - min(I(:)));

fprintf('Sharpness of Grayscaled image: %5.3f \n', ...
        getImageSharpness(G));
fprintf('Brightness of Grayscaled image: %5.3f \n', mean2(G));
fprintf('Contrast of Grayscaled image: %5.3f \n\n', max(G(:)) - min(G(:)));

fprintf('Sharpness of Enhanced image: %5.3f \n', ...
        getImageSharpness(E));
fprintf('Brightness of Enhanced image: %5.3f \n', mean2(E));
fprintf('Contrast of Enhanced image: %5d \n\n', max(E(:)) - min(E(:)));

end