【图像分割】基于压缩系数的粒子群和重力搜索算法实现图像的多级阈值分割matlab源码...

1 简介

提出了基于粒子群优化(PSO)与引力搜索(GSA)混合算法(PSOGSA)的多阈值图像分割方法来解决图像阈值搜寻过程中单一优化算法局部搜索能力不强的问题.在阈值寻优过程中提高群体多样性,增强了全局搜索能力;采用了全局最优解的正态变异策略,扩展了全局最优的搜索区域,避免了算法的早熟收敛.在此基础上,实现了基于粒子群与引力搜索混合算法的多阈值图像分割方法.最后,使用本方法对复杂多目标图像进行了多阈值分割实验,并与引力搜索算法和萤火虫算法进行了比较.实验结果表明,本文方法的分割精度优于引力搜索算法与萤火虫算法,其分割目标函数值在连续运行时的标准差降低了90%以上,是一种精度高,稳定性强的多阈值图像分割方法.

2 部分代码

clear all

close all

clc

% Parameter initialization

%      I = imread('Aeroplane.tiff');

   I = imread('Cameraman.tiff');

  level = 5; %% Threshold = level-1 

% 

 N_PAR = level;                          %number of thresholds (number of levels-1) (dimensiones)

 dim = N_PAR;  

% 

 n = 15;                                  % Size of the swarm " no of objects " %%% Default (n = 15)

 Max_Iteration  = 300;                    % Maximum number of "iterations"      %%% Default (Max_Iteration  = 300)

% 

if size(I,3) == 1 %grayscale image

[n_countR, x_valueR] = imhist(I(:,:,1));

end

Nt = size(I,1) * size(I,2); 

% % Lmax indicated color segments 0 - 256

Lmax = 256;   %256 different maximum levels are considered in an image (i.e., 0 to 255)

for i = 1:Lmax

    if size(I,3) == 1  

        %grayscale image

        probR(i) = n_countR(i) / Nt;

    end

end

if size(I,3) == 1

    up = ones(n,dim) * Lmax;

    low = ones(n,dim);

end

 tic

 RunNo  = 1;   

    for k = [ 1 : RunNo ]  

       [CPSOGSA_bestit,CPSOGSA_bestF,CPSOGSA_Fit_bests]= CPSOGSA(I, Lmax, n,Max_Iteration,low,up,dim, level, probR);

       BestSolutions1(k) = CPSOGSA_bestF; 

 disp(['Run # ' , num2str(k),'::' 'Best estimates =',num2str(CPSOGSA_bestit)]);         % CPSOGSA

    end  

% /* Boxplot Analysis */

   figure

   boxplot([BestSolutions1'],{'CPSOGSA'});

   color = [([1 0 0])];

   h = findobj(gca,'Tag','Box'); 

   for j=1:length(h) 

   patch(get(h(j),'XData'),get(h(j),'YData'),color(j));

   end 

   title ('\fontsize{15}\bf Aeroplane (k=2)');

   % %  title ('\fontsize{15}\bf  Cameraman (k=2)');

   xlabel('\fontsize{15}\bf Algorithms');

   ylabel('\fontsize{15}\bf Best Fitness Values');

   box on

% % % 

% /* Graphical Analysis*/

figure

 plot(CPSOGSA_Fit_bests,'DisplayName','CPSOGSA','Color','b','LineStyle','-','LineWidth',3);

 disp( ['Time_CPSOGSA =', num2str(toc)]); 

 title ('\fontsize{15}\bf Aeroplane (k=2)'); % k=2,4,6,8,10

 % %  title ('\fontsize{15}\bf Cameraman (k=2)');

 xlabel('\fontsize{15}\bf Iterations');

 ylabel('\fontsize{15}\bf Fitness values');

 legend('\fontsize{12}\bf CPSOGSA');

 %

 %

 gBestR = sort(CPSOGSA_bestit);

 Iout = imageGRAY(I,gBestR);

 Iout2 = mat2gray(Iout); 

% % Show results on images  

figure

imshow(Iout)

figure

imshow(I)

% % Show results

intensity = gBestR(1:dim-1);  

STDR  = std(CPSOGSA_Fit_bests)              %Standard deviation of fitness values       

MSEV = MSE(I, Iout)                         %Mean Square Error

PSNRV = PSNR(I, Iout)                       %PSNR between original image I and the segmented image Iout

SSIMV = ssim (I, Iout)                      %SSIM Quality Measure

FSIMV = FeatureSIM (I, Iout)                %FSIM Quality Measure

Best_Fitness_Value= CPSOGSA_Fit_bests(k)    %Best fitness

% % Plot the threshold values over the histogram

figure 

plot(probR)

    hold on

vmax = max(probR);

for i = 1:dim-1

    line([intensity(i), intensity(i)],[0 vmax],[1 1],'Color','r','Marker','.','LineStyle','-')

    title ('\fontsize{15}\bf Aeroplane (k=2)');

% title ('\fontsize{15}\bf  Cameraman (k=2)');

    xlabel('\fontsize{15}\bf Gray level');

    ylabel('\fontsize{15}\bf Frequency');

    hold off

end    

3 仿真结果

4 参考文献

[1]巢渊, 戴敏, 陈恺,等. 基于广义反向粒子群与引力搜索混合算法的多阈值图像分割[J]. 光学精密工程, 2015, 23(3):8.

博主简介：擅长智能优化算法、神经网络预测、信号处理、元胞自动机、图像处理、路径规划、无人机等多种领域的Matlab仿真，相关matlab代码问题可私信交流。