图像增强基于matlab GSA灰度图像增强含Matlab源码 1172期

Posted 2021-08-30 紫极神光

一、简介

基于matlab GSA灰度图像增强：引力搜索算法（GSA）是一种基于引力和质量相互作用定律的优化算法。该算法基于牛顿引力：“宇宙中的每个粒子都以与力乘积成正比的力吸引每个其他粒子。它们的质量与它们之间的距离的平方成反比”。

二、源代码

% Using the Regional Similarity Transformation Function and Dragonfly Algorithm. 

%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%THIS IS A DEMO FOR NOVEL IMAGE ENHANCEMENT USING GRAV脻TAT脻ONAL SEARCH ALGORITHM 
clc;
clear all;
close all;

%READ IMAGE
i=imread('test5.bmp');
i=imresize(i,1);  

%--------------------------------------------------------------------------
%GLOBAL MEAN
frekans=zeros(256,1);

% It records by counting how many times each pixel value is.

for k=1:size(i,1)
    for l=1:size(i,2)

        value=i(k,l);
       frekans(value+1)=frekans(value+1)+1; 

    end
end
deger=max(frekans);
for k=1:256
    if deger==frekans(k)
        D=k;
    end
end
D=double(D/255);
image = im2double(i);
global_mean = D;
%--------------------------------------------------------------------------
% LOCAL MEAN
Bmean = mean_n(image);
%--------------------------------------------------------------------------
%SIMILARITY
[S]=similarity(i);
B=double(S);
%--------------------------
C = std_n(image,Bmean);
im_size = size(image);
%-------------------------------------------------------------------------
%OPTIMIZATION PARAMETERS
N=15; % Agent number;
Max_iteration=15;
%-------------------------------------------------------------------------
%CALL FUNCTION 

[parameters Fbest BestChart MeanChart] = GSA_enhancement2(image,global_mean,B,C,im_size, N, Max_iteration);

enh = trans(i,image, B, C, global_mean, parameters(1),parameters(2),parameters(3),parameters(4) );

function [S]= similarity(x)


[xlen ylen] =size(x);

S=zeros(xlen, ylen);

for i=2:1:xlen-1
    for j=2:1:ylen-1
        
            %--------------------------------------------------------
        %Getting the values in the mask
        maskGray=zeros(1,9) ;
        resim=zeros(1,9) ; 
          for b=1:9;
          [m]=ikomsu(b,i);
          [n]=jkomsu(b,j);
          [res]=resimdemi(m,n,xlen,ylen);
          resim(b)=res;
                        if(resim(b)==1)
                            gray= x(m,n); 
                            maskGray (b) =gray;
                        else
                            maskGray (b) =0;
                        end
          end
  % Distance and similarity are calculated between 9 pixels in the mask and the result is determined.
       DN=150;
       sim=0;
       sum=0;
      for b=1:9
                   sim (b)=similarity1(maskGray(b),maskGray(9),DN);
                   sum=sim(b)+sum;    
      end
         %---------------------------------------------------------
     
         
          S(i,j)=sum/9;
          
              
    end
end 


end
%------------------------------------------------------------------------
%FUNCTIONS
%-----------------------------------------------------------------------
function [ res ] = resimdemi(x,y,w,h )

 if(x<1 || x>w-1 || y<1 || y>h-1)
    res=0;
 else
    res=1;


 end
end
 function [x] = ikomsu( komsuno, i)


 if(komsuno ==1)
     x=i-1;
 elseif(komsuno ==2)
    x= i;
 elseif(komsuno ==3)
    x= i+1;
 elseif(komsuno ==4)
    x= i-1;
 elseif(komsuno ==5)
    x= i+1;
 elseif(komsuno ==6)
   x= i-1;
 elseif(komsuno ==7)
   x= i;
 elseif(komsuno ==8)
   x= i+1;
 elseif(komsuno ==9)
    x=i;
 else
 x=0;
 end


 end

 function [y] = jkomsu( komsuno, j )

 if(komsuno ==1)
    y= j-1;
 elseif(komsuno ==2)
    y= j-1;
 elseif(komsuno ==3)
   y= j-1;
 elseif(komsuno ==4)
   y= j;
 elseif(komsuno ==5)
   y= j;
 elseif(komsuno ==6)
   y= j+1;
 elseif(komsuno ==7)
   y= j+1;
 elseif(komsuno ==8)
   y= j+1;
 elseif(komsuno ==9)
    y= j;
 else
   y=0;
 end


 end

 
function [sim] = similarity1( k, o, DN)

 
 a=0;b=0;c=0;
 a=abs(k-o);
%  b=(abs(l-p))^2;
%  c=(abs(m-r))^2;
%  //Distance calculation based on the Euclidean relation
 dist=a;

% %  //Calculation of similarity based on linear function
 arahes=dist/DN;
 sim=1 - arahes;
%  sim=exp(-dist/DN);


end

三、运行结果

四、备注

版本：2014a