Check my codes for the paper titled MUTICONTOURLET BASED ADAPTIVE FUSION OF INFRARED AND VISIBLE REMOTE SENSING IMAGESVol 7,N0.3,July 2010.Authors- Xia Cheng,Licheng Jiao,Fang Liu

warning off;
clc;
clear all;
close all;
% Input Image
[filename, pathname] = uigetfile( '*.*', 'Load input image A');
I1= imread([pathname, filename]);
[filename, pathname] = uigetfile( '*.*', 'Load input image A');
I2= imread([pathname, filename]);
% Resizing Image
A1 = imresize(I1, [256 256]);
A2 = imresize(I2, [256 256]);
figure,imshow(A1);
title('Input Image 1');
figure,imshow(A2);
title('Input Image 2');
 i1=im2double(A1);
i2=im2double(A2);
% Prefiltering
L1 = medfilt2(i1,[3 3]);
L2 = medfilt2(i2,[3 3]);
figure,imshow(L1);
title('Prefiltered Image 1');
figure,imshow(L2);
title('Prefiltered Image 2');
% Decomposition
[b1]=GHM(L1);
[b2]=GHM(L2);
figure,imshow(b1);
title('GHM dec1');
figure,imshow(b2);
title('GHM dec2');
y1 = pdfbdec(b1, 'db1', 'haar', 4);
y2 = pdfbdec(b2, 'db1', 'haar', 4);
figure,imshow(y1{1,1})
title('PDFB dec1');
figure,imshow(y2{1,1})
title('PDFB dec2');
y11=y1{1,1};
y21=y2{1,1};
w=0.5;
SL=w*y11+(1-w)*y21;
figure,imshow(SL)
% Optimal Fusion Weight
e=0.01;
a=0;b=1;
w1=a+0.382*(b-a);
w2=a+b-w1;
% Determine Objective Function
T1=edge_association(y11,y21,SL,w1);
T2=edge_association(y11,y21,SL,w2);
if T1<T2
    a=w1;
     if abs(b-a)<e
         optw=(a+b)./2;
         return
     else
         w1=w2;
         w2=a+0.618*(b-a);
         T1=edge_association(y11,y21,SL,w1);
          T2=edge_association(y11,y21,SL,w2);
         goto(52)
     end
else
    b=w2;
     if abs(b-a)<e
         optw=(a+b)./2;
         return
     else
         w2=w1;
         w1=a+0.382*(b-a);
         T1=edge_association(y11,y21,SL,w1);
         T2=edge_association(y11,y21,SL,w2);
         goto(52)
     end
end
weight=optw;
% Fusion Rule of Low-Frequency Subbands
FL=weight*y11+(1-weight)*y21;
figure,imshow(FL)
% Fusion Rule of High-Frequency Directional Subbands
kernel=[1/16 1/16 1/16;1/16 1/2 1/16;1/16 1/16 1/16];
ELA1=sqrt(conv2((y1{1,2}{1,1}.^2),kernel,'same'));
ELB1=sqrt(conv2((y2{1,2}{1,1}.^2),kernel,'same'));
[r c]=size(ELA1);
for i=1:r
    for j=1:c
        if(ELA1(i,j)>=ELB1(i,j))
            D1(i,j)=y1{1,2}{1,1}(i,j);
        else
            D1(i,j)=y2{1,2}{1,1}(i,j);
        end
    end
end

ELA2=sqrt(conv2((y1{1,2}{1,2}.^2),kernel,'same'));
ELB2=sqrt(conv2((y2{1,2}{1,2}.^2),kernel,'same'));
[r c]=size(ELA2);
for i=1:r
    for j=1:c
        if(ELA2(i,j)>=ELB2(i,j))
            D2(i,j)=y1{1,2}{1,2}(i,j);
        else
            D2(i,j)=y2{1,2}{1,2}(i,j);
        end
    end
end

ELA3=sqrt(conv2((y1{1,2}{1,3}.^2),kernel,'same'));
ELB3=sqrt(conv2((y2{1,2}{1,3}.^2),kernel,'same'));
[r c]=size(ELA3);
for i=1:r
    for j=1:c
        if(ELA3(i,j)>=ELB3(i,j))
            D3(i,j)=y1{1,2}{1,3}(i,j);
        else
            D3(i,j)=y2{1,2}{1,3}(i,j);
        end
    end
end
ELA4=sqrt(conv2((y1{1,2}{1,4}.^2),kernel,'same'));
ELB4=sqrt(conv2((y2{1,2}{1,4}.^2),kernel,'same'));
[r c]=size(ELA4);
for i=1:r
    for j=1:c
        if(ELA4(i,j)>=ELB4(i,j))
            D4(i,j)=y1{1,2}{1,4}(i,j);
        else
            D4(i,j)=y2{1,2}{1,4}(i,j);
        end
    end
end
ELA5=sqrt(conv2((y1{1,2}{1,5}.^2),kernel,'same'));
ELB5=sqrt(conv2((y2{1,2}{1,5}.^2),kernel,'same'));
[r c]=size(ELA5);
for i=1:r
    for j=1:c
        if(ELA5(i,j)>=ELB5(i,j))
            D5(i,j)=y1{1,2}{1,5}(i,j);
        else
            D5(i,j)=y2{1,2}{1,5}(i,j);
        end
    end
end
ELA6=sqrt(conv2((y1{1,2}{1,6}.^2),kernel,'same'));
ELB6=sqrt(conv2((y2{1,2}{1,6}.^2),kernel,'same'));
[r c]=size(ELA6);
for i=1:r
    for j=1:c
        if(ELA6(i,j)>=ELB6(i,j))
            D6(i,j)=y1{1,2}{1,6}(i,j);
        else
            D6(i,j)=y2{1,2}{1,6}(i,j);
        end
    end
end
ELA7=sqrt(conv2((y1{1,2}{1,7}.^2),kernel,'same'));
ELB7=sqrt(conv2((y2{1,2}{1,7}.^2),kernel,'same'));
[r c]=size(ELA7);
for i=1:r
    for j=1:c
        if(ELA7(i,j)>=ELB7(i,j))
            D7(i,j)=y1{1,2}{1,7}(i,j);
        else
            D7(i,j)=y2{1,2}{1,7}(i,j);
        end
    end
end
ELA8=sqrt(conv2((y1{1,2}{1,8}.^2),kernel,'same'));
ELB8=sqrt(conv2((y2{1,2}{1,8}.^2),kernel,'same'));
[r c]=size(ELA8);
for i=1:r
    for j=1:c
        if(ELA8(i,j)>=ELB8(i,j))
            D8(i,j)=y1{1,2}{1,8}(i,j);
        else
            D8(i,j)=y2{1,2}{1,8}(i,j);
        end
    end
end
ELA9=sqrt(conv2((y1{1,2}{1,9}.^2),kernel,'same'));
ELB9=sqrt(conv2((y2{1,2}{1,9}.^2),kernel,'same'));
[r c]=size(ELA9);
for i=1:r
    for j=1:c
        if(ELA9(i,j)>=ELB9(i,j))
            D9(i,j)=y1{1,2}{1,9}(i,j);
        else
            D9(i,j)=y2{1,2}{1,9}(i,j);
        end
    end
end
ELA10=sqrt(conv2((y1{1,2}{1,10}.^2),kernel,'same'));
ELB10=sqrt(conv2((y2{1,2}{1,10}.^2),kernel,'same'));
[r c]=size(ELA10);
for i=1:r
    for j=1:c
        if(ELA10(i,j)>=ELB10(i,j))
            D10(i,j)=y1{1,2}{1,10}(i,j);
        else
            D10(i,j)=y2{1,2}{1,10}(i,j);
        end
    end
end
ELA11=sqrt(conv2((y1{1,2}{1,11}.^2),kernel,'same'));
ELB11=sqrt(conv2((y2{1,2}{1,11}.^2),kernel,'same'));
[r c]=size(ELA11);
for i=1:r
    for j=1:c
        if(ELA11(i,j)>=ELB11(i,j))
            D11(i,j)=y1{1,2}{1,11}(i,j);
        else
            D11(i,j)=y2{1,2}{1,11}(i,j);
        end
    end
end
ELA12=sqrt(conv2((y1{1,2}{1,12}.^2),kernel,'same'));
ELB12=sqrt(conv2((y2{1,2}{1,12}.^2),kernel,'same'));
[r c]=size(ELA12);
for i=1:r
    for j=1:c
        if(ELA12(i,j)>=ELB12(i,j))
            D12(i,j)=y1{1,2}{1,12}(i,j);
        else
            D12(i,j)=y2{1,2}{1,12}(i,j);
        end
    end
end

ELA13=sqrt(conv2((y1{1,2}{1,13}.^2),kernel,'same'));
ELB13=sqrt(conv2((y2{1,2}{1,13}.^2),kernel,'same'));
[r c]=size(ELA13);
for i=1:r
    for j=1:c
        if(ELA13(i,j)>=ELB13(i,j))
            D13(i,j)=y1{1,2}{1,13}(i,j);
        else
            D13(i,j)=y2{1,2}{1,13}(i,j);
        end
    end
end
ELA14=sqrt(conv2((y1{1,2}{1,14}.^2),kernel,'same'));
ELB14=sqrt(conv2((y2{1,2}{1,14}.^2),kernel,'same'));
[r c]=size(ELA14);
for i=1:r
    for j=1:c
        if(ELA14(i,j)>=ELB14(i,j))
            D14(i,j)=y1{1,2}{1,14}(i,j);
        else
            D14(i,j)=y2{1,2}{1,14}(i,j);
        end
    end
end
ELA15=sqrt(conv2((y1{1,2}{1,15}.^2),kernel,'same'));
ELB15=sqrt(conv2((y2{1,2}{1,15}.^2),kernel,'same'));
[r c]=size(ELA15);
for i=1:r
    for j=1:c
        if(ELA15(i,j)>=ELB15(i,j))
            D15(i,j)=y1{1,2}{1,15}(i,j);
        else
            D15(i,j)=y2{1,2}{1,15}(i,j);
        end
    end
end
ELA16=sqrt(conv2((y1{1,2}{1,16}.^2),kernel,'same'));
ELB16=sqrt(conv2((y2{1,2}{1,16}.^2),kernel,'same'));
[r c]=size(ELA16);
for i=1:r
    for j=1:c
        if(ELA16(i,j)>=ELB16(i,j))
            D16(i,j)=y1{1,2}{1,16}(i,j);
        else
            D16(i,j)=y2{1,2}{1,16}(i,j);
        end
    end
end
t111={D1,D2,D3,D4,D5,D6,D7,D8,D9,D10,D11,D12,D13,D14,D15,D16};
tt={FL,t111};
% Reconstruction
x = pdfbrec(tt, 'db1', 'haar');
figure,imshow(x);
F=IGHM(x);
figure,imshow(F);
title('Reconstructed Image');
% Postfiltering
F = medfilt2(F,[3 3]);
figure,imshow(F);
title('Fused Image');

[Andreas Goser]

What's the goal of this? I cannot see a question.

[Jan]

I do not expect, that anybody will debug your code, if you do not explain with any details, which problems occur.

If a problem occur, post just the relevant section of the code together with either the complete error message or an exact description of the difference between the results and your expectations.

[John D'Errico]

So you expect people to read the paper, figure out what you want to do, then verify that this complete mess of code does what it should do? Yeah, right. We will jump right to it.

[Walter Roberson]

The paper is available (amongst other places) at http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=5427044

The above code calls upon the undefined function "edge_association". The paper itself does not refer to edge association at all, and seldom even refers to edges.

Answer 1

[Jan]

Starting a program with this header is suspicious:

warning off;
clc;
clear all;
close all;

• Disable all warnings? There should be a smarter way to get rid of warnings.
• About the clear all see good programming practice.
• I do not see a reason to close other GUIs.

Later I see some variables with an index in the name, e.g. ELA5, ELB5. It is much neater to use a cell array instead.

I think, these are poor programming techniques. I did not check, if the program creates correct results. But for reading and maintenance the code has a low quality only.

[rajesh veerabadran]

Respected Sir

I doimg my M.TECH in bsa university.I need some updation on this above project.For updation what algorithm we can use.

[Jan]

What do you mean by "updation"? Do you want to modify the above code? Then my advice is: Do not do this. The quality of the code is so low that deleting and re-writing it from the scratch will be an improvement already.

Please consider the comments by Walter, John and Andreas.

Answer 2

[rajesh veerabadran]

JAN SIMPSON SIR,

I want to use markov random principle in multicontourlet based image fusion.Is it possible to implement mrf in above source code.

[Walter Roberson]

No, the code posted here relies on an undefined function "edge_association" whose purpose cannot be determined by reference to the paper the code is supposedly an implementation of. The code must, until proven otherwise, be treated as having been written for some other purpose than multicontourlet based image fusion.