Thread Subject: Shadow Removal Using Entropy minimization

Hi all,

I am trying to implement a shadow removal algorithm in matlab. I follow the paper
written by Mr.Graham.Finalyson i.e, "Shadow removal using Entropy minimization".
First of all i am trying to get 1-D illuminant invariant image by assuming that we
know the projection angle. I took some known images from the above mentioned
paper so that i know the projection angle and there is no need of entropy calculation.
The program which wrote works with some images and seems to be failed with some images. Please help me if some one has some experience with this algorithm. I am appending small part of my matlab code which is written to produce a 1-D illumination invariant image.
------------------------------------------------------------------------------------------------------------------

       I = imread('file1.tiff');
       J = im2double(I);

      R = J(:,:,1);
      G = J(:,:,2);
      B = J(:,:,3);

     [len,wid] = size(R);

     % Generation of 2-D Log Chromaticity Image.
     for i = 1:len
        for j = 1:wid
           if ((R(i,j)*G(i,j)*B(i,j))~= 0)
              c1(i,j) = R(i,j)/((R(i,j)*G(i,j)*B(i,j))^(1/3));
              c2(i,j) = G(i,j)/((R(i,j)*G(i,j)*B(i,j))^(1/3));
              c3(i,j) = B(i,j)/((R(i,j)*G(i,j)*B(i,j))^(1/3));
           else
              c1(i,j) = 1;
              c2(i,j) = 1;
              c3(i,j) = 1;
        end
    end
end

rho1 = mat2gray(log(c1));
rho2 = mat2gray(log(c2));
rho3 = mat2gray(log(c3));

X1 = mat2gray(rho1*1/(sqrt(2)) - rho2*1/(sqrt(2))); %(1/sqrt(2); -1/sqrt(2); 0)
X2 = mat2gray(rho1*1/(sqrt(6)) + rho2*1/(sqrt(6)) - rho3*2/(sqrt(6))); %(1/sqrt(6); 1/sqrt(6); -2/sqrt(6))

theta = 120;

InvariantImage = cos(theta*pi/180)*X1 + sin(theta*pi/180)*X2;

-----------------------------------------------------------------------------------------------------

Thank you all,
Sandeep P

"Sandeep " <sandeepdevasrii@gmail.com> wrote in message <hafchq$bl1$1@fred.mathworks.com>...
>
> Hi all,
>
> I am trying to implement a shadow removal algorithm in matlab. I follow the paper
> written by Mr.Graham.Finalyson i.e, "Shadow removal using Entropy minimization".
> First of all i am trying to get 1-D illuminant invariant image by assuming that we
> know the projection angle. I took some known images from the above mentioned
> paper so that i know the projection angle and there is no need of entropy calculation.
> The program which wrote works with some images and seems to be failed with some images. Please help me if some one has some experience with this algorithm. I am appending small part of my matlab code which is written to produce a 1-D illumination invariant image.
> ------------------------------------------------------------------------------------------------------------------
>
> I = imread('file1.tiff');
> J = im2double(I);
>
> R = J(:,:,1);
> G = J(:,:,2);
> B = J(:,:,3);
>
> [len,wid] = size(R);
>
> % Generation of 2-D Log Chromaticity Image.
> for i = 1:len
> for j = 1:wid
> if ((R(i,j)*G(i,j)*B(i,j))~= 0)
> c1(i,j) = R(i,j)/((R(i,j)*G(i,j)*B(i,j))^(1/3));
> c2(i,j) = G(i,j)/((R(i,j)*G(i,j)*B(i,j))^(1/3));
> c3(i,j) = B(i,j)/((R(i,j)*G(i,j)*B(i,j))^(1/3));
> else
> c1(i,j) = 1;
> c2(i,j) = 1;
> c3(i,j) = 1;
> end
> end
> end
>
> rho1 = mat2gray(log(c1));
> rho2 = mat2gray(log(c2));
> rho3 = mat2gray(log(c3));
>
> X1 = mat2gray(rho1*1/(sqrt(2)) - rho2*1/(sqrt(2))); %(1/sqrt(2); -1/sqrt(2); 0)
> X2 = mat2gray(rho1*1/(sqrt(6)) + rho2*1/(sqrt(6)) - rho3*2/(sqrt(6))); %(1/sqrt(6); 1/sqrt(6); -2/sqrt(6))
>
> theta = 120;
>
> InvariantImage = cos(theta*pi/180)*X1 + sin(theta*pi/180)*X2;
>
> -----------------------------------------------------------------------------------------------------
>
> Thank you all,
> Sandeep P

Hi,
at the end did you find the solution of this problem?

I'm as you with the same problem. I want to implement the algorithm to find the minimum entropy value....

hi,i use this code to plot Entropy :

X1 = mat2gray(rho1*1/(sqrt(2)) - rho2*1/(sqrt(2)));
X2 = mat2gray(-rho1*sqrt(6)/4 + rho2*sqrt(6)/4 + rho3*(sqrt(6)/4));
for t=1:180

    delta = t*pi/180;
    img = cos(delta)*X1 + sin(delta)*X2 ;
    
    logimg = -log(img);
    r=img.*logimg;
    
    n(t)=mean(abs(r(:)));
end;

plot(abs(n(:)), 'DisplayName', 'n', 'YDataSource', 'n'); figure(gcf)
[v,idx]=min(n(:));

result are inconsistent amongst the papers. may be the paper use some high precision
bmp.

hi,i use this code to plot Entropy :

X1 = mat2gray(rho1*1/(sqrt(2)) - rho2*1/(sqrt(2)));
X2 = mat2gray(-rho1*sqrt(6)/4 + rho2*sqrt(6)/4 + rho3*(sqrt(6)/4));
for t=1:180

    delta = t*pi/180;
    img = cos(delta)*X1 + sin(delta)*X2 ;
    
    logimg = -log(img);
    r=img.*logimg;
    
    n(t)=mean(abs(r(:)));
end;

plot(abs(n(:)), 'DisplayName', 'n', 'YDataSource', 'n'); figure(gcf)
[v,idx]=min(n(:));

result are inconsistent amongst the papers. may be the paper use some high precision
bmp.

hi :
I want know how to solve the PDE in this paper, who can help me ? Thanks all.

On 10/18/2010 2:05 AM, hao yu wrote:
> hi :
> I want know how to solve the PDE in this paper, who can help me ? Thanks all.

Sure. Where is the paper btw?

--Nasser

 it is availialbe in IEEE. also you can google it" Shadow Removal Using Entropy minimization, IJCV 2009".
 If you want a talk, my MSN : yuhao@tju.edu.cn,
thanks