How to loop on a data cube of an image

Hi, I'm trying to calculate gradient at each pixel in a data cube of an image an i'm getting this error "index out of bounds because size(X)=[280,307,191]."

Any help would be much appreciated Thank you

[Walter Roberson]

We are going to need to see some of your code.

[Rawan hamdi]

Okay,here's my code below:

close all
Bands =191;                  %no of bands
pixels =280*307;              %no of pixels
SIZE = [280,307 ,191];       %WxHxBands

lam = 0.063;
B=12;
X0 = multibandread('dc.tif', SIZE, 'int16',1252*2, 'bsq', 'ieee-le');

%adding noise
std_n=40; var_noise=std_n^2;  % Gaussian noise standard deviation
reduced_pw = 1.5*var_noise;   % power to reduce in first phase

std_n=sqrt(var(X0(:)));
Xn = randn(size(X0))*std_n; %noise
X = X0 + Xn; %Noisy Image

%looping on datacube
x = (X(1,:,1));
y = (X(:,1,1));
bands = X(1,1,:);

for  i = 1:1: x

      for j = 1:1: y

          for k = 1 :1:bands

         current_grad(i,j,k) = gradient(i,j,k);

          end

      end
  end

Answer 1

[Image Analyst]

Somewhere in your code where you're indexing x, y, and z, the value(s) of at least one of those indexes is greater than 280, 307, and 191, respectively.

[Rawan hamdi]

Okay and how to fix this? Thank you

[Image Analyst]

You will be using the debugger. Go here if you don't know how yet: http://blogs.mathworks.com/videos/2012/07/03/debugging-in-matlab/

Answer 2

[Walter Roberson]

You establish x, y, and bands according to the content of the noisy image, but then your triple nested for loop has you looping using those values as indices, such as in

for  i = 1:1: x

Question: in your line

current_grad(i,j,k) = gradient(i,j,k);

is "gradient" the gradient function? If so then why are you applying the function to the indices i, j, k, rather than to the content of the image? And does it make sense to be applying the gradient function only to a point at a time?

[Rawan hamdi]

i need to calculate the gradient at each pixel of the image in all bands and put it in a matrix which is the same size of the image X so i thought looping on all 3 dimensions of the image would do so? What should i do to get it right?

[Image Analyst]

Then why are you not using imgradient() or imgradientxy() on each spectral band?

[Rawan hamdi]

i tried using it but it's not defined in Matlab, all i could find it "gradient" function Is it related to specific version of Matlab?

[Walter Roberson]

imgradient() is part of the Image Processing Toolkit. That toolkit is included with Student Version licenses but is not installed by default.

[Image Analyst]

You must have a really old version. Does your license include upgrades? If so, do it.

[Rawan hamdi]

I think yes,would you help me do it pls cause i'm a bit new to matlab Thank you

[Walter Roberson]

Which MATLAB version are you using? And are you using the Student license?

[Image Analyst]

I cannot help you do the upgrade. You'll have to contact the Mathworks to do that.

Answer 3

[Rawan hamdi]

I've tried this:

x= squeeze(X(1,:,1));
y =squeeze(X(:,1,1));

b = X(1,1,:);

for i=1:x
    for j =1 : y

          for k = 1 :191

              G = gradient(1:i,1:j,1:k);
          end
      end

end

but it keeps telling me 'Index exceeds matrix dimensions'

[Walter Roberson]

You establish x, and y according to the content of the noisy image, but then your triple nested for loop has you looping using those values as indices.

Also, you are overwriting G in each iteration of the triple-nested loop.

[Rawan hamdi]

Okay , could you help me fix the code?