How to index a for loop with negative subscript indices?

Dear MATLAB-users,

could anyone help me by indexing correct subscript indices in a for loop? Aim of the loop is to summarize several gradient images. Therefore, a for loop should run in integer steps from a given range from "-P" to "+P". When running the script, this error occurs: "Subscript indices must either be real positive integers or logicals." Unfortunately, I have to use the given parameter (P), because it is used at different parts of the main script. In the following a simple code of the described problem is attached.

Could anyone tell me how to calculate the sum of the several gradient images?

Many thanks in advance and best regards!

I = im2double(image.tif);
P=2;
I_sum = zeros(16,64);
for l = -P:P
   % SHIFT horizontal (l)
   I_temp = circshift(I, [0, l]);                                                  
   % figure('Name','l-shifted image'); imshow(I_temp);
   % SUBTRACT horizontally shifted HR-estimation from initial image
   I_Gradient = I - I_temp;                                                                  
   % figure('Name','Gradient image'); imshow(I_Gradient);
   % ANTISHIFT horizontal (-l)
   I_temp = circshift(I_Gradient, -[0, l]);
   % figure('Name','Backshifted Gradient image'); imshow(I_temp);
   I_sum_(l) = I_temp;
   I_sum = I_sum + I_sum(l);
   % figure('Name','I_sum'); imshow(I_sum);
end

Answer 1

[Azzi Abdelmalek]

You can't use negative indices in Matlab, but this is not a problem. Look at this example

k=0
for l = -P:P
  k=k+1
  A(k)=l
end

[Henrik]

Thank you for your fast answer. Unfortunately, it doesn't solve my problem. I have implemented it like this:

I = im2double(image.tif'));
P=2;
I_sum = zeros(16,64);
k=0;

for l = -P:P
    k=k+1; 
    % SHIFT horizontal (l)
     I_temp = circshift(I, [0, l]);                                                  
     % figure('Name','l-shifted image'); imshow(I_temp);
     % SUBTRACT horizontally shifted HR-estimation from initial image
     I_Gradient = (I - I_temp);                                                                  
     % figure('Name','Gradient image'); imshow(I_Gradient);
     % ANTISHIFT horizontal (-l)
     I_temp = circshift(I_Gradient, -[0, l]);
     % figure('Name','Backshifted Gradient image'); imshow(I_temp);
     I_sum_(k) = I_temp(k);
     I_sum = I_sum + I_sum(k)
     % figure('Name','I_sum'); imshow(I_sum);
end

But the result is not comprehensible to me. The result of my sum-operation should be an image with highlighted edges. Instead I get a black image as output. Could you or anybody else have a look on my code and check if there is a mistake?

Answer 2

[Walter Roberson]

Use this code pattern:

l_vals = -P:P;
num_l = length(l_vals);
for l_idx = 1 : num_l
  l = l_vals(l_idx);
  ...
  I_sum(l_idx) = ...
  ...
end

That is, prepare a variable with a list of all of the values you want to iterate over, in the order you want to iterate over them. Then iterate an index from 1 to the length of that list. At each step, use the index to pull out the particular value for this iteration, for use in your calculations. When you want to store a value that is associated with that particular value, use the index value as the offset to store it.

Answer 3

[Henrik]

Thank you very much! Regarding the initial question both answers work!

The problem with getting a wrong summation image had its origin in the following rows: I_sum_(k) = I_temp(k); I_sum = I_sum + I_sum(k);

Replacing them with I_sum(:) = I_sum + I_temp; solves this problem.