saving values of one array into another array using for loop statement

It may be little complicated to explain for me this but I'm trying my best, if not quite clear then please ask me again for clarification. So, I have some 3-4 array to process in this problem. First array is 'centroids' (mx2 double). My second array is 'point3D' (437x517x3 single). Now using the values of 'centroids' I want to find out the corresponding values of 'point3D' and store these values in another array named 'co_ord'. For example say in 'centroids' for X1 and Y1, I want to extract corresponding values in 'point3D' and so on upto Xm and Ym and store them into 'point3D' for further use. I got 'centroids' array through Image processing and 'point3D' array from stereo vision.

Code which I tried is as

 load('C:\Users\Naseeb\Desktop\centroids.mat'); % load centroids file
load('C:\Users\Naseeb\Desktop\point3D.mat'); %load point3D file
centroids = ceil(centroids); % change values into integer
x = cell(size(centroids,1),1); % create array X for storing first column values
y = cell(size(centroids,1),1); % create array Y for storing second column values
 i = cell(size(centroids,1),1);
for i = 1:size(centroids,1) 
   for row = 1:size(centroids,1)
    for column = 1:size(centroids,2)
        x{i} = centroids(size(centroids,1),1); % store values of first column into array X
        y{i} = centroids(size(centroids,1),2); % store values of second column into array Y
        co_ord = point3D(x{i},y{i},:); % store values into co_ord file
    end
   end 
end

I tried in this way, you can suggest me direct way to extract values of 'point3D' from 'centroids' and save them in another array 'co_ord'.

Thanks

[Stephen23]

Please upload your data files: without them it is difficult for us to run your code or try any changes to it. You need to click the paperclip button, then both the Choose file and Attach file buttons.

[Naseeb Gill]

Hello Stephen, I attached both files to my question as you asked and I'm attaching those files with this also.

Thanks.

Answer 1

[Stephen23]

Try this:

% load the data:
matc = load('C:\Users\Naseeb\Desktop\centroids.mat');
matp = load('C:\Users\Naseeb\Desktop\point3D.mat');
ctrd = ceil(matc.centroids);
pt3D = matp.point3D;
% get the indices:
N = size(pt3D,3);
X = repmat(ctrd(:,1),1,N);
[Y,Z] = ndgrid(ctrd(:,2),1:N);
I = sub2ind(size(pt3D),X,Y,Z);
% extract the required data points:
out = pt3D(I);

where each row of out is one of the rows of centroids:

>> out
out =
      -0.13502      -0.35063        1.3853
     -0.044997       0.30485        1.2942
      -0.55011       0.85439        1.9908

The variable out contains the values of point3D at the index positions given in centroids, with the rows corresponding between out and centroids. Note that MATLAB is optimized for indexing, not for-loops, so I used much faster linear indexing to get obtain all of the values at once rather than using a slow and inefficient for-loop. Also note that it is recommended to load data into variables, rather than simply into the workspace. This allows tracking of the variables, prevents variables being overwritten without warning, and makes debugging easier.

[Naseeb Gill]

Thanks Stephen for your help.

Answer 2

[Naseeb Gill]

I also tried to solve above problem myself and get the following solution.

 load('C:\Users\Naseeb\Desktop\centroids.mat'); % load centroids file
load('C:\Users\Naseeb\Desktop\point3D.mat'); %load point3D file
centroids = ceil(centroids); % change values into integer
null=[];
Co_ordinates=[];  % final matrix where Co_ordinates will be stored
for step=1:3;
    temp=[];
    for k=1:length(centroids);
        null=point3D(centroids(k,1),centroids(k,2),step);
        temp=[temp null];
    end
    Co_ordinates=[Co_ordinates; temp];
    step=step+1;
end

[Stephen23]

Note that in general nested loops are a slow way of coding in MATLAB, and vectorized code (as per my answer) is a usually the faster, neater and more efficient option. This is why I wrote vectorized code for you.

Also note that you do not preallocate the arrays Co_ordinates and temp, and expand these inside loops. A much more memory-efficient solution would be to allocate those arrays before the loops and use indexing to allocate into them.

For more tips on writing efficient MATLAB code read this:

http://www.mathworks.com/help/matlab/matlab_prog/techniques-for-improving-performance.html

[Naseeb Gill]

Thanks Stephen for your valuable suggestions and I'm using your code only. I just tried from my side also and I get above solution. So, I just put this code also here.