Interpolation of a matrix in 2D for complex numbers

Hello

I have a 3D matrix of a field E_x_y_t. say it's 600 x 700 x 2048 (correlated to x y and t)

At some point I'm deleting part of the Field in x and y. No need to touch T:

E_x_y_t([1:length(x)/4, length(x)/4+length(x)/2+1:length(x)],:,:) = [];
E_x_y_t(:,[1:length(y)/4, length(y)/4+length(y)/2+1:length(y)],:) = [];
x = [x(length(x)/4+1:length(x)/4+length(x)/2)];

But what I really need is from this point to make X and Y again 600 x 700 on all the T. (make interpolation of the complex 2d field for each value of time (T) to smaller grid for better resolution).

How can I do it? using interp2 somehow i guess?

Answer 1

[KSSV]

You may use imresize.

Let A be your complex matrix:

B = imresize(A,2) ; % this makes double the size of A, you can specify the dimensions also 

You can also use interp2.

[m,n,p] = size(A) ; 
[X,Y] = meshgrid(1:n,1:m) ;

[Xi,Yi] = meshgrid(linspace(1,n,100),linspace(1,m,100)) ;    % give your desired numbers instead of 100
Br = zeros(100,100,p) ;
Bc = zeros(100,100,p) ; 
for i = 1:p
    Br(:,:,i) = interp2(X,Y,real(A(:,:,i)),Xi,Yi) ; 
    Bc(:,:,i) = interp2(X,Y,imag(A(:,:,i)),Xi,Yi) ; 
end
B = Br+1i*Bc ; 

[elis02]

doesn't work.

I have 3D matrix with known values.

E_x_y_t

the final should also remain 3D.

and if i have 3mm for example dia in the cross section for x, it should remain after the interpolation as well.

think of a small gaussian, that i have 10 pixels in x and y but also i have the development in time in the third dimension.

for each segment of the time, the 10 pixels i want to make them 600 pixels and preserve the value of the real and phse. if i had 3mm beam diameter for the 10 pixels, if i plot it in the 600 pixels i will also get it. But now with higher resolution.

[KSSV]

It should work...attach your data.

[elis02]

attaching the code.

see line 224.

the next lines is why i need this :-)

The relavent variable is E_x_y_t.

[elis02]

So to solve it I'm using interp2 right now:

E_x_y_t([1:length(x)/4, length(x)/4+length(x)/2+1:length(x)],:,:) = [];
E_x_y_t(:,[1:length(x)/4, length(x)/4+length(x)/2+1:length(x)],:) = [];
x = [x(length(x)/4+1:length(x)/4+length(x)/2)];
xq = linspace(x(1),x(end),650);
E_new = zeros(length(xq),length(xq),length(T));
for index = 1:length(T)
E_new(:,:,index) = interp2(x,x.',E_x_y_t(:,:,index),xq,xq.');
end

E_x_y_t = E_new;

clear E_new

Answer 2

[elis02]

To solve this i'm using interp2 right now.

E_x_y_t([1:length(x)/4, length(x)/4+length(x)/2+1:length(x)],:,:) = [];
E_x_y_t(:,[1:length(x)/4, length(x)/4+length(x)/2+1:length(x)],:) = [];
x = [x(length(x)/4+1:length(x)/4+length(x)/2)];
xq = linspace(x(1),x(end),650);
E_new = zeros(length(xq),length(xq),length(T));
for index = 1:length(T)
E_new(:,:,index) = interp2(x,x.',E_x_y_t(:,:,index),xq,xq.');
end

E_x_y_t = E_new;

clear E_new