how to concatenate structure fields

Hello I have a structure s(i).f(j).d where i=1:40 and j=1:10.I want to concantenate all the fields in to one single matrix. for eg

    for j=1:10
m(:,j)=s(1).f(j).d

with the following code above

this concatenates 10 column matrix into m .But this is just for s(1) how to proceed for s1..s2...s40 for all the 40 fields?

Answer 1

[Azzi Abdelmalek]

Look at this example

s(1).f(1).d=11
s(1).f(2).d=12
s(2).f(1).d=21
s(2).f(2).d=22
a=s.f
n=numel(a)
for k=1:numel(s)
  M(k,:)=[s(k).f(1:n).d]
end

[Newman]

Subscripted assignment dimension mismatch. this is the error it is showing

[Azzi Abdelmalek]

attach your data

[Newman]

this is my structure file

[Azzi Abdelmalek]

load lbpstructure
a=s.f;
n=numel(a);
M=[]
for k=1:numel(s)
  b=[s(k).f(1:n).d];
  M=[M;b];
end
M

Answer 2

[Guillaume]

Use a double for loop, then:

m = zeros(numel(s(1).f(1).d), numel(s(1).f(1)), numel(s)); %preallocation is advised
for sidx = 1:numel(s) 
  for fidx = 1:numel(s(sidx).f)
     m(:, fidx, sidx) = s(sidx).f(fidx).d;
  end
end

Note that I've concatenated the s1, ... s40 along the 3rd dimension. You didn't specify what you wanted exactly.

[Newman]

This is exactly what I want: please see the image below

your code is returning m as 10304x10x40 double where as I want it as 10304x400.(400= 10 fields in s(1) or s(2) etc. so for 40 s(i) = 40x10=400)

[Guillaume]

Whichever shape you want for the output, the idea is still the same. Iterate over both arrays. To get the above:

m = zeros(numel(s(1).f(1).d, numel(s) * numel(s(1).f));
col = 1;
for sidx = 1 : numel(s)
   for fidx = 1 : numel(f)
      m(:, col) = s(sidx).f(idx).d;
      col = col + 1;
   end
end