reshape and save data in a for loop

I have a dataset of 100 rows and 3072 columns,and every single row represents an image.I want to divide every single image into 4 sub-images and get the mean of every color feature(rgb).The result should be 4,4,4(rgb) values,ie, for one image it should have 12 values.My code is,

function fr = features1(x)
  [row col]=size(x);
  f=[];
  for i=1:row
    r=x(i,(1:1024));
    g=x(i,(1025:2048));
    b=x(i,(2049:3072));
    rre=reshape(r,32,32);
    gre=reshape(g,32,32);
    bre=reshape(b,32,32);
    r1=mat2cell(rre,[32/2,32/2],[32/2,32/2]);
    g1=mat2cell(gre,[32/2,32/2],[32/2,32/2]);
    b1=mat2cell(bre,[32/2,32/2],[32/2,32/2]);
    fr1=mean(r1{1}(:));
    fr2=mean(r1{2}(:));
    fr3=mean(r1{3}(:));
    fr4=mean(r1{4}(:));
    fg1=mean(g1{1}(:));
    fg2=mean(g1{2}(:));
    fg3=mean(g1{3}(:));
    fg4=mean(g1{4}(:));
    fb1=mean(b1{1}(:));
    fb2=mean(b1{2}(:));
    fb3=mean(b1{3}(:));
    fb4=mean(b1{4}(:));
    f(:,:,i)=[f,fr1,fg1,fb1;fr2,fg2,fb2;fr3,fg3,fb3;fr4,fg4,fb4]
  end
end

It stores the values for only one image and rest is shown in command window but not stored in workspace. Moreover,I get these error messages:

Error using horzcat
Dimensions of matrices being concatenated are not consistent.

Error in features1 (line 26)
f(:,:,i)=[f,fr1,fg1,fb1;fr2,fg2,fb2;fr3,fg3,fb3;fr4,fg4,fb4]

Kindly help me in reshape and store all the values(1200) and also the values should be stored in the form red subimage1,green subimage1,blue subimage1 and in next line for subimage 2 and so on...

[dpb]

You've got LH variable in the RH expression...

f(:,:,i)=[f,fr1,fg1,fb1;fr2,fg2,fb2;fr3,fg3,fb3;fr4,fg4,fb4];

You can try

f(:,:,i)=[fr1,fg1,fb1;fr2,fg2,fb2;fr3,fg3,fb3;fr4,fg4,fb4];

and it possibly might work...

[Kalai Arasi Mehavarnam]

Thank you so much,it works...