multi channel Image reconstruction

MD RESHAD UL HOQUE

2017 10 27

0 个回答

2 次查看（30 天）

0 个投票

How can I reconstruct an image from [3410, 33, 33, 8] sub-images?.

3 个评论
显示 1更早的评论隐藏 1更早的评论

Walter Roberson 2017-10-27

How were the parts arranged relative to each other?

Do you currently have an numeric matrix, or do you have cell arrays?

MD RESHAD UL HOQUE 2017-10-27

编辑：Walter Roberson 2017-10-27

在 MATLAB Online 中打开

Here it is a pansharpening project where two images (single channel image, 8 channel image) are merged to make a new image (8 channel). I crop sub-images from both images ( 1 channel and 8 channel ) and tiled it to feed CNN for training. It is an image restoration problem (regression problem). After training, I tested it and model gave desired 8 channel sub-images [3410, 33, 33, 8]. Now I have to reconstruct an image from sub-images.

This is how I extract sub-images(Code)

load('D:\keras-program\practice\MS_data.mat') %load image
image=M;
g=size(image,3);
%%settings sub-image dimension according to requirement
size_input = 33;
size_label = 21;
scale = 4;
stride = 14;
%%initialization
data = zeros(size_input, size_input, 1, 1);
label = zeros(size_label, size_label, 1, 1);
padding = abs(size_input - size_label)/2;
count = 0;
%%generate data
im_label = modcrop(image, scale);
%%%%%%%%%%%%%%%%%%
function imgs = modcrop(imgs, modulo)
if size(imgs,3)==1
    sz = size(imgs);
    sz = sz - mod(sz, modulo);
    imgs = imgs(1:sz(1), 1:sz(2));
else
    tmpsz = size(imgs);
    sz = tmpsz(1:2);
    sz = sz - mod(sz, modulo);
    imgs = imgs(1:sz(1), 1:sz(2),:);
end
%%%%%%%%%%%%%%%%%%%%%%%%%
[hei,wid] = size(im_label);
im_input = imresize(imresize(im_label,1/scale,'bicubic'),[hei,wid],'bicubic');
for x = 1 : stride : hei-size_input+1
   for y = 1 :stride : wid-size_input+1
       count=count+1;
       for i = 1:1:g     
            subim_input = im_input(x : x+size_input-1, y : y+size_input-1, i);
            subim_label = im_label(x+padding : x+padding+size_label-1, y+padding : y+padding+size_label-1);
              data(:, :, i, count) = subim_input;
              label(:, :, i, count) = subim_label;
          end
      end
  end
order = randperm(count);
data = data(:, :, :, order);
label = label(:, :, :, order); 
save('MS_reduce.mat', 'data', '-v7.3')