How to crop and save plot square on an image using 4 points coordinates

I have a dataset with annotations for each image. when i read the annotation to the image it will look like this.

I want to crop all the boxes for each box will be a single image and save it to my dir. How i can do that if imcrop doesn't support 4 coordinate?

if true
  function displaydata
% This function displays training data with annotations overlaid. Other
% datasets could be seen in the same way by changing the paths of image
% and annotation directories
uf = dir('training_data/images/*.jpg');
for i = 1000:length(uf)
  dot = strfind(uf(i).name,'.');
  imname = uf(i).name(1:dot-1);
  load(['training_data/annotations/' imname '.mat']);
  im = imread(['training_data/images/' uf(i).name]);
  imshow(im);
  for j = 1:length(boxes)
      box = boxes{j};
      line([box.a(2) box.b(2)]',[box.a(1) box.b(1)]','LineWidth',3,'Color','y');
      line([box.b(2) box.c(2) box.d(2) box.a(2)]',[box.b(1) box.c(1) box.d(1) box.a(1)]','LineWidth',3,'Color','r');
      disp([box.a(1), box.a(2), box.b(1), box.b(2), box.c(1), box.c(2), box.d(1), box.d(2)]);
  end
  disp(['Press any key to move onto the next image', imname]);pause;
 end
end

the code above is how to generate the annotation file and show the bound box to the image. the box.a(1) is like x axis and the box.a(2) is y axis and so on to the others boxes. Really need some help for this to finish my assignment at college.

Answer 1

[KSSV]

I = imread('peppers.png') ;
[m,n] = size(I(:,:,1)) ;
%%Four corners 
data = [
  277.0000  154.0000
  350.0000  208.0000
  286.0000  268.0000
  182.0000  174.0000
];
x = data(:,1) ; y = data(:,2) ;
bI = poly2mask(x,y,n,m);
meas = regionprops(bI, 'BoundingBox');
cI = imcrop(I, meas.BoundingBox);