Guessing at what you mean about "compare with the original image and extract the components from it":
Assuming the control information is in YourMatrix and that the original image is YourImage:
ncol = size(YourMatrix, 1);
outputs = cell(ncol,1);
for K = 1 : ncol
x = YourMatrix(K,1);
y = YourMatrix(K,2);
high = YourMatrix(K,3);
wide = YourMatrix(K,4);
this_extract = YourImage(y:y+high-1, x:x+wide-1, :);
outputs{K} = this_extract;
end
Now outputs is a cell array containing the sub-sections of the image.
Somehow, I suspect that what you want to do is feature extraction based on the extracted portions and compare the results against a database of features of OCR'd characters to figure out which character you are probably looking at. Not that you said anything like that, but...
By the way, please re-check that the height is in the third position and the width is in the fourth position. If this matrix is being output by one of the extract_features routines then it almost certainly has the width in the third position and the height in the fourth position.
Best Answer