Attempting to Crop binary image

I want the white section of this image I took,

I used this code to attempt a crop,

[rows, columns] = find(ProperlyOrientedBinaryBlock);
topRow = min(rows);
bottomRow = max(rows);
leftColumn = min(columns);
rightColumn = max(columns);
croppedImage = ProperlyOrientedBinaryBlock(topRow:bottomRow, leftColumn:rightColumn);
imshow(croppedImage)

I'm not getting the result I expect. Anyone have any idea?

[Satyajeet Sasmal]

Could you provide us with your workflow? What does the "ProperlyOrientedBinaryBlock" function do?

Answer 1

[Image Analyst]

Try this:

grayImage = imread('binary.jpg');
binaryImage = grayImage > 128;
subplot(2, 1, 1);
imshow(binaryImage);
% Get convex hulls
binaryImage2 = bwconvhull(binaryImage, 'objects');
% Extract largest blob only
binaryImage2 = bwareafilt(binaryImage2, 1);
% Now you can crop:
[rows, columns] = find(binaryImage2);
topRow = min(rows);
bottomRow = max(rows);
leftColumn = min(columns);
rightColumn = max(columns);
croppedImage = binaryImage(topRow:bottomRow, leftColumn:rightColumn);
subplot(2, 1, 2);
imshow(croppedImage)

It's not super robust - if the blobs are broken up or nested, it's possible that it would need some more code to get it perfect, but it will work for rectangular Lego blocks as long as they're not too broken apart.

[manimuthu Venu]

Is it possible to get the original coordinates of the cropped image, so that it can be merged in the same position in another image? thanks a lot in advance

Answer 2

[Thorsten]

The image contains some spurious white points, e.g. at (950,3). You can delete them using erosion

se = strel('disk',1)
erodedB = imerode(B,se);