MATLAB detecting the largest open space in a binary image

I am trying to obtain the area of the largest open (black) space (non-contiguous) in a binary image. The actual binary image is shown in "Original_Image.png" attachment. What I am looking for is the area of the region roughly bounded by the red line in the "Original_Image_red.png". Thanks for your help.

Answer 1

[Image Analyst]

You didn't attach the image. But the steps are (untested)

% Call bwlabel or bwconncomp on the inverted image.
cc = bwconncomp(~binaryImage);
% Measure the area of all the blobs.
measurements = regionprops(cc, 'Area');
allAreas = [measurements.Area]
% Find the biggest area and blob number
[maxArea, indexOfBiggestBlob] = max(allAreas);

[Kanthi]

Thanks a lot for answering this question! I apologize that the attachments didn't show up. This is my first question on this forum. The problem is the black region is continuous around the picture. So, I need to figure out the largest black region surrounded by white blobs. The white blobs are not touching, but if I draw a boundary connecting the adjacent white blobs, it gives me the area of the black regions and I need to identify the maximum black region. I hope the images have attached this time.

Thanks for your time.

[Image Analyst]

I would just use imfreehand to hand draw a region since it's not really clear where the boundaries are. That's probably the fastest way unless you have more than 500 or so images.

[Image Analyst]

Or try using the Euclidean distance transform, bwdist(), and threshold it.

[Kanthi]

imfreehand is not an option for now as some of the images have areas close together visually and its not practical for me to mark all the areas on all the images.

I do think the second option seems good (Euclidean distance transform). Actually I did try the option to obtain Euclidean distances (See the attached image). However, I couldn't figure out how thresholding option should be used. Could you kindly assist me with that? Again, your assistance is greatly appreciated. Thanks.

[Image Analyst]

Find the value of the EDT at the lower "struts" and use that value.

binaryImage = edtImage > thresholdValue;
cc = bwconncomp(binaryImage);
measurements = regionprops(cc, 'Area');
allAreas = [measurements.Area]

[Kanthi]

BINGO...it worked!

Thanks a lot for this wonderful suggestion and all your prompt responses! Its greatly appreciated!

[Image Analyst]

Can you go ahead and mark my answer as "Accepted" then? Thanks.

[Kanthi]

I just did. Please let me know if I need to do anything else. Again...thanks for all your help!