How to calculate the length of image?

Hi

Regarding the image below:

I want to calculate the length of the large black region, i.e., the perimeter of this region plus the length of the small white objects within it. How may I do so?

I know there are some functions for calculating the perimeter, but how to include the small objects?

Also please note that this is a binary image, but I can also do the calculations on grayscale image, if needed.

Thanks so much.

Steven

Answer 1

[Image Analyst]

Something is wrong with that image. I saved it but when I read it in it looks totally different. Anyway, you need to get the perimeters of everything with regionprops and sum them

measurements = regionprops(binaryImage, 'Perimeter');
allPerimeters = [measurements.Perimeter];
theSum = sum(allPerimeters);

But make sure the big perimeter is the inside one and not the outer edge of the image. If it is, then invert the image and fill holes before calling regionprops

binaryImage = imfill(~binaryImage, 'holes');
measurements = regionprops(binaryImage, 'Perimeter');
theBigPerimeter = measurements.Perimeter;

I can't really test it because, like I said, it seems the image is messed up.

[Steven]

Thanks for your reply and sorry for the problematic image.

Regarding your comment, what I want is the length that I have specified by RED line in the following image (I have done this funny thing in Paint, so that I can show you what I want, sorry for the mess!)

So, how may I do so now? shall I invert it? (your second code or first code?)

By the way! when I invert the image, will I use any data?

Is there anyway for calculating the perimeter without inverting?

Thanks so much!

Steven

[Image Analyst]

How about just simply doing

totalPerimeter = sum(bwperim(binaryImage));

[Steven]

Thanks again.

I get (naturally!) three wide different values from three different codes.

How can I know which one is correct? (because yet I have no idea of the scales)

Thanks.

Steven

[Image Analyst]

There are different algorithms. Summing bwperim just sums up the pixels. regionprops computes the perimeter by calculating the distance between each adjoining pair of pixels around the border of the region. So the first method would have two pixels at 45 degrees to each other as a length of 2 while regionprops would give a distance of sqrt(2). You can use whichever method you like better. Optically the regionprops method is probably more accurate.