How to detect and count number of circles in overlap condition

Hello...I have questions about an image with multiple shapes that overlap.

1. How to count the number of disks in the image? (If I'm not mistaken,the mentioned disks are refer to the circle shapes)

2. How to count the boundary pixels in the image?

3. How to count the pixels of that are common to two or more objects?

[KSSV]

Read about imfindcircles.

[Syaw Werdas]

Thanks for your advice sir, I'll read about the function that you mentioned

Answer 1

[Image Analyst]

I'd try normalized cross correlation and scan the image with known templates. Attached is a demo.

[Syaw Werdas]

Thanks for your answer sir/miss. Really appreciate it. I'll give it a try soon

[Syaw Werdas]

Sir/miss, I already applied the example that you have given and it works! However, I still couldn't detect the other circles that are overlaps. How should I do this? I think the original shapes would be gone if I use imerode to separate them. Attached are the code and the result, tqvm.

[Image Analyst]

Maybe you want to try Hu's image moments: https://www.youtube.com/watch?v=Nc06tlZAv_Q

Or try feature detection in the Computer Vision Toolbox: https://www.mathworks.com/products/computer-vision.html#feature-detection

[Syaw Werdas]

Thanks for your advice, however I still don't find any advice that can give me idea about how I can detect and count the OVERLAP OR INTERCEPTED SHAPES. Most of the advices refer on focusing at one object in an image which are already separated significantly. But, I still want to thank you for share with me the new knowledges.

[Image Analyst]

Or try deep learning, perhaps YOLO v2. That's all I got. I've run out of suggestions.

Answer 2

[KALYAN ACHARJYA]

Here are my submission:

1. How to count the number of disks in the image? (If I'm not mistaken,the mentioned disks are refer to the circle shapes)

Some how this function help you to get the counts.

https://www.mathworks.com/help/images/ref/regionprops.html

2. How to count the boundary pixels in the image?

Extract the boundary pixels, and do the operation as you are looking for.

boundary_pixels=image([1,end],:) , (image([:,[1,end])');

Simple count is=2*rows*columns

3. How to count the pixels of that are common to two or more objects?

Please refer the following links

https://www.mathworks.com/matlabcentral/answers/82600-how-can-i-find-the-overlap-between-two-binary-volumes

https://www.mathworks.com/matlabcentral/answers/54580-how-can-i-find-the-location-of-the-pixels-between-adjacent-regions

Stay Safe!

[Syaw Werdas]

Thanks for your quick answer sir, I will give it a try