2
$\begingroup$

I am trying to check whether this black circle is exist or not on my images.

enter image description here

My current approach is like that;

  • Hough Circle detection

  • Crop the outer roi

  • Adaptive threshold

enter image description here

  • count zero pixels

  • if ( zero pixels >500 ) black hole exist else black hole does not.

I just want to know is this method useful if this algorithm go for real time check ? if not what kind of methods you could suggest me ?

Here is a image with blackhole

enter image description here

Here is a image without a blackhole.

enter image description here

$\endgroup$
2
  • $\begingroup$ Could you post images after each step of your current algorithm ? Will the black hole always be centered in the image ? If so, you can crop the image to begin with, which will mean less pixels to analyze, so more time to perform analysis. $\endgroup$
    – Loufylouf
    Mar 19, 2015 at 12:24
  • 1
    $\begingroup$ If the both the image and coin sizes keep the same, can you just check the pixel values within a small circle region around the center? Hough detection might be an overkill in your case. $\endgroup$
    – lennon310
    Mar 19, 2015 at 14:03

1 Answer 1

1
$\begingroup$

I think you can use a much simpler method.

The steps I did are:

  1. I filtered the image with an Edge Preserving Filter.
  2. I looked the profile of values along the 2 diagonals of the image.

I think once you look at it, it will be easy to build a model.
Let's see the image without the black circle:

enter image description here

enter image description here

enter image description here

From ~150 to ~250 we see mostly a flat zone.

Now, let's see the image with the black circle:

enter image description here

enter image description here

enter image description here

Now, on the same segment we see 2 levels.

This is a great feature to classify with.

The code is available at my StackExchange Codes Signal Processing GitHub Repository (Look at the SignalProcessing\Q22183 folder).

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.