I need to process a stack of ultrasonic images, where there is mostly a lot of noise to see, but I need to extract specific objects that look for example like this:

enter image description here

Here is an example from Wikipedia of pretty nice formed object enter image description here

But I'm not sure how to get the whole region.

What I did:

I already tried with mean & median filters to achieve a binary image in the end, which then is processed with the union-find-method to create regions. Unfortunately this leads to vague results. And it seems like the difference of the edge pixels aren't high enough to get a clean sobel-image (for trying to define the margins).

What I believe could work

On a different project I was working with the hough-algorithm to detect a (very clean shaped) ball in a picture. I heard this method can be applied for lines and thus for curves too. But would I be able to detect a region like this ? And if so, how would it be? is it multiple hough-detection for just one area? Since I need to process in real time I'm afraid this would exceed the processing frame (or wouldn't it?)

My question

Has anyone an idea how to easily detect such objects ? Would it be possible to apply hough under real-time conditions, on an ordinary (not an high end) computer too?

Thanks for any advises! Kind Regards

  • $\begingroup$ Welcome to DSP.SE! Can you give an example of the noisy image? $\endgroup$ – Peter K. Oct 23 '15 at 20:11
  • 1
    $\begingroup$ @PeterK. please have a look at the updated post $\endgroup$ – user3085931 Oct 23 '15 at 20:38

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.