I have a feature detection algorithm that is called FAST - Feature Accelerated Segmented Test.

It's a very fast algorithm for finding feature points, e.g "corners" inside an image. Here is one example.

enter image description here

The problem with these points is that they are moving around very rapidly, like gaussian noise. Still the shape can be viewn.


Is there a way to filter so these points will "move around" less? Like a particle filter, but I don't know if a particle filter can be useful here.

The "dots" comes from a $M \times N$ matrix where $N = 2$ and $M$ can vary a lot. That matrix contains the coordinates for each blue points.

This is how the "dots" looks like. enter image description here https://pastebin.com/wF34Fymr


The blue "dots" are coordinates, and they move around like "vibrating", even if the object inside the picture (hand) is static.

The dots move like this: https://www.youtube.com/watch?v=NhuE9P5gDJg

  • $\begingroup$ Could you try to explain the problem clearer? For instance, do they move because the hand moves? $\endgroup$
    – Royi
    Commented Nov 24, 2023 at 22:03
  • $\begingroup$ @Royi Thank you for your reply. I have updated my question with more information + a video. $\endgroup$
    – euraad
    Commented Nov 25, 2023 at 10:41
  • $\begingroup$ I am still not sure what do you mean the movement. Do you mean the movement of your hand holding the cards? $\endgroup$
    – Royi
    Commented Nov 25, 2023 at 10:46
  • $\begingroup$ @Royi No, the blue dots are like moving +- steps upwards, backwards, sides due to noise. $\endgroup$
    – euraad
    Commented Nov 25, 2023 at 10:47
  • $\begingroup$ @Royi The blue dots are not floating around uncontrolled. $\endgroup$
    – euraad
    Commented Nov 25, 2023 at 10:48


Your Answer

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