I am trying to make tracking for soccer player, I need to detect features from this player and then estimate the distance difference of these pixels over number of frames.

first I have to detect the main features of the player and then match them to those in the next frames, then calculate the distance difference.

My problem, there is no good feature detector at MATLAB to my case, these sample from the existing detectors:


enter image description here


enter image description here


enter image description here

Q: I see that I can improve feature detector by doing preprocessing such like: adaptive thresholding, opening & closing and sharping to highlight the player edges. I have do the first one to make adaptive threshloding, and then make the feature detector for the output logical image, such like the following

After adaptive thresholding Harris, for logical image

enter image description here

I tried to make the feature detector on gray image after thresholding but I couldn't change logical image to gray image.

Also any suggestion or alternative approach to do?



Your Answer

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