I'm trying to track a retroreflective sphere using the leap motion + OpenCV blob tracking and I'm able to do it successfully. following steps outlined in
Unfortunately, reflections from the environment seem to interfere with and throws the tracking off. Is there a way to solve this?
I'm looking to see if I can use epipolar constraints to narrow down my search region to pair blobs between left and right image more accurately. Is there a simple way to get the epipolar line/ window of probability on the right image, given a tracked point in the left image? That would be very helpful in reducing the interference.
Any help would be greatly appreciated.