Dense Optical flow from sparse motion vectors
Hi community,
I created a sparse optical flow (vector field) from KLT but I want to extend this to a dense optical flow (like Gunnar Farnback is doing). I found the function cv::ximgproc::EdgeAwareInterpolator() but this is very slow and without any acceleration. Could someone give me a hint?
can you show, what you tried here ?
maybe one of those is for you:
Try the DisparityWLSFilter technique, lines 326-332. You will need a mask/confidence image. This can be 1 for the sparse points and 0 for everything else.