FLANN clustering/search for dense optical flow image comparison

asked 2016-04-03 22:11:22 -0500

ttyy gravatar image

Hi -

I'm working on building an experimental content based image retrieval (CBIR) system using optical flow.

I'm pulling the images from video sequences where I'm calculating Farneback optical flow, so each frame has an associated dense 2D matrix of flow vectors. This matrix is then flattened into a long feature vector.

I'm also a little confused on how to use FLANN clustering with OpenCV. Is there a good C++ tutorial out there that I haven't found? (preferably for Open Cv 3)

I'm specifically quite confused on how to choose the distance measurement. I have some code that computes energy between flow fields and I'd like to use that for a distance measurement.

I'm just starting out with OpenCV, so not entirely sure if I'm on the right path here.

Thanks very much

edit retag flag offensive close merge delete