How do i keep the lukas kanade method keep running on my head even when i move it around, or twist the head. The the point i am only able to make a mask based on Haar cascade and use that to detect good features with for the lukas kanade method, but this method kinda mess up when i twist my head and so on.
any suggestions?
Take a look at: http://wiki.ros.org/pi_face_tracker
I think that your idea is more or less based on this tracking algorithm.
Once a face is detected (Viola & Jones), the "GoodFeaturesToTrack" filter is run over the face region. These points are then fed to the Lucas-Kanade Optical Flow Tracker, which follows the points from one frame to the next. Two additional methods are run to maintain the integrity of the tracked feature cluster.
feature points are pruned (dropped) if they lie too far outside the current cluster in the x-y camera plane or in depth
new feature points are added by expanding the current cluster ROI by a small amount (10%) and running "GoodFeaturesToTrack" over the expanded region. If the feature cluster shrinks below a specified minimum number of points, the algorithm returns to the Haar face detector and scans the video to relocate the face.
Thank you for the response..
I am not quite sure if understand the 2 additional methods correctly.
I am at moment only able to apply "GoodFeatureToTrack" in a rect from where i have detected the frame.. So when move my face, the same mask will be used, even though the face is not there.. which makes sense since the rect hasn't been updated.
I am not sure I understand how this even would happen.. my frames is now and pref framed is delayed with 1 ms, so it seem impossible for it to happen.
How long would you keep expanding it.. "GoodFeaturesToTrack" doesn't loose any of the point, it keeps having the max number of corners, even if i move me head away from the Rect .
I don't know the details, but you can see the source code of the face tracker.
Asked: 2015-04-14 19:46:27 -0600
Seen: 518 times
Last updated: Apr 15 '15
