How filtering the detected face position and size along time?

Question

I created an app that uses Core Image (CIDetector, CIDetectorTypeFace) for face detection (tracking) in realtime. Something similar to: http://www.youtube.com/watch?v=D_XiwtoorLk I wonder how can I filter the face position and size in order to have something smooth. I guess I have to use a low pass or high pass filter, but not sure how can I do it. Does anyone have an idea or recommendation? I saw an app doing something like that, but I don't remember it. Thanks for help.

EDIT: When I say filtering, this is what I mean: Left: What I receive from core image when a face is detected. Right: The filtered signal from the left.

Answer 1

You'll want to look more into object tracking in general. This post suggests using Kalman filters, which may be your best bet.