I just finished coding a simple implementation of the CONDENSATION algorithm.
The multi hypothesis framework of the PF is much more robust than the Kalman Filter.
I use Mean shift to track the particles around the HSV feature space.
Hopefully I'll find a clever heuristic way to automatically tweak the parameters.
The multi hypothesis framework of the PF is much more robust than the Kalman Filter.
I use Mean shift to track the particles around the HSV feature space.
Hopefully I'll find a clever heuristic way to automatically tweak the parameters.
