Online data smoothing for hand tracking using Kalman filters

I am using Kinect with OpenNI / NITE. OpenNI can track human hands using NITE. In addition, OpenNI can smooth out the monitored manual line, and I tried to figure out how it does it.

I tried to use Kalman filters, replaced the old hand position with a Kalman estimate of the hand position, but still smoother in OpenNI is much better.

I would appreciate any tips on how to smooth online data or how to set parameters in the Kalman filter (something specific for hand tracking, as I already know what the parameters do).

+5
source share
2 answers

, . , . , , 3D- , (x, y z ), :

State vector =[x, y, z, v_x, v_y, v_z]

Update equations: (x,y,z) = (x,y,z)+ (v_x,v_y,v_z)*delta(t)
velocity would be constant

, ( ).

. , . . , , , W ( Q), H A. Q R , .

, , .

+4

, , kalman.

. (, , ..).

, , , .

opencv.

, !

0

All Articles