Hi,
In the Traffic Monitoring Object Detection and Tracking demo, the process covariance matrix Q is computed in the function RADARDEMO_clusterTracker_updateFQ in the following：
float c = (float)(dt*dt*4.0); // (dt*2)^2 *d,
float b = (float)(c*dt*2); // (dt*2)^3 *d
float a = (float)(c*c); // (dt*2)^4 *d
float Q[16] = {
a, 0, b, 0,
0, a, 0, b,
b, 0, c, 0,
0, b, 0, c};
that is Q[16] = {
16*dt^4, 0, 8*dt^3, 0,
0, 16*dt^4, 0, 8*dt^3,
8*dt^3, 0, 4*dt^2, 0,
0, 8*dt^3, 0, 4*dt^2};
I think the mathematical modeling is as below:
x(k+1)=F*x(k)+Dv(k)
x(k)=[posx,posy,velx,vely]
F=[1,0,dt,0
0,1,0,dt,
0,01,0
0,0,0,1 ]
D=[dt^2/2,dt^2/2,dt,dt]
Q=D*D'q;
=q*[dt^4/4, 0, dt^3/2, 0
0, dt^4/4,0, dt^3/2
dt^3/2, 0, dt^2, 0
0, dt^3/2,0, dt^2
]
I don't know my understanding is right or not.
Thanks in advanced,
Regards,
Rata
