OpenCV Kalman Filter stack overflow

OpenCV Kalman Filter stack overflow

I'm trying to implement a kalman filter for 3D tracking in OpenCV 2.2. The
state variables are the coordinates x,y,z followed by the velocities Vx,Vy
and Vz and I can only measure x,y and z.
I used an example from the book Learning OpenCV from O'reilly to get
started, but when I tried to adapt the example to my problem things got a
little confusing.
This is my implementation (I've tried to reduce the code to just the
relevant parts, and I've commented a lot to hopefully ease the reading).
CvKalman* kalman = cvCreateKalman( 6, 3, 0 );
// Setting the initial state estimates to [0,0,0,0,0,0].
CvMat* x_k = cvCreateMat( 6, 1, CV_32FC1 );
cvZero(x_k);
// Setting the a posteriori estimate to zero.
cvZero(kalman->state_post);
// Creating the process noise vector.
CvMat* w_k = cvCreateMat( 2, 1, CV_32FC1 );
// Creating the measurement vector.
CvMat* z_k = cvCreateMat( 6, 1, CV_32FC1 );
cvZero( z_k );
// Initializing the state transition matrix.
float F_kalman[] = { 1,0,0,0.05,0,0, 0,1,0,0,0.05,0, 0,0,1,0,0,0.05,
0,0,0,1,0,0, 0,0,0,0,0,1 };
memcpy( kalman->transition_matrix->data.fl, F_kalman, sizeof(F_kalman));
// Initializing the other necessary parameters for the filter.
cvSetIdentity( kalman->measurement_matrix);
cvSetIdentity( kalman->process_noise_cov, cvRealScalar(1e-2) );
cvSetIdentity( kalman->measurement_noise_cov, cvRealScalar(1e-1) );
cvSetIdentity( kalman->error_cov_post, cvRealScalar(1));
// Updates the measurement vector with my sensor values, wich were in
the variable xyz (an array of CvScalar).
cvSetReal1D(z_k,0,xyz[i].val[0]);
cvSetReal1D(z_k,1,xyz[i].val[1]);
cvSetReal1D(z_k,2,xyz[i].val[2]);
cvKalmanPredict(kalman,0);
cvKalmanCorrect(kalman,z_k);
The problem is, when I run the code I get a "Unhandled exception at
0x55a3e757 in test.exe: 0xC00000FD: Stack overflow." at the
cvKalmanCorrect line.
Perhaps I've initialized one of the matrices to the wrong expected size,
but I'm really lost on how to check this.
Any thoughts?