Converting Match to Probability and Spatial Evidence
Define the constant H* =
Hi
The probability of M if the correct displacement D were k:
P( M | D=k ) = Gk H* / Hk
Bayes' theorem and H* cancellations convert this to
P( D=k | M ) =
Gk / HkP(D=k)
Gi / HiP(D=i)
where P(D=i) is the prior probability that the correct D is i.
A given prior and noise assumption derives a probability curve, which is convolved with
a stereoscopic range uncertainty model to make an evidence ray backbone, which is swept
to make a 2D cross-section, itself swept again to add a 3D cone of evidence to a gid map: