Table 1.
Algorithm to obtain samples from p(θ, x|y).
1. Initialize n = 1, {x, θ} = {x̂, θ̂} (optimal point estimates). |
2. FOR t = 1 to number of samples of x to draw: |
2A. FOR t’ = 1 to number of trajectories: |
2A-I. Sample m ~ 𝒩(0, I). |
2A-II. Track (x, m) with Equations B.1–B.3. |
2A-III. Accept move with probability given by Eq. B.4 |
END |
2B. FOR t” = 1 to number of samples of θ per sample of x: |
2B-I. FOR t”’ = 1 to number of samples to skip: |
2B-Ia. Sample the image labels with Equation 9. |
2B-Ib. Sample the image intensity parameters with Equations 16 and 17. |
END |
2B-II. Record sample: x(n) = x, θ(n) = θ. |
2B-III. n=n+1. |
END |
END |