bayesImageS

Various algorithms for segmentation of 2D and 3D images, such as computed tomography and satellite remote sensing. This package implements Bayesian image analysis using the hidden Potts model with external field prior of Moores et al. (2015) doi:10.1016/j.csda.2014.12.001. Latent labels are sampled using chequerboard updating or Swendsen-Wang. Algorithms for the smoothing parameter include pseudolikelihood, path sampling, the exchange algorithm, approximate Bayesian computation (ABC-MCMC and ABC-SMC), and the parametric functional approximate Bayesian (PFAB) algorithm. Refer to doi:10.1007/978-3-030-42553-1_6 for an overview and also to doi:10.1007/s11222-014-9525-6 and doi:10.1214/18-BA1130 for further details of specific algorithms.