×

Mechanical models as priors in Bayesian tomographic reconstruction. (English) Zbl 0901.62046

Hanson, Kenneth M. (ed.) et al., Maximum entropy and Bayesian methods. Proceedings of the 15th international workshop, Santa Fe, NM, USA, July 31–August 4, 1995. Dordrecht: Kluwer Academic Publishers. Fundam. Theor. Phys. 79, 117-124 (1996).
Summary: We introduce a new prior – the weak plate – to Bayesian tomographic reconstruction. The weak plate captures the piecewise ramplike spatial structure evident in primate autoradiograph source distributions. The weak plate is a part of a family of “mechanical” models – weak membrane (1st order), weak plate (2nd order), and weak quadric (3rd order) – in which a class of smoothness constraints derived from properties of ideal physical materials are used as models in the associated reconstruction problem. Since “weak” priors generate local minima in MAP estimation, we have designed novel Generalized Expectation-Maximization deterministic annealing algorithms to alleviate this problem. Our simulation studies qualitatively demonstrate the improvements over the weak membrane and maximum likelihood reconstructions.
For the entire collection see [Zbl 0855.00029].

MSC:

62F15 Bayesian inference
62P99 Applications of statistics