hal-00151999, version 2
CONSTRUCTION OF BAYESIAN DEFORMABLE MODELS VIA STOCHASTIC APPROXIMATION ALGORITHM: A CONVERGENCE STUDY
(2008-12-18)
Abstract: The problem of the definition and the estimation of generative models based on deformable templates from raw data is of particular importance for modelling non aligned data affected by various types of geometrical variability. This is especially true in shape modelling in the computer vision community or in probabilistic atlas building for Computational Anatomy (CA). A first coherent statistical framework modelling the geometrical variability as hidden variables has been given by Allassonnière, Amit and Trouvé (JRSS 2006). Setting the problem in a Bayesian context they proved the consistency of the MAP estimator and provided a simple iterative deterministic algorithm with an EM flavour leading to some reasonable approximations of the MAP estimator under low noise conditions. In this paper we present a stochastic algorithm for approximating the MAP estimator in the spirit of the SAEM algorithm. We prove its convergence to a critical point of the observed likelihood with an illustration on images of handwritten digits.
- 1:
- CNRS : UMR7641 – Université de Versailles Saint-Quentin-en-Yvelines – Polytechnique - X
- 2:
- CNRS : UMR7539 – Université Paris XIII - Paris Nord – Université Paris VIII - Vincennes Saint-Denis
- 3:
- CNRS : UMR8536 – École normale supérieure de Cachan - ENS Cachan
- Domain : Mathematics/Statistics
Statistics/Statistics Theory - Keywords : stochastic approximation algorithms – non rigid-deformable templates – shapes statistics – Bayesian modelling – MAP estimation
- Available versions : v1 (2007-06-06) v2 (2009-01-16)
- hal-00151999, version 2
- http://hal.archives-ouvertes.fr/hal-00151999
- oai:hal.archives-ouvertes.fr:hal-00151999
- From:
- Submitted on: Friday, 16 January 2009 16:24:16
- Updated on: Friday, 16 January 2009 16:47:43
Associated documents
Export