This paper deals with the problem of statistical unsupervised fusion of dependent sensors with its potential applications to multisensor image segmentation. On the one hand, Bayesian fusions can be of great efficiency, particularly when using hidden Markov models. On the other hand, we give some examples showing that there are situations in which the Dempster-Shafer fusion can be usefully integrated in the classical Bayesian models. The contribution of this paper is then to show how a recent parameter estimation of probabilistic models, valid in the case dependent and possible non Gaussian sensors case, can be extended to the situations in which some of sensors can be evidential. The proposed method allows one to imagine different unsupervised segmentation methods, valid in the Dempster-Shafer framework for dependent and possibly non Gaussian sensors.

. In this paper we discuss the efficient implementation of the STRIP (STrong Rule Induction in Parallel) algorithm in parallel using a transputer network. Strong rules are rules that are almost always correct. We show that STRIP is well suited for parallel implementation with scope for parallelism existing at four different levels of the algorithm. We present a performance study analysing the best topologies for the transputer network using different number of transputers. The choice of certain variables (the number and size of samples) in the STRIP algorithm affects the performance (speedup and efficiency) of the implementation. 1. Introduction Since 1970 when Codd introduced the relational model for databases [7], the database industry has matured a great deal and applications that were never envisaged earlier have become possible. Furthermore, heterogeneous data collections, perhaps distributed and multi-media, can now be integrated and used globally [6]. Despite these and other adv...

this paper.

The Dempster-Shafer combination rule can be of great utility in multisensor image segmentation. In addition, the approach based on theory of evidence can be seen as generalizations of the classical Bayesian approach, which is often used in the Hidden Markov Field Model context. Finally, some recent works allow one to use the DempsterShafer combination rule in the Markovian context, and different methods so obtained can greatly improve the effectiveness of Markovian methods working alone. The aim of this paper is to make these methods unsupervised by proposing some parameter estimation algorithms. In order to do so, we use some recent methods of generalized mixture estimation, which allows one to estimate mixtures in which the exact nature of components is not known.

This paper deals with the statistical segmentation of multisensor images. In a Bayesian context the interest of using Hidden Markov Random Fields, which allow one to take contextual information into account, has been well known for about twenty years. In other situations, the Bayesian context is inadequate and one has to make use of the theory of evidence. The aim of our work is to propose an evidential model which can take into account contextual information via Markovian fields. We define a general evidential Markovian model and show that it is usable in practice. Some simulation results attest to the practical interest of the proposed model.