Foundations of modeling for image analysis (Q1962867)

This book (in French) treats of the use of probabilistic models (namely Markovian fields) in image modelization. The aim is to present mathematical foundations as well as examples and to show the link between the probabilistic and variational approaches. The book is divided in three parts, two presenting the theory and one treating many examples are: Contents: The first one presents parametric and non parametric splines models which, though they can be viewed as peculiar cases of Markovian modelization, are treated on their own. The case in which more than one observation is available is also investigated in the context of auto-associative models. The second part is about finite Markov fields and Bayesian estimation. The way to resolution of (stochastic) optimization and parameter estimation problems are investigated. The third part presents various applications of the methods: de-noising and de-blurring, Detection of filar structures, Reconstruction from projections (tomography), Matching of two-dimensional hand drawings. If the book requires a good knowledge (or at least habit) of probability theory, at this level it will provide a good introduction to the effective use of Markovian modelization in the image framework.