Advanced search
Start date
Betweenand

Bayesian inference for signal segmentation in dyed textile images

Grant number: 17/25469-2
Support type:Scholarships in Brazil - Doctorate
Effective date (Start): May 01, 2018
Effective date (End): April 30, 2021
Field of knowledge:Physical Sciences and Mathematics - Probability and Statistics
Principal Investigator:Nancy Lopes Garcia
Grantee:Victor Freguglia Souza
Home Institution: Instituto de Matemática, Estatística e Computação Científica (IMECC). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil
Associated research grant:17/10555-0 - Stochastic modeling of interacting systems, AP.TEM

Abstract

Markov random fields are often used when analyzing image data. In some applications, these analyses become difficult due to the presence of textures in images, which hinders the use of spatial models with simple interaction neighborhoods because they induce very sophisticated correlation structures. The main goals of the project are (i) to study and develop probabilistic methods for identifying the most sparse possible interaction structures for texture images; (ii) to study and develop estimation methods for these models in a Bayesian approach; (iii) apply the developed theory to real textile dyeing data that motivate the project. (AU)