Random Field Models in Pattern Recognition and Image Processing

点击翻译按钮获取中文摘要

This research will further develop Markov random field (MRF) models for image processing, pattern analysis, and engineering or scientific data analysis for decision making. The mathematical models are domain-dependent: discrete MRFs for image lattices, spatial point processes for patterns in a feature space, and Markov graphs for proximity data. The researchers will study fundamental problems of model definition, sampling, parameter estimation, and goodness of fit, with specific applications in image segmentation and restoration, texture analysis and synthesis, cluster validation, and multisensor fusion.