Topics in mathematical morphology for multivariate images

This thesis contributes to the field of mathematical morphology and illustrates how multivariate statistics and machine learning techniques can be exploited to design vector ordering and to include results of morphological operators in the pipeline of multivariate image analysis. In particular, we make use of supervised learning, random projections, tensor representations and conditional transformations to design new kinds of multivariate ordering, and morphological filters for color and multi/hyperspectral images. Our key contributions include the following points:• Exploration and analysis of supervised ordering based on kernel methods.• Proposition of an unsupervised ordering based on statistical depth function computed by random projections. We begin by exploring the properties that an image requires to ensure that the ordering and the associated morphological operators can be interpreted in a similar way than in the case of grey scale images. This will lead us to the notion of background/foreground decomposition. Additionally, invariance properties are analyzed and theoretical convergence is showed.• Analysis of supervised ordering in morphological template matching problems, which corresponds to the extension of hit-or-miss operator to multivariate image by using supervised ordering.• Discussion of various strategies for morphological image decomposition, specifically, the additive morphological decomposition is introduced as an alternative for the analysis of remote sensing multivariate images, in particular for the task of dimensionality reduction and supervised classification of hyperspectral remote sensing images.• Proposition of an unified framework based on morphological operators for contrast enhancement and salt- and-pepper denoising.• Introduces a new framework of multivariate Boolean models using a complete lattice formulation. This theoretical contribution is useful for characterizing and simulation of multivariate textures.