Macrocanonical Models for Texture Synthesis - Archive ouverte HAL Access content directly
Conference Papers Year : 2019

Macrocanonical Models for Texture Synthesis

Abstract

In this article we consider macrocanonical models for texture synthesis. In these models samples are generated given an input texture image and a set of features which should be matched in expectation. It is known that if the images are quantized, macrocanonical models are given by Gibbs measures, using the maximum entropy principle. We study conditions under which this result extends to real-valued images. If these conditions hold, finding a macrocanonical model amounts to minimizing a convex function and sampling from an associated Gibbs measure. We analyze an algorithm which alternates between sampling and minimizing. We present experiments with neural network features and study the drawbacks and advantages of using this sampling scheme.
Fichier principal
Vignette du fichier
main.pdf (26.4 Mo) Télécharger le fichier
Origin : Files produced by the author(s)
Loading...

Dates and versions

hal-02093364 , version 1 (08-04-2019)

Identifiers

Cite

Valentin de Bortoli, Agnès Desolneux, Bruno Galerne, Arthur Leclaire. Macrocanonical Models for Texture Synthesis. Scale Space and Variational Methods in Computer Vision. SSVM 2019, Jun 2019, Hofgeismar, Germany. ⟨10.1007/978-3-030-22368-7_2⟩. ⟨hal-02093364⟩
153 View
83 Download

Altmetric

Share

Gmail Facebook X LinkedIn More