Image Processing and Analysis with Graphs

4

A Parametric Maximum Flow Approach for Discrete Total Variation Regularization

Chambolle Antonin

CMAP, Ecole Polytechnique, CNRSF-91128 Palaiseau, France

Email: [e-mail]

Jérôme Darbon

CMLA, ENS Cachan, CNRS, PRES UniverSudF-94235 Cachan, France

Email: [e-mail]

CONTENTS

4.1    Introduction

4.2    Idea of the approach

4.3    Numerical Computations

4.3.1    Binary Optimization

4.3.2    Discrete Total Variations optimization

4.4    Applications

Bibliography

4.1    Introduction

In this chapter, we consider the general reconstruction problem with total variation (TV) that takes the following form

$E\left(u\right)=F\left(u\right)+\text{λ}J\left(u\right)$

with m ≤ u ≤ M, and where F and J, respectively, correspond to the data fidelity ...