Stopping criteria for anisotropic PDEs in image processing

A. Ilyevsky, E. Turkel*

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

15 Scopus citations

Abstract

A number of nonlinear diffusion-like equations have been proposed for filtering noise, removing blurring and other applications. These equations are usually developed as time independent equations. An artificial time is introduced to change these equations to parabolic type equations which are then marched to a steady state. In practice the time iteration is stopped before the steady state is reached. The time when to stop the iteration is usually determined manually for each case. In this study we develop a more automatic procedure for stopping the time integration.

Original languageEnglish
Pages (from-to)333-347
Number of pages15
JournalJournal of Scientific Computing
Volume45
Issue number1-3
DOIs
StatePublished - Oct 2010

Keywords

  • Anisotropic PDE
  • Image processing
  • Stopping criteria

Fingerprint

Dive into the research topics of 'Stopping criteria for anisotropic PDEs in image processing'. Together they form a unique fingerprint.

Cite this