Suscripción a Biblioteca: Guest
Portal Digitalde Biblioteca Digital eLibros Revistas Referencias y Libros de Ponencias Colecciones
Telecommunications and Radio Engineering
SJR: 0.202 SNIP: 0.2 CiteScore™: 0.23

ISSN Imprimir: 0040-2508
ISSN En Línea: 1943-6009

Volumes:
Volumen 79, 2020 Volumen 78, 2019 Volumen 77, 2018 Volumen 76, 2017 Volumen 75, 2016 Volumen 74, 2015 Volumen 73, 2014 Volumen 72, 2013 Volumen 71, 2012 Volumen 70, 2011 Volumen 69, 2010 Volumen 68, 2009 Volumen 67, 2008 Volumen 66, 2007 Volumen 65, 2006 Volumen 64, 2005 Volumen 63, 2005 Volumen 62, 2004 Volumen 61, 2004 Volumen 60, 2003 Volumen 59, 2003 Volumen 58, 2002 Volumen 57, 2002 Volumen 56, 2001 Volumen 55, 2001 Volumen 54, 2000 Volumen 53, 1999 Volumen 52, 1998 Volumen 51, 1997

Telecommunications and Radio Engineering

DOI: 10.1615/TelecomRadEng.v68.i18.80
pages 1659-1686

Joint Estimation of Remote Sensing Images and Mixed Noise Parameters

M. L. Uss
National Aerospace University (Kharkov Aviation Institute), 17, Chkalov St., Kharkov, 61070
Benoit Vozel
University of Rennes 1, Enssat, Lannion, 22300, France
Kacem Chehdi
University of Rennes I, 6, Rue de Kerampont, 22 305 Lannion cedex, BP 80518, France

SINOPSIS

We address a joint task of remote sensing image enhancement and noise parameters’ estimation within a maximum likelihood framework. Estimation (blind determination) of noise parameters is an important operation in pre-processing images formed in varying or unknown imaging conditions. One peculiarity of our approach is that fractals (fBm-model) are used for modeling real-life images. Another peculiarity and advantage of the proposed approach consists in simultaneous evaluation of additive correlated noise variance and impulse noise occurrence probability. The core of our method is an iterative procedure of impulse noise detection and estimation of additive noise variance using pixels that are considered uncorrupted by impulses. Image model parameters are estimated as well with providing additional information for image interpretation. The designed method is tested for simulated and real life remote sensing images.