Please use this identifier to cite or link to this item: http://repositorio.unicamp.br/jspui/handle/REPOSIP/244218
Type: Artigo de periódico
Title: Improving Land Cover Classification Through Contextual-based Optimum-path Forest
Author: Osaku
D.; Nakamura
R. Y. M.; Pereira
L. A. M.; Pisani
R. J.; Levada
A. L. M.; Cappabianco
F. A. M.; Falco
A. X.; Papa
Joao P.
Abstract: Traditional machine learning algorithms very often assume statistically independent data samples. However, this is clearly not the case in remote sensing image applications, in which pixels present spatial and/or temporal dependencies. In this work, it has been presented an approach to improve land cover image classification using a contextual approach based on optimum-path forest (OPF) and the well-known Markov random fields (MRFs), hereinafter called OPF-MRF. In addition, it is also introduced a framework to the optimization of the amount of contextual information used by OPF-MRF. Experiments over high- and medium-resolution satellite (CBERS-2B, Landsat 5 TM, lkonos-2 MS and Geoeye) and radar (ALOS-PALSAR) images covering the area of two Brazilian cities have shown the proposed approach can overcome several shortcomings related to standard OPF classification. In some cases, the proposed approach outperformed traditional OFF in about 9% of recognition rate, which is crucial for land cover classification. (C) 2015 Elsevier Inc. All rights reserved.
Subject: Remotely-sensed Data
Hyperspectral Images
Resolution Images
Neural Networks
Sensing Images
Framework
Fusion
Models
System
Country: NEW YORK
Editor: ELSEVIER SCIENCE INC
Citation: Improving Land Cover Classification Through Contextual-based Optimum-path Forest. Elsevier Science Inc, v. 324, p. 60-87 DEC-2015.
Rights: embargo
Identifier DOI: 10.1016/j.ins.2015.06.020
Address: http://www.sciencedirect.com/science/article/pii/S0020025515004521
Date Issue: 2015
Appears in Collections:Unicamp - Artigos e Outros Documentos

Files in This Item:
File SizeFormat 
wos_000362307200005.pdf9.54 MBAdobe PDFView/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.