@InProceedings{AnjosAlmeGalv:2015:IdMaUr,
author = "Anjos, Camila Souza and Almeida, Cl{\'a}udia Maria de and
Galv{\~a}o, L{\^e}nio Soares",
affiliation = "{} and {Instituto Nacional de Pesquisas Espaciais (INPE)} and
{Instituto Nacional de Pesquisas Espaciais (INPE)}",
title = "Identifica{\c{c}}{\~a}o de materiais urbanos por meio de
m{\'e}todos inovadores de classifica{\c{c}}{\~a}o de imagens",
booktitle = "Anais...",
year = "2015",
editor = "Gherardi, Douglas Francisco Marcolino and Arag{\~a}o, Luiz
Eduardo Oliveira e Cruz de",
pages = "4377--4384",
organization = "Simp{\'o}sio Brasileiro de Sensoriamento Remoto, 17. (SBSR)",
publisher = "Instituto Nacional de Pesquisas Espaciais (INPE)",
address = "S{\~a}o Jos{\'e} dos Campos",
abstract = "Urban areas represent one of the most challenging environments for
remote sensing analysis. In general, urban areas represent
spatially and spectrally complex environments, where exposed
targets show profuse geometric and compositional variations. In
this context, the use of images with high spatial and spectral
resolutions is an excellent alternative for urban studies. The
combination of these imagery characteristics allow a potential
improvement for detection and discrimination of urban targets,
especially using automatic classifiers. This research uses optical
ultispectral data with very high spatial resolution (VHR) acquired
by the WorldView-2 satellite in eight spectral channels. The study
area is a transect in the campus of the State University of
Campinas (UNICAMP), located in the Campinas municipality,
southeast of S{\~a}o Paulo State Brazil. The area comprises a
diversity of urban targets, such as French tiles, metal roofs,
concrete/asphalt, water, low vegetation, woody vegetation, among
others. The imagery dataset were processed by means of
nonparametric classifiers. Three classification experiments were
performed using the following techniques: (i) Decision Tree,
Support Vector Machines (SVM), and (ii) Random Forest (RF). A
comparative evaluation among these three nonparametric
classification methods was also produced, seeking to examine the
confusion matrix and the Kappa index. The results indicated that
all classifiers showed high performance with Kappa values greater
than 0.8. The SVM got the best Kappa result (0.93).",
conference-location = "Jo{\~a}o Pessoa",
conference-year = "25-29 abr. 2015",
isbn = "978-85-17-0076-8",
label = "859",
language = "pt",
organisation = "Instituto Nacional de Pesquisas Espaciais (INPE)",
ibi = "8JMKD3MGP6W34M/3JM4CN2",
url = "http://urlib.net/ibi/8JMKD3MGP6W34M/3JM4CN2",
targetfile = "p0859.pdf",
type = "Classifica{\c{c}}{\~a}o e minera{\c{c}}{\~a}o de dados",
urlaccessdate = "27 abr. 2024"
}