@InProceedings{RomaniAmaGonZulSou:2013:ApTéCl,
author = "Romani, Luciana Alvim Santos and Amaral, Bruno Ferraz do and
Gon{\c{c}}alves, Renata Ribeiro do Valle and Zullo Junior,
Jurandir and Sousa, Elaine Parros Machado de",
title = "Aplica{\c{c}}{\~a}o de t{\'e}cnicas de
classifica{\c{c}}{\~a}o semissupervisionada para an{\'a}lise de
s{\'e}ries multitemporais de imagens de sat{\'e}lite",
booktitle = "Anais...",
year = "2013",
editor = "Epiphanio, Jos{\'e} Carlos Neves and Galv{\~a}o, L{\^e}nio
Soares",
pages = "1750--1757",
organization = "Simp{\'o}sio Brasileiro de Sensoriamento Remoto, 16. (SBSR)",
publisher = "Instituto Nacional de Pesquisas Espaciais (INPE)",
address = "S{\~a}o Jos{\'e} dos Campos",
abstract = "In the last years remote sensing has become an important tool to
support the agricultural crop monitoring in the whole world. In
the most of studies involving agriculture specialists have applied
medium and high spatial resolution satellites, preferentially.
However, an important question is: can low spatial resolution
satellites also be used in this monitoring task? In this context,
the focus of this work is how to take advantage of the high
temporal resolution which is characteristic of this kind of
satellite to identify agricultural fields such as sugarcane.
Accordingly, this paper proposes to apply two semi-supervisioned
classification methods to classify multitemporal satellite images.
Thus, we adapted the LNP (Linear Neighborhood Propagation) and
HC-LGT (Hierarchical Clustering and Local Graph Transduction)
methods to be employed in classification of time series extracted
from NDVI-NOAA images. The training dataset was generated with
time series of six (agricultural crops, sugarcane, urban areas,
forest, pasture and perennial crops) different classes defined by
agrometeorologists. The study area encloses state of S{\~a}o
Paulo in Brazil where is cultivated large sugarcane fields.
Results with both classification methods showed that time series
from low spatial resolution satellites can be satisfactorily used
to identify regions of agricultural fields as well as forests,
pasture and urban areas. Additionally, the classification
generated by both methods can also be used to identify the
vegetation cover of a specific region as an initial step before
applying more complex strategies.",
conference-location = "Foz do Igua{\c{c}}u",
conference-year = "13-18 abr. 2013",
isbn = "{978-85-17-00066-9 (Internet)} and {978-85-17-00065-2 (DVD)}",
label = "1237",
language = "pt",
organisation = "Instituto Nacional de Pesquisas Espaciais (INPE)",
ibi = "3ERPFQRTRW34M/3E7GK25",
url = "http://urlib.net/ibi/3ERPFQRTRW34M/3E7GK25",
targetfile = "p1237.pdf",
type = "An{\'a}lise e Aplica{\c{c}}{\~a}o de Imagens Multitemporais",
urlaccessdate = "17 jun. 2024"
}