@InProceedings{PortesScJuFeCaGl:2011:AvPoCl,
author = "Portes, Raquel de Castro and Scudeller, Alice Azevedo and Jusksch,
Ivo and Fernandes Filho, Elp{\'{\i}}dio In{\'a}cio and Cardoso,
Irene Maria and Gleriani, Jos{\'e} Marinaldo",
affiliation = "{Universidade Federal de Vi{\c{c}}osa – UFV} and {Universidade
Federal de Vi{\c{c}}osa – UFV} and {Universidade Federal de
Vi{\c{c}}osa – UFV} and {Universidade Federal de Vi{\c{c}}osa –
UFV} and {Universidade Federal de Vi{\c{c}}osa – UFV} and
{Universidade Federal de Vi{\c{c}}osa – UFV}",
title = "Avalia{\c{c}}{\~a}o do potencial de classificadores
autom{\'a}ticos para mapeamento de uso e cobertura do solo sob
manejo agroecol{\'o}gico",
booktitle = "Anais...",
year = "2011",
editor = "Epiphanio, Jos{\'e} Carlos Neves and Galv{\~a}o, L{\^e}nio
Soares",
pages = "576--583",
organization = "Simp{\'o}sio Brasileiro de Sensoriamento Remoto, 15. (SBSR).",
publisher = "Instituto Nacional de Pesquisas Espaciais (INPE)",
address = "S{\~a}o Jos{\'e} dos Campos",
keywords = "Remote Sensing, Maximum Likelihood, Artificial Neural Netwoks,
SAFs, IKONOS, sensoriamento remoto, M{\'a}xima
verossimilhan{\c{c}}a, redes neurais artificiais e
Bhattacharya.",
abstract = "The production areas based in agroecology systems are being
implemented in Brazil and present on small farms intercropping
different species of plants and making the diverse agricultural
landscape. The classification of land cover and soil of these
areas requires the use of images with high spatial resolution for
detailed mapping and to identify the best method to rank areas
with heterogeneous patterns of use. This study aimed to evaluate
the potential of automatic classifiers for mapping land cover and
soil under agro-ecological management in the S{\~a}o Joaquim
River basin in Araponga, MG - Brazil. In the methodology were
performed field expeditions to collect the training samples and
validation using GPS. In the laboratory, supervised
classifications were performed on IKONOS image by the algorithms
of Maximum Likelihood and Artificial Neural Networks
(Backpropagation Error) and regions (Bhattacharya). Among the
classifiers evaluated in this experiment, the classification by
regions had the best result, with Kappa of 0.76. The ratings by
Maximum Likelihood and Artificial Neural Networks were
respectively 0.48 and Kappa 0.51. This demonstrates the great
potential that the supervised classification by segmentation have
on classifying areas with many classes of land cover and soil and
heterogeneous intra-class patterns. Thus, the findings of this
study besides being useful for future planning in the watershed,
will serve as universal knowledge to use classification and land
cover in other areas with agro-ecological management.",
conference-location = "Curitiba",
conference-year = "30 abr. - 5 maio 2011",
isbn = "{978-85-17-00056-0 (Internet)} and {978-85-17-00057-7 (DVD)}",
language = "pt",
organisation = "Instituto Nacional de Pesquisas Espaciais (INPE)",
ibi = "3ERPFQRTRW/3A645NL",
url = "http://urlib.net/ibi/3ERPFQRTRW/3A645NL",
targetfile = "p1357.pdf",
type = "Processamento de Imagens",
urlaccessdate = "04 maio 2024"
}