@InProceedings{NevesKortGiroFons:2017:MiDaSe,
               author = "Neves, Alana Kasahara and Korting, Thales Sehn and Girolamo Neto, 
                         Cesare Di and Fonseca, Leila Maria Garcia",
          affiliation = "{Instituto Nacional de Pesquisas Espaciais (INPE)} and {Instituto 
                         Nacional de Pesquisas Espaciais (INPE)} and {Instituto Nacional de 
                         Pesquisas Espaciais (INPE)} and {Instituto Nacional de Pesquisas 
                         Espaciais (INPE)}",
                title = "Minera{\c{c}}{\~a}o de dados de sensoriamento remoto para 
                         detec{\c{c}}{\~a}o e classifica{\c{c}}{\~a}o de {\'a}reas de 
                         pastagem na Amaz{\^o}nia Legal",
            booktitle = "Anais...",
                 year = "2017",
               editor = "Gherardi, Douglas Francisco Marcolino and Arag{\~a}o, Luiz 
                         Eduardo Oliveira e Cruz de",
                pages = "2508--2515",
         organization = "Simp{\'o}sio Brasileiro de Sensoriamento Remoto, 18. (SBSR)",
            publisher = "Instituto Nacional de Pesquisas Espaciais (INPE)",
              address = "S{\~a}o Jos{\'e} dos Campos",
             abstract = "Most of deforested areas in the Brazilian Amazon are occupied by 
                         pasture lands. The main cause of pasture degradation in this 
                         region is related to the condition of vegetation cover because of 
                         the fast regrowth and the competition with invasive plants. The 
                         aim of this study is to semi-automatically detect and classify 
                         patterns of pasture lands in the Legal Amazon, using time series 
                         of remote sensing images and data mining techniques, according to 
                         the conditions of the vegetation cover. The study site is the 
                         path/row 001/67 from Landsat 8 satellite. 28 images of surface 
                         reflectance, from 2013 to 2015, were used to construct the time 
                         series. Two classification methods were used: per pixel and object 
                         based. The following features were extracted from each image: 
                         vegetation indexes, fractions from the Spectral Linear Unmixing 
                         Model and components from the Tasseled Cap Transformation. The 
                         first step of the classification consisted in identifying pasture 
                         pattern, distinguishing class Pasture from Vegetation and Others. 
                         Later on, the pasture areas were reclassified into Clear Pasture 
                         (herbaceous pasture) and Dirty Pasture (shrubby pasture). In order 
                         to better evaluate the results, a classification procedure 
                         involving all classes was performed. The classification was 
                         validated by visual interpretation of a high spatial resolution 
                         image (RapidEye). The best accuracy was obtained on the object 
                         based approach, where it reached around 90%. Considering the 
                         per-pixel approach, it was difficult to identify some pasture due 
                         to the great amount of mixed elements in the images, like patterns 
                         of grass, tree, bush and others.",
  conference-location = "Santos",
      conference-year = "28-31 maio 2017",
                 isbn = "978-85-17-00088-1",
                label = "59263",
             language = "pt",
         organisation = "Instituto Nacional de Pesquisas Espaciais (INPE)",
                  ibi = "8JMKD3MGP6W34M/3PSLQN4",
                  url = "http://urlib.net/ibi/8JMKD3MGP6W34M/3PSLQN4",
           targetfile = "59263.pdf",
                 type = "Classifica{\c{c}}{\~a}o e minera{\c{c}}{\~a}o de dados",
        urlaccessdate = "27 abr. 2024"
}