@InProceedings{LeiteSouz:2009:ClSuIm,
               author = "Leite, Emilson Pereira and Souza Filho, Carlos Roberto de",
          affiliation = "{Instituto de Geoci{\^e}ncias - Universidade Estadual de 
                         Campinas} and {Instituto de Geoci{\^e}ncias - Universidade 
                         Estadual de Campinas}",
                title = "Classifica{\c{c}}{\~a}o Supervisionada de Imagens Texturais 
                         Utilizando Redes Neurais Artificiais",
            booktitle = "Anais...",
                 year = "2009",
               editor = "Epiphanio, Jos{\'e} Carlos Neves and Galv{\~a}o, L{\^e}nio 
                         Soares",
                pages = "7821--7828",
         organization = "Simp{\'o}sio Brasileiro de Sensoriamento Remoto, 14. (SBSR)",
            publisher = "Instituto Nacional de Pesquisas Espaciais (INPE)",
              address = "S{\~a}o Jos{\'e} dos Campos",
             keywords = "Semivariogramas, Classifica{\c{c}}{\~a}o Supervisionada, Redes 
                         Neurais de Alimenta{\c{c}}{\~a}o Direta, Imagens Texturais, 
                         Imagens de RADAR.",
             abstract = "A methodology to perform supervised classification of textural 
                         images using Artificial Neural Networks for applications in the 
                         Geosciences is presented in this work. Feature vectors are built 
                         with textural information composed of semivariogram values, 
                         histogram measures of mean, standard deviation and weighted-rank 
                         fill ratio. Feed-forward back-propagation Artificial Neural 
                         Networks are designed and trained so as to minimize the mean 
                         squared error of the differences between feature and target 
                         vectors of training sets. At each training iteration, the mean 
                         squared error for validation and test sets are also evaluated. 
                         Global accuracy and kappa coefficient are calculated for training, 
                         validation and test sets, allowing a quantitative appraisal of the 
                         predictive power of the Neural Networks. The best model for 
                         classification of all pixels in a given textural image is obtained 
                         from a k-fold cross-validation. The methodology was tested using 
                         synthetic images and airborne, multi-polarized SAR imagery for 
                         geologic mapping, and the overall results are considered quite 
                         positive.",
  conference-location = "Natal",
      conference-year = "25-30 abr. 2009",
                 isbn = "978-85-17-00044-7",
             language = "pt",
         organisation = "Instituto Nacional de Pesquisas Espaciais (INPE)",
                  ibi = "dpi.inpe.br/sbsr@80/2008/11.17.10.14",
                  url = "http://urlib.net/ibi/dpi.inpe.br/sbsr@80/2008/11.17.10.14",
           targetfile = "7821-7828.pdf",
                 type = "T{\'e}cnicas de Classifica{\c{c}}{\~a}o e Minera{\c{c}}{\~a}o 
                         de Dados",
        urlaccessdate = "06 maio 2024"
}