@InProceedings{HappFeitStre:2012:AsOpMe,
               author = "Happ, Patrick and Feitosa, Raul and Street, Alexandre",
                title = "Assessment of optimization methods for automatic tunning of 
                         segmentation parameters",
            booktitle = "Proceedings...",
                 year = "2012",
               editor = "Feitosa, Raul Queiroz and Costa, Gilson Alexandre Ostwald Pedro da 
                         and Almeida, Cl{\'a}udia Maria de and Fonseca, Leila Maria Garcia 
                         and Kux, Hermann Johann Heinrich",
                pages = "490--495",
         organization = "International Conference on Geographic Object-Based Image 
                         Analysis, 4. (GEOBIA).",
            publisher = "Instituto Nacional de Pesquisas Espaciais (INPE)",
              address = "S{\~a}o Jos{\'e} dos Campos",
             keywords = "Image Segmentation, Parameter Adjustment, Optimization, Genetic 
                         Algorithm, Derivative-Free Optimization.",
             abstract = "The image segmentation is a key step in the image classification 
                         process since its quality will directly affects the classification 
                         result. The quality measure of image segmentation has been widely 
                         discussed in image analysis leading to the development of 
                         different metrics in order to try to automate the process and 
                         replace the subjective analysis of a specialist. These metrics are 
                         also known as similarity metrics (or functions) and evaluate the 
                         segmentation outcome comparing it with a given image containing 
                         some reference objects and returning a numerical value that 
                         express the similarity between the result and the expected 
                         references. As the quality can be expressed by a metric, the 
                         problem lies in achieving a small similarity value. This task is 
                         related to the input segmentation parameters that vary according 
                         to the image features and the classes of objects of interest. 
                         Given that the relation between the parameters and the 
                         segmentation quality can not be formulated, this procedure is 
                         generally done by a trial and error process. To avoid misleading 
                         and time consuming, automatic parameter tuning are proposed using 
                         genetic algorithms. However, this solution tends to have a high 
                         computational cost and another several parameters to tune. This 
                         work compares this solution with some derivative-free optimization 
                         methods to present some alternatives that have smaller 
                         computational cost.",
  conference-location = "Rio de Janeiro",
      conference-year = "May 7-9, 2012",
                 isbn = "978-85-17-00059-1",
             language = "en",
         organisation = "Instituto Nacional de Pesquisas Espaciais (INPE)",
                  ibi = "8JMKD3MGP8W/3BTFEM5",
                  url = "http://urlib.net/ibi/8JMKD3MGP8W/3BTFEM5",
           targetfile = "131.pdf",
                 type = "Segmentation",
        urlaccessdate = "09 maio 2024"
}