Avaliação de métodos de classificação por regiões em imagens de alta resolução do sensor Rapideye

Main Article Content

Luiz Felipe Parente Santiago
Wagner Barreto da Silva

Abstract

This work aims in the assessment of region based image classification methods, Support Vector Machine (SVM) and Minimum Stochastic Distance (SMD), and to compare them with pixel based image classification methods, SVM and Maximum Likelihood, applied to high resolution imagery from RapidEye satellite. The assessment will be carried out according to the confusion matrix and the Kappa Concordance Coefficient. Images from Rondônia State were used. Such images have statial resolution of 5 meters. In order to get some samples from the land cover classes, cartographic features acquired from 5ª DL staff were used. From
the performed evaluations, one can say that the classifications by regions were on average 13% higher than SVM by pixel and 34% higher than the MaxVer by pixel in Area 1, 3% and 11% in Area 2 and 10% and 11% In Area 3.

Downloads

Download data is not yet available.

Article Details

How to Cite
Santiago, L. F. P., & Silva, W. B. da. (2018). Avaliação de métodos de classificação por regiões em imagens de alta resolução do sensor Rapideye. Revista Militar De Ciência E Tecnologia, 35(2), 22–26. Retrieved from https://ebrevistas.eb.mil.br/index.php/CT/article/view/1950
Section
Artigos