Determining the accuracy in image supervised classification problems

Abstract

A large number of accuracy measures for crisp supervised classification have been developed in supervised image classification literature. Overall accuracy, Kappa index, Kappa location, Kappa histo and user accuracy are some well-known examples. In this work, we will extend and analyze some of these measures in a fuzzy framework to be able to measure the goodness of a given classifier in a supervised fuzzy classification system with fuzzy reference data. In addition with this, the measures here defined also take into account the preferences of the decision maker in order to differentiate some errors that must not be considered equal in the classification process.