Expectation-Maximization x Self-Organizing Maps for Image Classification

Abstract

To deal with the huge volume of information provided by remote sensing satellites, which produce images used for agriculture monitoring, urban planning, deforestation detection and so on, several algorithms for image classification have been proposed in the literature. This article compares two approaches, called Expectation-Maximization (EM) and Self-Organizing Maps (SOM) applied to unsupervised image classification, i.e. data clustering without direct intervention of specialist guidance. Remote sensing images are used to test both algorithms, and results are shown concerning visual quality, matching rate and processing time.