Back to search

Aristotle University Of Thessaloniki (AUTH)

Ανάπτυξη αυτο-οργανούμενων ασαφών νευρωνικών πολυστρωματικών ταξινομητών: εφαρμογή στο πρόβλημα ταξινόμησης κάλυψης γης με χρήση πολυφασματικών δορυφορικών εικόνων υπερ-υψηλής ευκρίνειας

Abstract

dc:description

The purpose of this doctoral dissertation is to develop an innovative neuro-fuzzy multilayered classifier with self-organizing and feature selection capabilities. The proposed architecture implements a hierarchical interconnection of small scale elementary neuro-fuzzy classifiers, which are distributed in layers. In each layer, two parent classifiers are combined to produce a new elementary classifier with enhanced classification capabilities. In each neuron, two new units are introduced: the first fuses the parent classifiers’ decisions and while the latter splits the data set into confident classified patterns and ambiguous ones. The network of the elementary classifiers evolves in a self-organizing manner, following a systematic structure learning procedure based on the GMDH algorithm. Improved versions of the GMDH algorithm are suggested for the effective recognition of classifiers combination and to reduce the computational cost. Simulation results on benchmark problems indicate the effectiveness of the proposed methods. Finally, the suggested classifiers are tested on the land cover classification of the wetland of international ecological importance of Lake Koroneia. To this end, a very high resolution multispectral IKONOS satellite image was used, and additional features were extracted not only using linear transformations of the original bands but also texture transformations. New methods are suggested for the computation of the essential parameters used in texture features derivation and for the effective use of the available information. The high final classification accuracy and the quality of the derived thematic maps indicate the superior performance of the proposed methods comparing to other popular methods.

Degree

thesis:*
Grantor dc:publisher
Aristotle University Of Thessaloniki (AUTH)
Year dc:date
2008

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Mitrakis, Nikolaos
  • Μητράκης, Νικόλαος

Subjects

dc:subject × 16

Rights

Language dc:language
gre

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/19678

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Mitrakis, Nikolaos; Μητράκης, Νικόλαος. Ανάπτυξη αυτο-οργανούμενων ασαφών νευρωνικών πολυστρωματικών ταξινομητών: εφαρμογή στο πρόβλημα ταξινόμησης κάλυψης γης με χρήση πολυφασματικών δορυφορικών εικόνων υπερ-υψηλής ευκρίνειας. Aristotle University Of Thessaloniki (AUTH), 2008. http://hdl.handle.net/10442/hedi/19678