Back to search

Aristotle University Of Thessaloniki (AUTH)

Τηλεπισκοπικές και γεωπληροφοριακές μέθοδοι στη μελέτη περιβαλλοντικών θεμάτων: η αντικειμενοστραφής ανάλυση δορυφορικών δεδομένων στην ανίχνευση ορίων επιφανειακών χωρικών οντοτήτων

Abstract

dc:description

The science of remote sensing is an important "tool" for the investigation of various fields of Earth and environmental sciences, offering the possibility of studying extensive areas in a cost-effective way, while at the same time recording valuable information about the Earth's surface. The integration of remote sensing data into a Geographic Information System (GIS), enables the synergistic processing of spatial data from multiple sources, through processes such as visual interpretation, modeling, visualization, land-use/cover classification and geospatial analysis, contributing to decision-making processes. The remote sensing community, after significant advances in sensor technology, faced the challenge of searching for new techniques for utilizing and analyzing image data. On this basis, a framework has emerged that uses the object as the main unit for analysis and has contributed to the exploitation of this new data. This framework is object-based image analysis (OBIA). OBIA is based on dividing a satellite image into meaningful objects (rather than individual pixels). In particular, it evaluates their discrete features through spectral and spatial scale as opposed to traditional pixel-based techniques that exclusively examine the spectral signatures of features to extract useful thematic information. Image segmentation forms the initial and integral part in the production of the fundamental elements of OBIA. It is about dividing an image into spatially connected and homogeneous (spectrally) regions (segments) that are the basis for further analysis. In this thesis, environmental problems are considered based on the surface spatial entity, facilitating the clear identification of each object of interest. The definition of this entity is particularly important for the design and development of the appropriate methodology for detecting its boundaries. The integration of object-based image analysis into such a methodological framework relied on the fact that OBIA is based on the division of a satellite imagery into meaningful objects, which in our case are the spatial reference entities. Initially, the theoretical background of the thesis includes extensive bibliographic research concerning object-based image analysis and in particular segmentation methods, from the established to the most modern ones. Following, a thorough review and meta-analysis of modern applications of image segmentation methods is presented. The theoretical part is completed with the basic building blocks of the Python programming language that was mainly utilized in the experimental part of the thesis. Subsequently, the research that was implemented through the initial experimental approaches is presented. This led to the development of an image segmentation plugin (Terminus), its integration into QGIS software and the formulation of object-based analysis methodological frameworks for satellite data. The Terminus image segmentation plugin was developed in the Python programming language to provide a user-friendly and easy-to-use tool for segmenting satellite images. In addition, it enables the user to calculate various descriptive statistics for each segment, which can be used to measure certain quantitative characteristics of the image segmentation result, as well as for the subsequent stage of image classification (if necessary). It should also be noted that each of the algorithms included in the plugin produces two results, in the form of a vector and a raster file. The first can be used to implement an object-based classification, while the second can be used to display a user-defined color composite and integrate it into a semantic segmentation methodology. In the framework of the doctoral thesis, methodologies and techniques for detecting spatial entity boundaries were designed, developed and evaluated for the study of environmental issues based on object-based image analysis. The developed methodological frameworks and corresponding experimental approaches include the integration of spectral indices and geoinformation methods for detecting surface mine boundaries and areas affected by a catastrophic phenomenon (fire) in optical data and surface water bodies with SAR data. Also, an object-based analysis of thermal satellite data and convolutional neural network models (CNN) to detect boundaries of land cover areas and flood-prone areas (forecast) are included.

Degree

thesis:*
Grantor dc:publisher
Aristotle University Of Thessaloniki (AUTH)
Year dc:date
2022

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Kotaridis, Ioannis
  • Κοταρίδης, Ιωάννης

Subjects

dc:subject × 18

Rights

Language dc:language
gre

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/51936

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

Kotaridis, Ioannis; Κοταρίδης, Ιωάννης. Τηλεπισκοπικές και γεωπληροφοριακές μέθοδοι στη μελέτη περιβαλλοντικών θεμάτων: η αντικειμενοστραφής ανάλυση δορυφορικών δεδομένων στην ανίχνευση ορίων επιφανειακών χωρικών οντοτήτων. Aristotle University Of Thessaloniki (AUTH), 2022. http://hdl.handle.net/10442/hedi/51936