Mining satellite image database of landscapes and application for urban zones: clustering, consensus and categorisation

Abstract : Remote sensed satellite images have found a wide application for analysing and managing natural resources and human activities. Satellite images of high resolution, e.g., SPOT5, have large sizes and are very numerous. This gives a large interest to develop new theoretical aspects and practical tools for satellite image mining. The objective of the thesis is unsupervised satellite image mining and includes three main parts. In the first part of the thesiswe demonstrate content of high resolution optical satellite images. We describe image zones by texture and geometrical features. Unsupervised clustering algorithms are presented in the second part of the thesis. A review of validity criteria and information measures is given in order to estimate the quality of clustering solutions. A new criterion based on Minimum Description Length (MDL) is proposed for estimating the optimal number of clusters. In addition, we propose a new kernel hierarchical clustering algorithm based on kernel MDL criterion. A new method of
Document type :
Theses
Complete list of metadatas

Cited literature [127 references]  Display  Hide  Download

https://pastel.archives-ouvertes.fr/pastel-00004084
Contributor : Ecole Télécom Paristech <>
Submitted on : Friday, January 9, 2009 - 8:00:00 AM
Last modification on : Wednesday, February 20, 2019 - 2:40:43 PM
Long-term archiving on : Friday, September 10, 2010 - 12:49:26 PM

Identifiers

  • HAL Id : pastel-00004084, version 1

Citation

Ivan Kyrgyzov. Mining satellite image database of landscapes and application for urban zones: clustering, consensus and categorisation. domain_other. Télécom ParisTech, 2008. English. ⟨pastel-00004084⟩

Share

Metrics

Record views

554

Files downloads

916