Observatorio de I+D+i UPM

Memorias de investigación
Communications at congresses:
Multisource analysis strategies for remote sensing change detection
Year:2014
Research Areas
  • Remote sensing,
  • Images pursuit,
  • Image processing
Information
Abstract
In remote sensing, the change detection process can be approached from different perspectives. However, they all require a common set of previous operations as image registration and radiometric normalization. In this respect, these processes can be undertaken either in a relative or absolute manner. Among the so called relative methods, basic statistic operations and more complex procedures as methods based on Point Invariant Features (PIF) are identified. The latter can also be undertaken by means of zonal features. The second set of radiometric adjustments requires the application of atmospheric corrections. Instead of deciding between either of these methods, and in order to improve the radiometric similarity between the two images involved in the change detection process, this work describes first a methodology that combines both types of radiometric normalization techniques, and also applies the concept of zonal invariant features. Another important aspect to consider in image change detection is the use of Change Indices (CI). Traditionally, the change detection process is based on the determination of change and no_change categories based on the calculation of a specific change index, to which thresholding techniques or probabilistic decision rules are applied to separate these categories. The second part of this paper suggests the derivation of a set of different Change Indices, as well as the evaluation of the most representatives, in order to be subsequently integrated into a multi-source analysis, where a parametric probabilistic algorithm and a non-parametric as the Support Vector Machine are evaluated and compared. Both, the proposed radiometric normalization methodology as well as the resulting assignation processes of change and no_change categories based on multisource analysis have shown to be properly combined, reaching also correct user and producer accuracies for the purchased categories.
International
Si
Congress
RECENT ADVANCES IN QUANTITATIVE REMOTE SENSING
960
Place
Torrent (Spain)
Reviewers
Si
ISBN/ISSN
1682-1750
Start Date
22/09/2014
End Date
26/09/2014
From page
139
To page
139
Abstract Book
Participants
  • Autor: M.Estibaliz Martinez Izquierdo (UPM)
  • Autor: Iñigo Molina Sanchez (UPM)
  • Autor: Agueda Arquero Hidalgo (UPM)
Research Group, Departaments and Institutes related
  • Creador: Grupo de Investigación: Geovisualización, Espacios Singulares y Patrimonio
  • Departamento: Ingeniería Topográfica y Cartografía
  • Departamento: Arquitectura y Tecnología de Sistemas Informáticos
S2i 2020 Observatorio de investigación @ UPM con la colaboración del Consejo Social UPM
Cofinanciación del MINECO en el marco del Programa INNCIDE 2011 (OTR-2011-0236)
Cofinanciación del MINECO en el marco del Programa INNPACTO (IPT-020000-2010-22)