Fuzzy Regions for Handling Uncertainty in Remote Sensing Image Segmentation

  • Authors:

  • Ivan Lizarazo;Paul Elsner

  • Affiliations:

  • Cadastral Engineering and Geodesy Department, Universidad Distrital Francisco Jose de Caldas, Bogota, Colombia;School of Geography, Birkbeck College, University of London, London, UK

  • Venue:
  • ICCSA '08 Proceeding sof the international conference on Computational Science and Its Applications, Part I
  • Year:
  • 2008

Quantified Score

Hi-index 0.01

Visualization

Abstract

Increasing availability of satellite imagery is demanding robust image classification methods to ensure a better integration between remote sensing and GIS. Segmentation-based approaches are becoming a popular alternative to traditional pixel-wise methods. Hard segmentation divides an image into a set of non-overlapping image-objects and regularly requires significant user-interaction to parameterise a functional segmentation model. This paper proposes an alternative image segmentation method which outputs fuzzy image-regions expressing degrees of membership to target classes. These fuzzy regions are then defuzzified to derive the eventual land-cover classification. Both steps, fuzzy segmentation and defuzzification, are implemented here using simple statistical learning methods which require very little user input. The new procedure is tested in a land-cover classification experiment in an urban environment. Results show that the method produces good thematic accuracy. It therefore provides a new, automated technique for handling uncertainty in the image analysis process of high resolution imagery.