Land Use Mapping with Evidential Fusion of Polarimetric Synthetic Aperture Radar and Hyperspectral Imagery
Abstract
As part of the Earth Observation Application Development Program (EOADP) program sponsored by the Canada Space Agency, Lockheed Martin Canada has developed the Intelligent Data Fusion System (IDFS) for evidential fusion of features extracted from polarimetric SAR and Hyperspectral imagery. This paper presents the use of IDFS for land use mapping. IDFS is made of three modules. The polarimetric SAR module contains polarimetric classifiers (Cloude decomposition, polarization response parameters), textural classifiers (GLCM, backscattering coefficient) that provide hypotheses about the likelihood that some object of interest may be present in the scene based on textural and scattering properties of the analysed surface. The Hyperspectral module contains the Iterative Error Analysis endmembers selection technique proposed by the Canada Center for Remote Sensing to provide a set of pixel-based hypotheses reflecting the likelihood that some typical material may be present in the scene based on the spectral properties of the analysed surface. Hypotheses provided by each module represent an incomplete, inaccurate and imprecise description of the reality. The data fusion module combines PolSAR and HSI hypotheses using the evidence theory proposed by Dempster-Shafer. This paper presents an overview of the current functionality of IDFS. Results of evidential fusion are shown for land use mapping. The data-sets acquired over Indian-Head (Saskatchewan) with an airborne C-Band CV-580 PolSAR sensor and HSI Probe-1 imagery were provided by the Canada Center for Remote Sensing.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 01, 2002
- Accession Number
- ADA467018
Entities
People
- Alexandre Jouan
- Steve Allen
- Yannick Allard
Organizations
- Lockheed Martin Canada