Land use mapping with evidential fusion of features extracted from polarimetric synthetic aperture radar and hyperspectral imagery

Abstract

Abstract As part of the Earth Observation Application Development Program (EOADP) program sponsored by the Canada Space Agency, Lockheed Martin Canada––in partnership with Defense Research and Development Canada––has initiated a study aiming at studying the exploitation of the synergy between polarimetric SAR and hyperspectral imagery with the development of a system prototype called Intelligent Data Fusion System (IDFS). This system, based on the evidential fusion of features extracted from each type of imagery, is made of three modules. The polarimetric SAR module contains polarimetric classifiers, textural classifiers to provide hypotheses related to the textural and scattering properties of the illuminated area. The hyperspectral module contains an end-members selection technique and a collection of spectral indices. Each feature giving an incomplete representation of an object of interest, it is expected that the combination of complementary, redundant features will contribute to reduce the false alarm rate, improve the confidence in the target identification and the quality of the scene description as a whole. The combination of information provided by each feature extractor is performed according to the theory of evidence (Dempster–Shafer) in the third module. This paper presents an overview of the current functionality of IDFS. First results of evidential fusion are shown for land use mapping. Data sets were acquired over Indian-Head (Saskatchewan) with an airborne C-band polarimetric SAR and HSI Probe-1 sensors and were provided by the Canada Center for Remote Sensing.