Fast Huynen–Euler Decomposition and its Application in Disaster Monitoring

Abstract

Huynen–Euler (H–E) parameters proposed based on the diagonalization of the scattering matrix are of significant importance for single target information interpretation because of their explicit physical meanings. However, their application in target classification/recognition seems unsuccessful hitherto. Besides, the process of extracting the five H–E parameters by existing approaches, i.e., eigen-decomposition and unitary transformation, is a bit tedious and relatively time consuming, especially for large-sized polarimetric synthetic aperture radar (PolSAR) data. In this article, new H–E parameters are proposed to improve the performance of H–E decomposition in describing the scattering characteristics of the target. Furthermore, a fast decomposition approach is presented to derive the parameters directly and simultaneously with high computational efficiency. Another advantage of the fast H–E decomposition (FHED) is that, different from the original algorithm, which can only be applied to the single target, FHED is also effective for distributed targets, which expands the application range of the H–E decomposition. Experimental results on the temporal PolSAR data show that the changes in the disaster-affected area can be detected according to the changes in the newly proposed parameters, and the degree of change in the newly proposed skip angle has a linear relationship with the degree of urban damage. This indicates that FHED has a good prospect in disaster monitoring, especially for the estimation of building damage level (DL). It is also proved that the building DL mapped by the new skip angle and by the double-bounce scattering power, the most widely used parameter for such a situation, are highly consistent, while the former consumes much less time. Therefore, FHED can be applied to disaster monitoring and damage detection effectively, which is conducive to rescue operations by providing important information with a quick response.