Unsupervised classification of SAR imagery using polarimetric decomposition to preserve scattering characteristics

Abstract

We propose an unsupervised classification method using polarimetric synthetic aperture radar data to detect anomalies on earthen levees. This process mainly involves two stages: 1. Apply the scattering model-based decomposition developed by Freeman and Durden to divide pixels into three scattering categories: surface scattering, volume scattering, and double-bounce scattering. A class initialization scheme is also performed to initially merge clusters from many small clusters in each scattering category by applying a merge criterion developed based on the Wishart distance measure. 2. The iterative Wishart classifier is applied, which is a maximum likelihood classifier based on the complex Wishart distribution. This method not only uses a statistical classification, but also preserves the purity of dominant polarimetric scattering properties, and is superior to the entropy/anisotropy/Wishart classifier. An automated color rendering scheme is applied, based on the classes' scattering category to code the pixels. The effectiveness of the algorithms is demonstrated using fully quad-polarimetric L-band SAR imagery from the NASA Jet Propulsion Laboratory's (JPL's) Uninhabited Aerial Vehicle Synthetic Aperture Radar (UAVSAR). The study area is a section of the lower Mississippi River valley in the southern USA, where earthen flood control levees are maintained by the US Army Corps of Engineers.