Coherent Change Detection Techniques for Repeat Pass Polarimetric SAR Interferometry

SAR Coherent Change Detection (CCD) is a technique that exploits the sensitivity of the transduced SAR image phase and speckle pattern to detect subtle scene disturbanes that may occur in the interval between a repeat pass imaging collection. Polarimetric SAR provides additional scene information that may be exploited to provide enhanced CCD products and yield improved detection performance. In this paper two approaches to exploiting polarimetry to yield improved CCD products are examined. In the first approach proposed by Novak, the polarimetric coherent change detection problem is modelled via a single correlation parameter that describes the loss in coherence across all polarimetric channels. In the second approach, referred to as the fully polarimetric Log Likelihood Change Statistic LLCS, the change detection problem is formulated in a Bayesian hypothesis testing framework. Through simulation and application to experimental data it is shown that the additional polarimetric channels available in a fully polarimetric SAR can yield significant improvements in detection performance. Furthermore the LLCS outperforms the polarimetric coherence approach however, this is at the expense of significant complexity in establishing the models for the hypothesis test.