Accurate DInSAR stack coherence estimation exploiting phase statistics

Differential Synthetic Aperture Radar (SAR) Interferometry (DInSAR) coherence, a measurement of the similarity between two radar acquisitions, is a widely used index to quantitatively evaluate the quality of derived interferograms considering its direct relationship with Signal to Noise Ratio (SNR). When interferometric measurements originate from both point-like and distributed targets, the coherence has to be correctly estimated by adaptively grouping and averaging pixels sharing similar scattering properties. In this paper, we present a method to estimate coherence for DInSAR stack by exploiting the multi-temporal phase statistics. The key point of the proposed solution relies on the capability of averaging each pixel with its statistical homogeneous neighborhood, thus allowing a more accurate coherence estimation. Experimental results on real SAR data acquired by ENVISAT satellite, over Shanghai City, China demonstrate the validity of proposed method.