Adaptive Unscented Kalman Filter Phase Unwrapping Method and Its Application on Gaofen-3 Interferometric SAR Data

Abstract

Phase unwrapping (PU) is a key step in the reconstruction of digital elevation models (DEMs) and the monitoring of surface deformation from interferometric synthetic aperture radar (SAR, InSAR) data. In this paper, an improved PU method that combines an amended matrix pencil model, an adaptive unscented kalman filter (AUKF), an efficient quality-guided strategy based on heapsort, and a circular median filter is proposed. PU theory and the existing UKFPU method are covered. Then, the improved method is presented with emphasis on the AUKF and the circular median filter. AUKF has been well used in other fields, but it is for the first time applied to interferometric images PU, to the best of our knowledge. First, the amended matrix pencil model is used to estimate the phase gradient. Then, an AUKF model is used to unwrap the interferometric phase based on an efficient quality-guided strategy based on heapsort. Finally, the key results are obtained by filtering the results using a circular median. The proposed method is compared with the minimum cost network flow (MCF), statistical cost network flow (SNAPHU), regularized phase tracking technique (RPTPU), and UKFPU methods using two sets of simulated data and two sets of experimental GF-3 SAR data. The improved method is shown to yield the greatest accuracy in the interferometric phase maps compared to the methods considered in this paper. Furthermore, the improved method is shown to be the most robust to noise and is thus most suitable for PU of GF-3 SAR data in high-noise and low-coherence regions.

Modeling the Optimal Baseline for a Spaceborne Bistatic SAR System to Generate DEMs

“…Further studies of the influence of other coherence values are necessary to obtain an accurate optimal baseline range. In addition, to suppress the related unwrapping error, this paper chose the adaptive unscented Kalman filter phase unwrapping algorithm [13,14], which is superior to the traditional phase unwrapping methods. In future research, a better processing method will be selected to obtain a higher-precision DEM; therefore, choosing a better phase unwrapping algorithm for experimental analysis in this paper will be more conducive to follow-up research.…”

“…Because interferogram filtering would introduce more uncertainty factors, it is not performed. To keep up with the development of phase unwrapping algorithm and ensure a certain success rate of phase unwrapping, this paper adopts an adaptive unscented Kalman filter phase unwrapping method [13,14], which is relatively optimal. Compared to traditional phase unwrapping methods, this method that combines an amended matrix pencil model, an adaptive unscented Kalman filter, an efficient quality-guided strategy based on heapsort, and a median filter could achieve more accurate interferometric phase maps.…”

“…Thus, the interferometric phase error model should be statistically analyzed for the different simulated terrain slope data based on the phase unwrapping. To keep up with the development of the phase unwrapping algorithm and facilitate follow-up studies, the adaptive unscented Kalman filter phase unwrapping method [13,14], which is a relatively optimal algorithm, will be selected to suppress the related unwrapping error. Meanwhile, to avoid unnecessary phase error sources, such as temporal decorrelation and volume decorrelation, we prefer the bistatic interferometric platform for modeling.…”

Refined Two-Stage Programming Approach of Phase Unwrapping for Multi-Baseline SAR Interferograms Using the Unscented Kalman Filter

“…In Equations (23) and (24), P i ψψ (a, s) is the estimation error covariance of the unwrapped phase at pixel (a, s) in interferogram i, SNR i (a, s) is the signal-to-noise ratio (SNR) of pixel (a, s) in interferogram i, and γ i(a,s) is the coherence coefficient for pixel (a, s) in interferogram i.…”

“…The Kalman filter PU technique is famous for its strong noise robustness in the SB PU domain because it allows PU to be performed simultaneously with noise filtering [24][25][26]. Due to its robust PU accuracy, the unscented Kalman filter (UKF) technique has been widely used in SB PU.…”