A Sequential Monte Carlo Framework for Noise Filtering in InSAR Time Series

Abstract

This study proposes an alternative filtering tech-1 nique to improve interferometric synthetic aperture radar (In-2 SAR) time series by reducing residual noise while retaining the 3 ground deformation signal. To this end, for the first time, a 4 data-driven approach is introduced, which is based on Takens's 5 method within the sequential Monte Carlo framework, allowing 6 for a model-free approach to filter noisy data. Both Kalman-7 based and particle filters are applied within this framework 8 to investigate their impact on retrieving the signals. More 9 specifically, particle filter (PF) and particle smoother (PaSm; 10 to avoid confusion with persistent scatterers PS) are tested for 11 their ability to deal with non-Gaussian noise. A synthetic test 12 based on simulated InSAR time series, as well as a real test, are 13 designed to investigate the capability of the proposed approach 14 compared with spatio-temporal filtering of InSAR time series.

A Deep Convolutional Neural Network With Multiscale Feature Dynamic Fusion for InSAR Phase Filtering

“…(Corresponding author: Yi He) large-scale terrain reconstruction and deformation monitoring During the InSAR processing, two SAR complex images are multiplied by conjugation to obtain an InSAR interferometric phase image [4]. However, the inherent characteristics of SAR imaging systems inevitably introduce phase noise due to spatial and temporal decorrelations, atmospheric delay effect, thermal noise of the system, and others [5]- [7], resulting in inaccurate terrain or deformation inversion. InSAR interferometric phase filtering can suppress this phase noise and preserve phase details [8], thereby improving the accuracy of terrain reconstruction and deformation monitoring.…”

An Adaptive Particle Filter for Target Tracking Based on Double Space-Resampling

“…Over the past decades, particle filters have been applied with great success to a variety of state estimation problems [1][2][3], especially in nonlinear and non-Gaussian systems for which there is no analytical optimal solution [4][5]. In simple words, particle filter (PF) is based on Sequential Monte Carlo approach [6][7][8][9], which utilizes a large number of samples (particles) to represent the posterior probability distributions. The samples are propagated over time using a combination of sequential importance sampling and resampling steps [10][11].…”