Efficient InSAR phase noise reduction via total variation regularization

Abstract

We consider the phase noise filtering problem for Interferometric Synthetic Aperture Radar (InSAR) using a total variation regularized complex linear least squares formulation. Although the original formulation is convex, solving it directly with the standard CVX package is time consuming due to the large problem size. In this paper, we introduce the effective and efficient alternating direction method of multipliers (ADMM) to solve the equivalent well-defined complex formulation for the real and imaginary parts of the optimization variables. Both the iteration complexity and the computational complexity of the ADMM are established in the forms of theorems for our InSAR phase noise problem. Simulation results based on simulated and measured data show that this new InSAR phase noise reduction method not only is 3 orders of magnitude faster than the standard CVX solver, but also has a much better performance than the several existing phase filtering methods.

创新点

• 1. 提出一种基于复的线性二乘最小化项加上一个全差分正则项的凸贝叶斯解决方案。

• 2. 由于凸优化问题规模大, 仅仅采用标准的CVX软件包求解该模型非常耗时。为了有效 且高效地进行求解, 文本给出等效的相位滤波模型并引入ADMM算法求解该等效模型。

• 3. 为了分析用于求解该模型的ADMM算法的性能, 本文建立了其迭代复杂性和计算复杂 性。