Interferometric SAR Phase Denoising Using Proximity-Based K-SVD Technique.

Area of Science:

• Geosciences
• Remote Sensing
• Signal Processing

Background:

• Interferometric noise degrades the quality of Synthetic Aperture Radar (SAR) interferograms.
• Accurate phase information is crucial for various SAR applications, including deformation monitoring and topographic mapping.

Purpose of the Study:

• To develop and validate a novel noise reduction technique for SAR interferometry.
• To effectively extract phase image content using sparse and redundant representations over a trained dictionary.

Main Methods:

• Implementation of the Proximity-based K-SVD (ProK-SVD) algorithm for dictionary training on interferometric data.
• Comparison of different training dictionaries (data-extracted, random, DCT) on simulated data.
• Application of the strategy to real interferometric data from ERS/ENVISAT, ALOS, and COSMO-SkyMed SAR sensors.

Main Results:

• Significant noise reduction observed in the fringe patterns of both simulated and real interferometric phase images.
• Preservation of useful phase information without considerable loss.
• Demonstrated effectiveness across various SAR sensor resolutions and bands.

Conclusions:

• The proposed ProK-SVD based sparse representation method is effective for interferometric noise reduction in SAR.
• The approach offers a valuable tool for improving the quality and reliability of SAR interferometric products.
• Successful application across diverse SAR datasets validates the robustness of the method.