Combining ADMM DIP WTV and Deconvolution Algorithm for Enhancing Microwave Radiometer Image Resolution

Microwave radiometers are indispensable instruments for global environmental monitoring. However, constraints imposed by imaging environments, storage, and transmission bandwidth make it challenging to acquire high-spatial-resolution imagery. While traditional analytical deconvolution methods enhance spatial resolution, they often come with noise amplification and Gibbs fluctuations. To enhance the spatial resolution of microwave radiometer data while improving data integrity, this study proposes a novel approach. Building upon traditional algorithm-repaired images, it employs a deep image prior and alternates between solving the optimization network parameters and the total variation (TV) regularization term using Alternating Direction Method of Multipliers (ADMM) for denoising. This method significantly enhances data integrity and achieves resolution enhancement. Experiments using simulated and real Microwave Radiometer Imaging (MWRI) data demonstrate the robustness and effectiveness of this approach.