Approximate Observation Weighted ℓ_2/3 SAR Imaging under Compressed Sensing.

Area of Science:

• Remote Sensing
• Signal Processing
• Computational Imaging

Background:

• Compressed Sensing SAR Imaging relies on accurate observation matrices, leading to high resource consumption with larger scenes.
• Existing approximate observation models using ℓq-norm (q=1, 1/2) regularization struggle with insufficient sparsity and imaging detail.

Purpose of the Study:

• To develop a Compressed Sensing SAR Imaging method that addresses the limitations of accurate observation models and existing approximate methods.
• To enhance sparsity and improve imaging detail in SAR imaging through novel regularization techniques.

Main Methods:

• An approximate observation operator based on the Chirp Scaling Algorithm was employed to replace the precise observation model.
• A weighted ℓ2/3-norm regularization was applied, aligning with natural image gradient distributions.
• A weighted matrix was used to further constrain the regularization, balancing sparsity and detail.

Main Results:

• The proposed method demonstrates enhanced sparsity compared to traditional ℓq-norm regularization.
• The weighted ℓ2/3-norm regularization effectively balances detail insufficiency issues.
• Experimental results confirm the excellent performance of the developed SAR imaging method.

Conclusions:

• The weighted ℓ2/3-norm regularization SAR imaging method based on approximate observation offers superior performance.
• This approach effectively overcomes the limitations of existing methods in terms of sparsity and imaging detail.
• The findings suggest a promising direction for resource-efficient and high-fidelity SAR imaging.