Logarithmic Laplacian Prior Based Bayesian Inverse Synthetic Aperture Radar Imaging

Area of Science:

• Signal Processing
• Radar Imaging
• Computational Imaging

Background:

• Inverse Synthetic Aperture Radar (ISAR) imaging is crucial for target identification.
• Traditional ISAR algorithms face challenges with sparse data and phase errors.
• Sparse representation techniques offer potential for improved ISAR performance.

Purpose of the Study:

• To develop a novel ISAR imaging algorithm utilizing a new sparse prior.
• To enhance the focusing and autofocusing capabilities of ISAR.
• To improve resolution and noise suppression in ISAR images.

Main Methods:

• Implementation of a novel logarithmic Laplacian prior within a Bayesian framework.
• Joint estimation of phase errors using the minimum entropy criterion for autofocusing.
• Utilization of Maximum A Posteriori (MAP) and Maximum Likelihood Estimation (MLE) for parameter estimation.
• Application of Fast Fourier Transform (FFT) and Hadamard product for computational efficiency.

Main Results:

• The proposed logarithmic Laplacian prior demonstrates superior sparse representation compared to the standard Laplacian prior.
• The algorithm achieves better-focused ISAR images through effective autofocusing.
• Experimental results show significant improvements in resolution and noise suppression over traditional sparse ISAR methods.
• Parameter estimation is automated, avoiding manual tuning.

Conclusions:

• The novel ISAR algorithm based on the logarithmic Laplacian prior offers superior performance.
• The method effectively addresses challenges in sparse representation and phase error correction in ISAR imaging.
• The algorithm provides a computationally efficient and robust solution for enhanced ISAR imaging.