Three-Dimensional Microwave Imaging: Fast and Accurate Computations with Block Resolution Algorithms

Area of Science:

• Electromagnetics and Applied Physics
• Computational Imaging
• Signal Processing

Background:

• Microwave imaging reconstruction involves computationally intensive calculations.
• Solving multiple high-dimensional linear systems is required for each cost function evaluation.

Purpose of the Study:

• To develop a more efficient computational strategy for microwave imaging reconstruction.
• To reduce the time complexity of gradient-based optimization algorithms in this field.

Main Methods:

• Implementation of a block inversion strategy utilizing the block-biconjugate gradient stabilized (BiCGStab) algorithm.
• Development of efficient implementations tailored for the microwave imaging context.
• Application of additive penalization and gradient-based optimization.

Main Results:

• Achieved computational time savings of up to a factor of two compared to standard sequential BiCGStab.
• Demonstrated enhanced performance for challenging reconstruction problems, including high-frequency illuminations and highly contrasted objects.
• Obtained satisfactory imaging estimates for objects in the Fresnel database.

Conclusions:

• The proposed block inversion strategy offers a significant computational advantage for microwave imaging.
• This method is particularly effective for complex and high-contrast imaging scenarios.
• The approach provides accurate reconstruction results for challenging objects.