Optically coherent image formation and denoising using a plug and play inversion framework

Area of Science:

• Optics
• Image Processing
• Signal Processing

Background:

• Optically coherent imaging systems face performance limitations due to measurement and speckle noise.
• Speckle noise in coherent imaging is signal-dependent and follows an exponential distribution, posing challenges for traditional noise reduction methods.

Purpose of the Study:

• To develop an image formation framework for maximum a posteriori (MAP) estimation in coherent imaging.
• To enable the use of Gaussian denoising algorithms (GDAs) for mitigating noise in coherent imaging without modification.
• To compare the performance of various GDAs using simulated and experimental data.

Main Methods:

• Development of a novel image formation framework based on MAP estimation.
• Application of unmodified Gaussian denoising algorithms (GDAs) to address noise in coherent imaging.
• Comparative analysis of different GDAs using both simulated and experimental coherent imaging data.

Main Results:

• The proposed framework successfully mitigates exponentially distributed and signal-dependent noise.
• Gaussian denoising algorithms (GDAs) can be directly applied within the new framework.
• The framework demonstrates robustness to noise and significantly reduces reconstruction error compared to standard inversion methods.

Conclusions:

• The developed image formation framework enhances the performance of optically coherent imaging systems.
• The framework provides a robust and effective method for noise reduction in coherent imaging.
• This approach offers a significant improvement in image reconstruction accuracy for coherent imaging applications.