Polarization-informed deep learning for 3D integral imaging restoration in turbidity.

Area of Science:

• Optics and Photonics
• Computer Vision
• Machine Learning

Background:

• Polarimetric imaging captures light polarization information, crucial for understanding material properties and environmental conditions.
• Degradations like scattering, attenuation, and occlusion in turbid media severely impair polarimetric image quality.
• Recovering accurate polarimetric data (Stokes parameters, degree of linear polarization) is essential for reliable analysis.

Purpose of the Study:

• To develop and evaluate a novel polarimetric image restoration approach using polarization-informed deep learning and 3D integral imaging.
• To recover degraded Stokes parameters and the degree of linear polarization.
• To estimate the Mueller matrix for characterizing transmission media and objects.

Main Methods:

• Utilized an unsupervised image-to-image translation (UNIT) framework for Stokes parameter restoration.
• Employed a multi-output convolutional neural network (CNN) for Mueller matrix estimation.
• Integrated 3D integral imaging to mitigate degradations in turbid media.
• Evaluated performance under varying turbidity and partial occlusion conditions.

Main Results:

• The proposed method successfully recovered Stokes parameters and the degree of linear polarization from degraded images.
• Mueller matrix estimation provided insights into transmission media and object characteristics.
• 3D integral imaging demonstrated effectiveness in reducing turbidity-induced degradations.
• Experimental results confirmed the approach's promise under diverse environmental degradations.

Conclusions:

• The developed polarization-informed deep learning approach shows significant promise for polarimetric image restoration in degraded environments.
• This work represents the first application of polarization-informed deep learning within 3D imaging for recovering polarimetric information and Mueller matrix estimates.
• The method offers a robust solution for applications requiring accurate polarimetric analysis in challenging conditions.