Polarization-enhanced super-resolution imaging reconstruction based on a denoising diffusion probabilistic model.

Area of Science:

• Optics and Photonics
• Computer Vision
• Artificial Intelligence

Background:

• Super-resolution imaging is crucial for applications like target detection and material recognition.
• Existing methods struggle to fully leverage complementary information for enhanced reconstruction.

Purpose of the Study:

• To develop a novel super-resolution model that integrates polarization information.
• To improve the accuracy and performance of super-resolution image reconstruction.

Main Methods:

• Proposed a polarization-enhanced super-resolution diffusion model (PSRDM).
• Introduced a double-branch feature extraction module (DBFEM) to fuse intensity and degree of linear polarization (DoLP) image features.
• Employed a diffusion and denoising network for high-resolution image reconstruction.

Main Results:

• The PSRDM effectively extracts and fuses low-frequency and high-frequency features.
• Experimental results show superior performance compared to state-of-the-art super-resolution methods.
• Demonstrated the model's capability in reconstructing high-resolution images from low-resolution inputs.

Conclusions:

• The proposed PSRDM significantly enhances super-resolution imaging by incorporating polarization data.
• This approach offers a promising direction for advanced image reconstruction in various scientific and technical fields.