Efficient hybrid aberration spatial-calibrated 3D network for liquid-lens-based diffractive spectral imaging

Area of Science:

• Optics and Photonics
• Computational Imaging
• Artificial Intelligence in Imaging

Background:

• Liquid-lens diffractive spectral imaging offers miniaturization potential but faces image degradation from aberrations and spatial variations.
• Existing methods struggle to effectively correct these combined degradations, limiting performance.

Purpose of the Study:

• To develop a highly accurate and efficient computational method for reconstructing high-quality spectral images from liquid-lens-based diffractive systems.
• To address image degradation issues caused by aberrations and spatial variations in miniaturized spectral imaging.

Main Methods:

• Proposed a hybrid aberration spatial-calibrated 3D (HASC3D) network, integrating a novel calibration module with a 3D convolution-enhanced U-Net.
• Developed a custom imaging system using a diffractive lens and a tunable liquid lens for nine-channel multispectral reconstruction (510-590 nm, 10 nm resolution).

Main Results:

• The HASC3D network significantly improved spectral image reconstruction quality compared to conventional algorithms.
• Achieved a peak signal-to-noise ratio (PSNR) of 24.67 dB and a structural similarity index (SSIM) of 0.67 for reconstructed images.
• Demonstrated effective correction of aberrations and spatial variations, leading to superior imaging performance.

Conclusions:

• The HASC3D algorithm offers a promising solution for high-performance spectral image reconstruction in liquid-lens-based systems.
• This advancement paves the way for compact, portable, and high-quality spectral imaging technologies.
• The proposed method shows strong potential for various applications requiring miniaturized spectral imaging.