Minimalist infrared computational imaging with double-sided diffractive single-lens system.

The hybrid-order monolithic imaging diffractive (HMID) lens was designed to address the issue of negative dispersion of diffractive optical elements (DOEs) in infrared minimalist optical systems. However, the imaging quality is negatively impacted by processing errors and aberrations of the HMID when it is expanded for large aperture use. The initial goal of minimalist imaging is contradicted by traditional optical techniques like adding additional lenses to eliminate aberrations, so we proposed a joint optical-algorithmic processing method in this paper to address the above issues. Specifically, the method is divided into two parts: (1) Upgraded HMID (UHMID) is designed at the optical end to improve the image quality through Strehl ratio (SR) enhancement, which is realized by fine-tuning the height of the microstructure; (2) A denoiser based on Bayesian theory as well as a deep neural network structure, using the optical system point spread function (PSF) data to solve the recover image. The experimental results show that the resolution of the optical system after image restoration is improved from 21.13 p/mm to 31.33 lp/mm, representing a 48.27% increase. The principle proposed in this paper provides a framework for the use of monolithic imaging systems in large aperture optical systems.