Abstract:Lensless imaging has gained popularity in various applications due to its userfriendly nature, cost-effectiveness, and compact design. However, achieving high-quality image reconstruction within this framework remains a significant challenge. Lensless imaging measurements are associated with distinct point spread functions (PSFs), resulting in many PSFs introducing artifacts into the underlying physical model. This discrepancy between the actual and prior models poses challenges for standard reconstruction methods to effectively address high-quality image reconstruction by solving a regularization-based inverse problem. To overcome these issues, we propose MN-FISTA-Net, an unrolled neural network that unfolds the fast iterative shrinkage/thresholding algorithm for solving mixed norm regularization with a deep denoiser prior. Our method enhances mask-based lensless imaging performance by efficiently addressing noise and model mismatch, as evidenced by significant improvements in image quality compared to existing approaches. (c) 2024 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement

Robust Unrolled Network for Lensless Imaging with Enhanced Resistance to Model Mismatch and Noise