The code and data from other publications that do not have a GitHub repository may be available upon request at nichen.opticsATgmail.com.
Differentiable imaging introduces novel methodologies to bridge the gap between physical and numerical models, enabling joint optimization of optical hardware and computational algorithms. This approach facilitates end-to-end co-design of imaging systems.
Learn more: Differentiable Imaging Project
Physics-aware computing develops computational reconstruction algorithms that integrate physical priors and domain knowledge to enhance imaging quality and robustness. By leveraging understanding of the underlying physical processes, these methods achieve superior performance compared to purely data-driven approaches.
Uncertainty-aware computing advances algorithmic frameworks to identify, quantify, and compensate for system uncertainties such as misalignment, aberrations, and calibration errors. These methods enable robust imaging even under imperfect operating conditions.