Learning to Model and Calibrate Optics via a Differentiable Wave Optics Simulator
This method addresses the challenge of designing and calibrating microscopes more efficiently, offering a novel approach for optical engineering.
The authors tackled the problem of calibrating real optical setups by introducing a differentiable wave optics simulator composed of trainable modules, enabling direct parameter fitting from data and achieving improved reconstruction results for various optical elements.
We present a novel learning-based method to build a differentiable computational model of a real fluorescence microscope. Our model can be used to calibrate a real optical setup directly from data samples and to engineer point spread functions by specifying the desired input-output data. This approach is poised to drastically improve the design of microscopes, because the parameters of current models of optical setups cannot be easily fit to real data. Inspired by the recent progress in deep learning, our solution is to build a differentiable wave optics simulator as a composition of trainable modules, each computing light wave-front (WF) propagation due to a specific optical element. We call our differentiable modules WaveBlocks and show reconstruction results in the case of lenses, wave propagation in air, camera sensors and diffractive elements (e.g., phase-masks).