Microlens Array-Based Wavefront and Intensity Profile Sensing with Physical Constraint Learning

The wavefront and intensity profiles of laser beams are crucial in many applications, including micromachining. This letter proposes a microlens array (MLA)-based methodology, similar to a Shack-Hartmann wavefront sensor (SHWS), to simultaneously measure the wavefront and intensity profiles of an incoming laser beam using a physical constraint learning approach. First, a pretrained network based on MLA simulation datasets is developed and then refined through transfer learning. To create a practical MLA-based measurement system, this pretrained network is further trained with datasets modulated by a spatial light modulator. Experimental results show that the proposed network can reconstruct both the wavefront and intensity profiles in real time. Compared to conventional SHWS methods, this approach improves computation speed by over 100 times and increases wavefront sensing accuracy by about 5 times.