全息术
计算机科学
光场
微尺度化学
计算
光学
调制(音乐)
领域(数学)
波前
物理
人工智能
声学
算法
数学
纯数学
数学教育
作者
Jin Wu,Xinkuo Li,Ke Sun,Kai Gao,Chenduan Chen,Jianrong Qiu,Dezhi Tan
标识
DOI:10.1002/lpor.202401742
摘要
Abstract Phase‐only computer‐generated holography (CGH) is an effective technique to manipulate 3D light field distribution in the tight focusing volume for numerous applications in micro/nano‐manufacturing, optical tweezers, and optical communication. Unfortunately, hologram computation is slow and generally takes several seconds or longer for a single instance, which hinders broad applications in real time light modulation. Here, fast hologram computation is reported with the calculation time for a single instance down to 3.7 ms. A depth‐adaptive 3D tight‐focusing holographic network framework driven by a vectorial diffraction model is developed. The network adequately considers the tight‐focusing property and the spherical aberration effect in high NA objectives and employs a layer‐based learning strategy to reinforce the global constraints on reconstructed 3D focusing fields. This network enables the generation of high‐quality holographic phases in real time and facilitates large‐scale computations of focused fields with arbitrary spatial, intensity, and axial spacing distributions with high speed and high accuracy (up to 0.93). The proposed network is deployed in ultrafast laser direct writing and microscale fluorescence display applications, which indicates that the current 3D tight‐focusing field modulation technique will play a vital role in broad optical and photonic engineering.
科研通智能强力驱动
Strongly Powered by AbleSci AI