DeepFilters：基于学习数字滤波重构的散射感知瞳构工程技术
光学
2026-05-14 v1 计算机视觉与模式识别
摘要
Extended depth of field显微镜通过工程化点扩散函数(PST)对轴向信息进行单次采集，但传统和深度光学方法在散射组织中受到降解。我们引入DeepFilters，一个散射感知的深度光学框架，通过校准的可微前向模型联合优化参数化瞳滤波器和数字滤波重构网络，以实现广泛的泛化而无需重新训练。融合经验散射核、物理引导的正则化和混合遗传-梯度初始化策略，DeepFilters将PST从16微米扩展到>400微米的clear介质，并实现生物组织>120微米深处的信号恢复，通过固定脑切片和海urchin胚胎验证。
引用
@article{arxiv.2605.13619,
title = {DeepFilters: Scattering-Aware Pupil Engineering with Learned Digital Filter Reconstruction for Extended Depth of Field Microscopy},
author = {Joseph L. Greene and Suet YIng Chan and Qilin Deng and Jeffrey Alido and Alexandra Lion and Guorong Hu and Ruipeng Guo and Tongyu Li and Kivilcim Kiliç and Ian Davison and Lei Tian},
journal= {arXiv preprint arXiv:2605.13619},
year = {2026}
}
备注
38 pages (18 main text, 20 supplement), 23 Figures (7 main text, 16 supplement)