Displacement-agnostic coherent imaging through scatter with an interpretable deep neural network
Displacement-agnostic coherent imaging through scatter with an interpretable deep neural network
复制标题
DOI：
10.1364/oe.411291
复制
发表时间：
2021-01-18
期刊：
影响因子：
3.8
通讯作者：
Tian, Lei
中科院分区：
文献类型：
--
作者：
Li, Yunzhe;Cheng, Shiyi;Tian, Lei
通过散射进行连贯的成像是一项具有挑战性的任务。已经探索了基于模型的和数据驱动的方法来解决反散射问题。在我们以前的工作中，我们已经表明，深度学习方法可以通过看不见的扩散器来做出高质量且高度可推广的预测。在这里，我们提出了一个新的深神经网络模型，该模型对更广泛的扰动不可知，包括散射器变化，位移和系统发达，最高可达10倍。此外，我们开发了一个新的分析框架来解释我们深度学习模型的机制，并基于无监督的降低技术来可视化其普遍性。我们表明，我们的模型可以解开特定于散射的信息，并提取特定于对象的信息并在不同的散射条件下实现概括。我们的作品为通过散射媒体采用了强大而可解释的深度学习方法铺平了道路。 （c）2021美国光学学会根据OSA开放访问发布协议条款
Coherent imaging through scatter is a challenging task. Both model-based and data-driven approaches have been explored to solve the inverse scattering problem. In our previous work, we have shown that a deep learning approach can make high-quality and highly generalizable predictions through unseen diffusers. Here, we propose a new deep neural network model that is agnostic to a broader class of perturbations including scatterer change, displacements, and system detbcus up to 10x depth of field. In addition, we develop a new analysis framework for interpreting the mechanism of our deep learning model and visualizing its generalizability based on an unsupervised dimension reduction technique. We show that our model can unmix the scattering-specific information and extract the object-specific information and achieve generalization under different scattering conditions. Our work paves the way to a robust and interpretable deep learning approach to imaging through scattering media. (C) 2021 Optical Society of America under the terms of the OSA Open Access Publishing Agreement