计算光学成像在散射中的应用

自然界中普遍存在光散射现象。如何通过散射介质实现高分辨率成像是光学成像领域亟待解决的重要问题。在早期研究中，多重光散射被认为是雾霾、云层、生物组织等复杂介质成像中的障碍。然而，最近研究表明，散射并不是成像的基本限制:光子在经过多次散射后仍然包含了大量信息。为了深入了解新兴的计算光学成像是如何解决多重光散射问题的，文中主要介绍了波前整形、散斑相关及深度学习等方法在散射成像领域中的研究进展。最新的研究成果表明:波前整形可以实现动态散射介质内部的高分辨率快速聚焦；散斑相关能够利用单帧散斑实现非侵入式成像；基于深度学习的成像技术能恢复出隐藏在光学厚度为13.4的白色聚苯乙烯平板背后的物体。

Application of computational optical imaging in scattering

Light scattering is a common phenomenon in nature. How to realize high resolution imaging through turbid media is an important problem to be solved urgently in the field of optical imaging. In early studies, multiple light scattering has been regarded as a barrier in imaging through haze, cloud, biological tissue and other complex media. However, recent studies have shown that scattering is not the basic limitation of imaging:photons still contain a lot of information after multiple scattering. In order to provide insight into how new computational optical techniques can address the issues of multiple light scattering, the recent progress of scattering imaging method based on wavefront shaping, speckle correlation and deep learning was summarized. The latest research shows that, wavefront shaping technology can achieve fast optical focusing inside dynamic scattering medium with high resolution; speckle correlation method can realize non-invasive imaging by single-shot speckle pattern; deep learning is able to recover the object hidden behind the white polystyrene plate with optical thickness of 13.4.