Optical imaging through scattering media is important in a variety of fields ranging from microscopy to autonomous vehicles. Although advanced wavefront shaping techniques have offered several breakthroughs in the past decade, current techniques still require a known guide star and a high-resolution spatial light modulator or a very large number of measurements and are limited in their correction field of view. Here we introduce a guide-star-free, non-invasive approach that can correct more than 190,000 scattered modes using only 25 incoherently compounded, holographically measured, scattered light fields, obtained under unknown random illuminations. This is achieved by computationally emulating an image-guided wavefront shaping experiment, where several virtual spatial light modulators are simultaneously optimized to maximize the reconstructed image quality. Our method shifts the burden from the physical hardware to a digital, naturally parallelizable computational optimization, leveraging state-of-the-art automatic differentiation tools. We demonstrate the flexibility and generality of this framework by applying it to imaging through various complex samples and imaging modalities, including epi-illumination, anisoplanatic multi-conjugate correction of highly scattering layers, lensless endoscopy in multicore fibres and acousto-optic tomography. The presented approach offers high versatility, effectiveness and generality for fast, non-invasive imaging in diverse applications.

通过散射介质进行光学成像在从显微镜到自动驾驶汽车的各个领域都非常重要。尽管先进的波前整形技术在过去十年中取得了多项突破，但当前的技术仍然需要已知的导星和高分辨率空间光调制器或大量的测量，并且其校正视野有限。在这里，我们介绍了一种无导星、非侵入性的方法，该方法只需使用在未知随机照明下获得的 25 个不相干复合、全息测量的散射光场即可校正超过 190,000 个散射模式。这是通过计算模拟图像引导波前整形实验来实现的，其中多个虚拟空间光调制器同时优化，以最大限度地提高重建的图像质量。我们的方法利用最先进的自动微分工具，将负担从物理硬件转移到数字化、自然可并行的计算优化。我们通过将其应用于各种复杂样本和成像模式的成像来证明该框架的灵活性和通用性，包括落射照明、高散射层的各平面多共轭校正、多芯光纤中的无晶状体内窥镜检查和声光断层扫描。所提出的方法为各种应用中的快速、无创成像提供了高度的通用性、有效性和通用性。