Deep learning is progressively emerging as a vital tool for image reconstruction in light field microscopy. The present review provides a comprehensive examination of the latest advancements in light field image reconstruction techniques based on deep learning algorithms. First, the review briefly introduced the concept of light field and deep learning techniques. Following that, the application of deep learning in light field image reconstruction was discussed. Subsequently, we classified deep learning-based light field microscopy reconstruction algorithms into three types based on the contribution of deep learning, including fully deep learning-based method, deep learning enhanced raw light field image with numerical inversion volumetric reconstruction, and numerical inversion volumetric reconstruction with deep learning enhanced resolution, and comprehensively analyzed the features of each approach. Finally, we discussed several challenges, including deep neural approaches for increasing the accuracy of light field microscopy to predict temporal information, methods for obtaining light field training data, strategies for data enhancement using existing data, and the interpretability of deep neural networks.

Light-field imaging has wide applications in various domains, including microscale life science imaging, mesoscale neuroimaging, and macroscale fluid dynamics imaging. The development of deep learning-based reconstruction methods has greatly facilitated high-resolution light-field image processing, however, current deep learning-based light-field reconstruction methods have predominantly concentrated on the microscale. Considering the multiscale imaging capacity of light-field technique, a network that can work over variant scales of light-field image reconstruction will significantly benefit the development of volumetric imaging. Unfortunately, to our knowledge, no one has reported a universal high-resolution light-field image reconstruction algorithm that is compatible with microscale, mesoscale, and macroscale. To fill this gap, we present a real-time and universal network (RTU-Net) to reconstruct high-resolution light-field images at any scale. RTU-Net, as the first network that works over multiscale light-field image reconstruction, employs an adaptive loss function based on generative adversarial theory and consequently exhibits strong generalization capability. We comprehensively assessed the performance of RTU-Net through the reconstruction of multiscale light-field images, including microscale tubulin and mitochondrion dataset, mesoscale synthetic mouse neuro dataset, and macroscale light-field particle imaging velocimetry dataset. The results indicated that RTU-Net has achieved real-time and high-resolution light-field image reconstruction for volume sizes ranging from 300 μm × 300 μm × 12 μm to 25 mm × 25 mm × 25 mm, and demonstrated higher resolution when compared with recently reported light-field reconstruction networks. The high-resolution, strong robustness, high efficiency, and especially the general applicability of RTU-Net will significantly deepen our insight into high-resolution and volumetric imaging.

To uncover the underlying fluid mechanisms, it is crucial to explore imaging techniques for high-resolution and large-scale three-dimensional (3D) measurements of the flow field. Plenoptic background-oriented schlieren (Plenoptic BOS), an emerging volumetric method in recent years, has demonstrated being able to resolve volumetric flow dynamics with a single plenoptic camera. The focus-stack-based plenoptic BOS system can qualitatively infer the position of the density gradient in 3D space based on the relative sharpness of the refocused BOS image. Plenoptic BOS systems based on tomography or specular enhancement techniques are realized for use in high-fidelity 3D flow measurements due to the increased number of acquisition views. Here, we first review the fundamentals of plenoptic BOS, and then discuss the system configuration and typical application of single-view and multi-view plenoptic BOS. We also discuss the related challenges and outlook on the potential development of plenoptic BOS in the future.

High-resolution and three-dimensional measurements at large scales represent a crucial frontier in flow diagnostics. Color-encoded illumination particle imaging velocimetry has emerged as a promising non-contact volumetric measurement technique in recent years. By employing chromatic gradient illumination to excite tracer particles, this method encodes depth information into color signatures, which are then correlated with two-dimensional positional data in images to reconstruct three-dimensional flow fields using a single camera. This review first introduces the fundamental principles of particle image velocimetry/particle tracking velocimetry and chromatic-depth encoding. Subsequently, we categorize color-depth-encoded particle velocimetry methods based on different illumination strategies, including LED-based, projector-based, and laser-based systems, discussing their respective configurations and representative applications. Finally, we summarize the current research progress in color-encoded particle image velocimetry techniques, provide a comparative analysis of their advantages and limitations, and discuss existing challenges along with future development prospects.