Towards Next-Generation Smart Seed Phenomics: A Review and Roadmap for Metasurface-Based Hyperspectral Imaging and a Light-Field Platform for 3D Reconstruction

Jingrui Yang, Qinglei Zhao, Shuai Liu, Jing Guo, Fengwei Guan, Shuxin Wang, Qinglong Hu, Q. C. Liu, Qi Song, Mingdong Zhu, Chao LiPhotonics 2026 · DOI

TL;DR

This review proposes a revolutionary imaging platform for seed phenomics by integrating metasurface optics with light-field and hyperspectral imaging. This approach promises to capture detailed 3D structural and biochemical information of seeds in a single shot, overcoming the limitations of current bulky and complex systems.

What Problem Does This Paper Address?

Understanding seed germination and traits is crucial for global food security. Current seed phenotyping methods struggle with scalability, efficiency, and capturing the full complexity of seed characteristics. Traditional hyperspectral and light-field imaging systems are often bulky, have limited depth resolution, and lack seamless spectral-spatial integration. This paper addresses the need for a compact, high-throughput, and information-rich imaging solution for advanced seed phenomics, aiming to bridge the gap between genotype and observable traits for improved crop breeding.

Methods

The paper reviews and proposes a novel approach that leverages metasurfaces – ultrathin optical components with subwavelength nanostructures – to achieve unprecedented control over light. Specifically, it focuses on:

  • Metasurface-based Light-Field Imaging: Using metasurface lens arrays (MLAs) to capture both spatial and angular information of light rays in a single exposure. This enables 3D reconstruction without the need for multiple cameras or complex synchronization.
  • Metasurface-based Hyperspectral Imaging: Employing dispersive metasurfaces to precisely control how different wavelengths of light are focused. This allows for spectral encoding and separation, enabling the acquisition of hyperspectral data.
  • Computational Imaging and Reconstruction: Discussing advanced algorithms, including geometric methods and deep learning (e.g., physics-informed neural networks, Transformers), to decode the complex optical information captured by metasurface systems and reconstruct high-fidelity 3D hyperspectral data.
  • End-to-End Co-Design: Proposing a framework where the metasurface hardware and the reconstruction algorithms are jointly optimized, allowing the system to learn optimal optical encoding and decoding strategies.

Datasets

Not explicitly detailed as the paper is a review and roadmap. However, it discusses the need for large datasets for training deep learning models and proposes strategies for generating synthetic data using physics-driven models and domain adaptation techniques for real-world scenarios.

Key Insights and Ideas

  • Metasurfaces as a Game-Changer: Metasurfaces offer a pathway to miniaturize complex optical systems, integrating functionalities like dispersion control and light-field modulation into ultra-compact, flat optics.
  • Single-Shot 5D Imaging: The ultimate goal is a platform that captures spatial (x, y), angular (θ, φ), and spectral (λ) information simultaneously, providing a complete "phenotypic fingerprint" of a seed.
  • Physics-Informed AI: Integrating physical constraints of metasurface optics into deep learning models is crucial for robust and accurate reconstruction, especially for complex, ill-posed inverse problems.
  • Bridging the "Sim-to-Real" Gap: Generating realistic synthetic data and using domain adaptation techniques are key to training models that perform well on real-world biological samples, where ground truth data is scarce.

Key Results

As this is a review and roadmap, specific quantitative results are not presented. However, the paper highlights:

  • Demonstrated Capabilities: Existing research showcases metasurface-based systems achieving high-fidelity 3D reconstruction, broadband spectral imaging, and compact optical designs.
  • Potential for Seed Phenomics: The proposed integrated platform has the potential to significantly improve the throughput and depth of analysis for seed traits, including morphology, germination, and biochemical composition.
  • Roadmap for Implementation: The paper provides a structured plan for developing and validating such a system, addressing key technical challenges.

Limitations and Future Work

  • Optical Efficiency: A significant challenge for large-area metasurfaces is maintaining high transmission efficiency across broad spectral bands, which impacts signal-to-noise ratio.
  • Computational Complexity: Reconstructing high-dimensional 5D data is computationally intensive, requiring advanced algorithms and powerful hardware.
  • "Sim-to-Real" Gap: Generating accurate ground-truth data for complex biological samples remains a hurdle, necessitating robust synthetic data generation and domain adaptation strategies.
  • Future Work: The authors emphasize the need for continued research in advanced inverse design, novel materials, physics-informed neural networks, and efficient self-supervision strategies to overcome these limitations and realize the proposed platform. The focus will be on achieving higher optical efficiency, developing more sophisticated reconstruction algorithms, and enabling real-time processing for practical applications.

Summary updated

Abstract

Seed phenomics is a critical research field for understanding seed germination mechanisms. Metasurfaces, composed of subwavelength nanostructures, offer a promising pathway to achieve both dispersion control and imaging functionalities within an ultra-compact form factor. Recent advances in micro–nano-optics and computational imaging have opened new avenues for high-dimensional, multimodal imaging. However, conventional hyperspectral and light-field systems still face limitations in compactness, depth resolution, and spectral–spatial integration. This review summarizes recent progress in metalens and metasurface lens array-based light-field systems for hyperspectral imaging and 3D reconstruction, with a focus on the underlying principles, design strategies, and reconstruction algorithms that enable single-shot 3D hyperspectral acquisition. We further present a forward-looking roadmap toward the realization of a revolutionized imaging paradigm: a metasurface-based light-field platform that fully integrates 3D and hyperspectral imaging capabilities. In particular, we examine how dispersive metasurfaces serve as core optical elements for precise dispersion control in hyperspectral imaging systems, while metalens arrays enable accurate modulation of spatial–angular distributions in light-field configurations. We systematically review both 3D and spectral reconstruction algorithms, highlighting their roles in decoding complex optical encodings. The application of these integrated systems in seed phenotyping is emphasized, demonstrating their capability to capture 3D spatial–spectral distributions in a single exposure. This approach facilitates high-throughput analysis of morphological traits, germination potential, and internal biochemical composition, offering a comprehensive solution for advanced seed characterization. Finally, we outline a practical roadmap for implementing a metasurface-based light-field platform that integrates hyperspectral imaging and computational 3D reconstruction. This review offers a comprehensive overview of the state of the art in compact 3D light-field systems and multimodal hyperspectral imaging platforms, while providing forward-looking insights aimed at advancing smart seed phenotyping, precision agriculture, and next-generation optical imaging technologies.
BibTeX
@inproceedings{yang2026towards,
  author = {Jingrui Yang and Qinglei Zhao and Shuai Liu and Jing Guo and Fengwei Guan and Shuxin Wang and Qinglong Hu and Q. C. Liu and Qi Song and Mingdong Zhu and Chao Li},
  title = {{Towards Next-Generation Smart Seed Phenomics: A Review and Roadmap for Metasurface-Based Hyperspectral Imaging and a Light-Field Platform for 3D Reconstruction}},
  booktitle = {Photonics},
  year = {2026},
  doi = {10.3390/photonics13010061},
}
Figure