AI 中文总结
该研究针对传统光谱仪的权衡问题，通过共同设计色散莫尔超光学与递归采样算法，利用相位包裹色差实现高分辨率光谱扫描，引入自适应采样协议加速采集，以实验室原型验证，建立了智能任务自适应超光学传感器框架。
AI 中文摘要
光谱学是材料表征、化学传感和天文学的基础，但传统仪器在占地面积、光谱范围和分辨率之间面临严格权衡。我们展示了一种内容自适应光谱仪，通过将色散莫尔超光学与递归采样算法共同设计来克服这一问题。我们利用亚波长超表面结构中相位包裹产生的强色差，将其相互旋转角与聚焦波长进行确定性一一映射，用作高分辨率光谱扫描仪。为加速数据采集，引入内容自适应递归采样协议，利用光谱结构稀疏性。通过实验室原型在405 - 980nm范围验证，能以更少测量重建多种光谱，比传统均匀采样快6.7倍，建立了智能任务自适应超光学传感器框架。
英文摘要
Optical spectroscopy underpins material characterization, chemical sensing, and astronomy, but conventional instruments face a rigid trade-off between footprint, spectral range, and resolution. We demonstrate a content-adaptive spectrometer that overcomes this by co-designing dispersive Moiré meta-optics with a recursive sampling algorithm. Instead of using Moiré metalenses solely for varifocal tuning, we harness the strong chromatic aberration arising from phase-wrapping in their subwavelength metasurface architecture. This hyperchromaticity enables a deterministic, one-to-one mapping between the metasurfaces' mutual rotation angle and the sharply focused wavelength, repurposing the pair as a high-resolution spectral scanner. To accelerate data acquisition, we introduce a content-adaptive recursive sampling protocol that exploits the structural sparsity of physical spectra: a fast coarse sweep identifies high-information regions, followed by successively finer angular refinement only where needed. Using a laboratory prototype spanning 405-980 nm, we reconstruct diverse spectra -- from smooth broadband to sparse multi-line laser emissions -- with nearly 3x fewer measurements on average at matched fidelity (up to 7x for sparse line spectra), achieving 30 dB reconstruction 6.7x faster than conventional uniform sampling. This establishes a framework for intelligent, task-adaptive meta-optical sensors that tightly integrate physical dispersion with computational signal processing for real-time spectrometry.
