【摘要】 光谱成像技术能够同时获取目标的空间形态与光谱特征，在生物医学、环境监测及工业检测等领域具有重要应用价值。然而，传统扫描式光谱成像在时间分辨率与系统体积上存在局限，难以满足实时与便携化应用的需求。近年来，兴起的计算光谱成像技术通过光学编码与计算重构的有机结合，打破了传统成像在物理采样上的限制。系统综述了可调谐型计算光谱成像技术的研究进展，重点分析了基于不同可调谐滤波器的光谱成像系统，包括液晶可调谐滤波器、法布里-珀罗谐振腔、声光可调谐滤波器以及新兴的超表面与先进光学材料器件。文中详细阐述了计算光谱成像技术的原理与数学模型，并介绍了基于不同种类的可调谐滤波器的光学编码机制及成像系统，对比了其在光通量、光谱分辨率、系统紧凑性及成像速度等方面的性能表现。最后，总结了当前技术面临的问题和挑战，并对计算光谱成像技术的未来发展趋势及研究方向进行了展望。更多还原

【Abstract】 Significance Humans mainly rely on visual information such as shape, color, and brightness when observing the objective world. Traditional RGB color imaging systems can only simulate the three broad-band responses of the human eye, inevitably losing the rich spectral details contained in the objects. In fact, the spectral characteristics of substances are like their unique "fingerprints", containing key information such as chemical composition, physical structure, and physiological state. Spectral imaging technology perfectly solves this problem. Based on two-dimensional spatial imaging, it adds a spectral dimension to generate a three-dimensional data cube. Each spatial pixel corresponds to a continuous spectral curve. However, traditional spectral imaging technology is constrained by limitations in spatial resolution, spectral resolution, temporal resolution, and light transmission efficiency, and thus has certain limitations, making it difficult to meet the requirements of real-time and portable applications. To break through these limitations, tunable computational spectral imaging technology based on computational optics theory has emerged, continuously expanding the performance boundaries of spectral imaging technology and has become an emerging technical route.Progress Tunable filtering technology is a type of spectral selection technology that can dynamically change its central wavelength or transmission spectral characteristics under the influence of an external control signal. It can adjust the physical parameters of the filter through electrical control, magnetic control, or thermal control, and achieve selective transmission or suppression of different spectral information without mechanical movement,thereby enabling flexible acquisition of spectral information. Applying tunable filters to computational spectral imaging technology allows for equivalent regulation of the system measurement matrix by changing the working state of the tunable device, enabling flexible and efficient spectral sampling without the need for complex mechanical scanning or fixed spatial encoding structures, as shown in Fig.2. Compared to traditional spectral imaging systems and computational spectral imaging methods relying on static encoding, tunable computational spectral imaging has significant advantages in terms of system compactness, spectral acquisition flexibility, and imaging efficiency, and is particularly suitable for real-time imaging and resource-constrained application scenarios.The tunable filter can adjust the light of different wavelengths, providing a more flexible way for spectral acquisition in computational spectral imaging technology. The liquid crystal tunable filter(LCTF), as a common tunable filter, is widely used in agriculture and biomedical imaging. By adjusting the electro-optic birefringence property of the liquid crystal molecules, LCTF can precisely regulate the transmitted wavelength and selectively collect spectral information of different bands. The Fabry-Pérot(FP) filter, with its high spectral resolution and good light transmission efficiency, has significant advantages in fine spectral imaging and array integration. The FP filter can achieve continuous wavelength tuning by adjusting the optical length of the cavity,and its high spectral resolution makes it have unique advantages in fine spectral analysis. However, the application of the FP filter still faces limitations in tuning speed and spectral response range, especially in dynamic scenarios, where there are certain limitations. The acoustic-optic tunable filter(AOTF) modulates the light wave by applying a radio frequency signal, featuring fast tuning speed, no mechanical components, and high spectral resolution. In snapshot spectral imaging in dynamic scenarios, AOTF provides an important technical path. AOTF is widely used in environmental monitoring, biomedicine, and military fields, becoming an important technical tool. The sub-wavelength surface tunable filter, with its unique advantages in optical encoding through sub-wavelength structures, has also gained more applications in computational spectral imaging. The subwavelength surface can precisely control the phase, amplitude, and polarization of light by designing different geometric structures of nano-elements, thereby regulating spectral information and providing a new technical approach for computational spectral imaging. In addition, with the rapid development of materials science and micro-nano processing technology, a series of new functional materials with electrical, acoustic, optical, or nonlinear tunable properties have been introduced into computational spectral imaging systems. Their optical responses can be continuously, reversibly, and rapidly regulated under external field excitation and are suitable for integration into on-chip or compact optical systems.Conclusions and Prospects By introducing tunable filtering devices, active modulation of the incident light in the spectral dimension is achieved, and the high-dimensional spectral data cube is recovered from limited measurements through computational reconstruction algorithms. This encoding and decoding mode of computational spectral imaging technology effectively breaks through the inherent trade-off relationship between spatial, spectral and temporal resolutions in traditional spectral imaging, and has obvious advantages in system compactness, imaging efficiency and reliability. It has become one of the important development directions in the field of computational spectral imaging. In the future, with the development of microelectronics technology and nanofabrication processes, tunable filters will evolve towards faster tuning speed, higher light transmission efficiency and stronger integration capabilities, providing strong support for the realization of portable and efficient spectral imaging systems. At the same time, the collaborative optimization of physical models and algorithms will become the key to technological development. Image reconstruction based on deep learning and perception task-driven computational spectral imaging methods will become important research directions. With the continuous maturation of technology, computational spectral imaging systems based on tunable filters are expected to be more widely promoted and applied in various industries.更多还原