Upgrading fiber-optic spectrometers to hyperspectral imagers

Upgrading fiber-optic spectrometers to hyperspectral imagers | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Upgrading fiber-optic spectrometers to hyperspectral imagers Xueli Chen, Zhong Ji, Yujin Liu, Jingyang Xing, Hanyan Zhao, hongling wan, and 2 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7333525/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Traditional fiber-optic spectrometers, while featuring compact designs and high sensitivity, are constrained to single-point spectral acquisition with limited spatial resolution, thereby restricting their broder applications. Although existing commercial spectral imaging systems can provide three-dimensional spectral cubes, they often require complex optical configurations, resulting in high costs and operational complexity. To overcome these limitations, this paper proposes a fiber-optic hyperspectral single-pixel imager (FHSPI) based on single-pixel computational imaging technology. In addition, by integrating a fiber-optic probe inspired by the compound eyes of insects, the FHSPI significantly expands the imaging field of view (FOV) to 130 degrees, representing a substantial increase from the 10-degree FOV of a single fiber. The FHSPI we proposed can achieve spectral image data cubes with a resolution of 1 nm, and by developing a spectral flux integration method, it enables high-quality spectral imaging under low-light conditions. Operating at a sub-0.1% sampling rate (0.09%), the FHSPI enables rapid distinguishing between authentic from artificial green leaves with a theoretical spectral identification speed of up to 1250 spectra per second (sp/s). We present a novel approach for developing cost-effective, high-performance fiber-optic spectral imaging technology that harnesses the potential advantages of FHSPI in addressing the data redundancy issues faced by conventional hyperspectral imagers. This advancement holds significant potential for fostering innovations in various applications, including precision agriculture, environmental monitoring, and biomedical diagnostics. Physical sciences/Optics and photonics/Optical techniques/Optical spectroscopy Physical sciences/Optics and photonics/Optical techniques/Imaging and sensing Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Introduction Fiber-optic spectrometers are widely recognized for their compact design, high sensitivity, and ability to deliver rapid and precise spectral analysis across a broad wavelength range. These advantages have made them indispensable in fields such as chemical analysis [ 1 , 2 ] , in-situ environmental monitoring [ 3 , 4 ] , and label-free biomedical diagnostics [ 5 , 6 ] . However, a significant drawback of fiber-optic spectrometers is their inability to capture spatial image information, as they are limited to providing spectral data from a single point or an integrated area. This limitation severely restricts their application in scenarios where spatial context is critical, such as remote sensing [ 7 , 8 ] , material characterization [ 9 , 10 ] , and biological imaging [ 11 , 12 ] . Without spatial resolution, the interpretation of data from heterogeneous samples or dynamic processes becomes incomplete, potentially leading to inaccurate conclusions. In contrast, commercial spectral imaging systems overcome the dimensionality limitation of fiber-optic spectrometers by capturing three-dimensional data (two spatial dimensions and one spectral dimension), enabling spatially resolved spectral analysis [ 13 – 15 ] . These systems typically employ methodologies such as push-broom scanning [ 16 – 19 ] , point-scanning [ 20 , 21 ] , or snapshot imaging [ 22 – 24 ] . For instance, push-broom systems utilize a line-scan approach combined with a moving platform to build spatial-spectral data cubes, while snapshot systems rely on optical filters or coded apertures to simultaneously acquire spatial and spectral information. However, such implementations often require complex optical components (e.g., high-resolution detector arrays, precision scanning mechanisms, or specialized filters), which drastically increase manufacturing costs. Given the limitations of fiber-optic spectrometers and the high cost of commercial spectral imaging systems, developing a low-cost solution to transform fiber-optic spectrometers into a spectral imager holds profound significance. Scanning-based methods, such as motorized stage [ 25 , 26 ] and galvanometer mirrors [ 27 ] , have been developed to upgrade fiber-optic spectrometers for spatial-spectral imaging. However, these approaches face challenges such as slow acquisition, mechanical complexity, and high costs, limiting their practicality. In recent decades, single-pixel computational imaging [ 28 , 29 ] emerges as a promising alternative to enhance fiber-optic spectrometers into spectral imaging systems. By leveraging compressed sensing and computational algorithms [ 30 , 31 ] , single-pixel imaging eliminates the need for complex scanning mechanisms or expensive detector arrays, significantly reducing system cost and complexity [ 32 , 33 ] . Notably, advanced single-pixel hyperspectral imaging systems have demonstrated exceptional capabilities, such as compressive hyperspectral video imaging systems based on single-pixel detectors [ 34 ] , compressed single-pixel hyperspectral imaging systems using RGB sensors [ 35 ] , Fourier single-pixel hyperspectral imaging systems [ 36 ] , hyperspectral imaging systems achieving spectral multiplexing through digital micromirror devices (DMD) [ 37 ] , rapid hyperspectral single-pixel imaging systems based on frequency-division multiplexed illumination [ 38 ] , quantum dot-enabled infrared hyperspectral imaging technologies [ 39 ] , and deep learning-based highly compressed single-pixel macroscopic fluorescence lifetime imaging techniques [ 40 ] . However, our approach specifically focuses on enhancing existing fiber-optic spectrometer architectures. This approach uses a single-pixel detector combined with spatially modulated light patterns to reconstruct high-resolution images, offering advantages such as faster data acquisition, robustness to misalignment, and compatibility with existing fiber-optic spectrometer hardware. Therefore, by combining advanced single-pixel imaging technology with commercial fiber-optic spectrometers, it is expected to develop a cost-effective fiber-optic spectral imager. This paper integrates state-of-the-art single-pixel imaging technology based on Fourier orthogonal matrix patterns with a commercial fiber-optic spectrometer, developing a novel fiber-optic hyperspectral single-pixel imager (FHSPI). This system effectively acquires spectral hypercubes, enabling the retrieval of spectral information for each pixel in the image. In fact, the small aperture of fiber-optic spectrometers results in a limited field of view (FOV) for the FHSPI [ 41 ] . To address this, inspired by the wide FOV of insect compound eyes [ 42 – 44 ] , we developed a compound eye-like FHSPI, effectively increasing the imaging FOV area by nearly 164.8 times. This compound-eye optic fiber probe not only significantly enhances the imaging field of view but also improves the system's light utilization efficiency. Using compound eye-like FHSPI, we successfully acquired three-dimensional hypercube data of a color chart and resolved the spectral curves of each color patch. Additionally, leveraging the tunable spectral resolution of fiber-optic spectrometers, we can enhance the effective light throughput by adjusting the spectral bandwidth, significantly improving the imaging quality for specific spectral bands. More importantly, we found that the FHSPI can distinguish between real and artificial leaves at an ultra-low sampling rate, which significantly enhances the system's imaging speed and enables target spectral identification while reducing data redundancy exponentially. Our proposed FHSPI system holds great promise for advancing high-speed, low-cost spectral imaging applications, with potential breakthroughs in fields such as precision agriculture, environmental monitoring, and biomedical diagnostics, where rapid and accurate spectral analysis is critical. Results As a mature commercial product, fiber-optic spectrometers offer significant advantages, including compact size, high sensitivity, and the ability to perform real-time, cost-effective spectral measurements across a wide wavelength range. Their working principle, illustrated in Fig. 1 a, involves transmitting light through optical fibers to a diffraction grating, which disperses the light into its spectral components for detection by a linear array detector, enabling precise spectral analysis. Unfortunately, its capability to acquire only a single-point spectral curve at a time restricts its range of applications. Therefore, if it could be upgraded to a fiber-optic hyperspectral imager, it would significantly enhance its spectral information density (Fig. 1 b). Based on this concept, we propose a single-pixel computational imaging approach for fiber-based spectral imaging, as illustrated in Fig. 1 c. This method requires no modification to the existing fiber-optic spectrometer and achieves spatial-spectral information by simply encoding the illumination source using a structured light modulator (e.g., DMD or a projector). The system achieves three-dimensional spectral imaging by algorithmically reconstructing the temporal signals of structured light at different wavelengths captured by the fiber-optic spectrometer. The single-pixel imaging method assigns the task of acquiring spatial information to light source encoding and reconstruction algorithms, eliminating the need for detectors to have spatial resolution capabilities. This bypasses the highly challenging manufacturing processes required for spectral imaging detectors and offers a new possibility for developing low-cost hyperspectral imagers. The detailed architecture and imaging method principles of the FHSPI system are presented in Fig. S1 and S2. It's well known that the small aperture of a single optical fiber leads to a very limited FOV. As a result, the single-pixel fiber imaging system developed using a single optical fiber, as shown in Fig. 2 a, also faces the issue of a narrow FOV. Figure 2 b and 2 c show the spectrum and reconstruction images obtained using the single-fiber system at a sampling rate of 15.2%, respectively. These experimental results further confirm the issue of the narrow FOV. The imaging area has a diameter of 10 cm when the fiber is positioned 30 cm away from the scene. Based on geometric calculations, the imaging FOV angle ( θ ) of the single fiber system is determined to be 10° (Fig. 2 j), which matches the fiber's acceptance angle. To address the issue of limited FOV in single-fiber imaging systems, we took inspiration from the compound eyes of insects and designed a hemispherical fiber-optic port with a compound-eye structure. The probe was fabricated using high-precision 3D printing technology, featuring 16 multimode optical fibers (NA = 0.22) arranged in a concentric 1-6-9 configuration (1 central fiber, 6 middle fibers, and 9 outer fibers) within a 20-mm-diameter housing. This fiber-optic port consists of a 1-to-16 fiber optic interface and a hemispherical shell, with 16 fibers evenly distributed across the shell, as shown in Fig. 2 d. This design effectively expands the fiber's FOV. To address the issue of limited FOV in single-fiber imaging systems, we took inspiration from the compound eyes of insects and designed a hemispherical fiber-optic port with a compound-eye structure. This fiber-optic port consists of a 1-to-16 fiber optic interface and a hemispherical shell, with 16 fibers evenly distributed across the shell, as shown in Fig. 2 d. This design effectively expands the fiber’s FOV. Figures 2 e and 2 f show the spectrum and reconstruction results obtained by the compound-eye-type FHSPI system at a sampling rate of 15.2%, with the entire image being well captured. Figures 2 g and 2 h present the 3D mapping of the two images (Fig. 2 c, f), more clearly demonstrating the wide FOV advantage of the compound-eye system. Moreover, based on the compound-eye structure and the calculated field of view of a single fiber, we estimate the field of view angle (β) of the compound-eye system to be 130 degrees (Fig. 2 j). Therefore, it can be calculated that the theoretical imaging FOV area of the compound-eye system can reach 12941.9 square centimeters (Fig. 2 k), which is 164.8 times that of the single-fiber system. The photograph of the compound-eye fiber-optics has been presented in Fig. S3. To verify the spectral imaging and color spectral recognition capabilities of the FHSPI system, we performed imaging on a standard color card (Fig. 3 a). Figure 3 b shows the spectral-spatial data cube of the standard color card obtained using the FHSPI system, which records spectral images within the wavelength range of 450 nm to 650 nm, with a spectral resolution of 1 nm. This covers almost the entire visible light spectrum. The cross-sectional view of data cube clearly demonstrates that the intensity of different color blocks varies significantly with wavelength changes. To better illustrate the intensity differences of different color blocks at various wavelengths, we extracted 16 spectral images at equal intervals of 15 nm, as shown in Fig. 3 c. In addition, we extracted the spectral curves of each color block on the color card from the data cube, as shown in Fig. 3 d. These experiments demonstrate that the FHSPI system has a powerful capability for acquiring spectral information of scene. It can obtain the spectral curves of each point in a scene with just a single imaging process. This significantly enhances the spectral analysis capabilities of traditional fiber-optic spectrometers. The voltage curves (V-m curves) of the spectrometer at different wavelengths have been presented in Fig. S4, where m refers to the sequence of patterns. In the FHSPI system, in addition to the signal sensitivity of the spectral detector, the spectral throughput also significantly impacts imaging quality. For example, the spectral curve of the UHE-type light source (ELPLP96) we used, as shown in Fig. 4 a, reveals that the light signal intensity is significantly lower near the cutoff wavelengths around 420 nm and 680 nm compared to the mid-wavelength region. This results in noticeably poorer imaging quality in these spectral bands, as demonstrated in Fig. 4 b. Thanks to the ultra-high spectral resolution (less than 1 nm) of the fiber-optic spectrometer (AMOS LBTEK), the effective light throughput can be determined by spectral integration—summing the intensity values over progressively increased FWHM while maintaining central wavelength stability. As a result, we can perform spectral integration in cutoff wavelengths regions with weaker light intensity to effectively enhance the light throughput in specific wavelength ranges, thereby potentially improving imaging quality. We compared two spectral sampling methods for the spectral images in the 425 to 440 nm wavelength range. As shown in Fig. 4 c, the blue curve represents the uniform FWHM interval sampling method, while the gray curve corresponds to the increasing FWHM interval sampling method. Figure 4 d displays the spectral images obtained using the two sampling methods, revealing that image quality improves significantly as the FWHM increases. Taking the 430 nm spectral image as an example, when the FWHM increases from 1 nm to 10 nm, the signal-to-noise ratio (SNR) of the image rises from 10.8 to 26.5. The method for calculating the image signal-to-noise ratio (SNR) is presented in Fig. S5. Therefore, in practical applications where the focus is on acquiring spatial information, the system can significantly enhance image quality by increasing spectral throughput. Additionally, we compared the color image synthesis capabilities of the spectral sampling methods. Figure 4 f shows the original image used for imaging, while Figs. 4 g and 4 h demonstrate that the wide-spectrum overlapping sampling method achieves a slight improvement in color accuracy in the synthesized images. Notably, under conditions of spectral intensity imbalance, this improvement may become more pronounced. As demonstrated in Fig. 4 e, the adaptive sampling strategy of this method effectively compensates for intensity deficiencies in specific spectral bands, potentially reducing color distortion compared to conventional methods in controlled low-intensity scenarios. This characteristic suggests the method's particular value for applications involving non-uniform illumination conditions. To validate the performance of the FHSPI system in practical applications, we conducted imaging tests on real leaf and fake leaf (Figure. 5a), aiming to distinguish between them based on spectral differences. Figure 5 b presents the spectral images of real and fake leaves extracted using FHSPI. As shown, the spectral curves of real and fake leaves exhibit significant differences beyond 650 nm, which can be effectively utilized to distinguish between them. More importantly, we found that by selecting appropriate imaging wavelengths, here is 670 nm, even low-quality imaging results with low sampling rates can effectively distinguish between real and fake leaves, as shown in Fig. 5 c, d. Figure 5 e shows spectral mapping of real and fake leaves obtained by FHSPI with different sampling rates. This further indicates that the sampling rate has no impact on the extraction of spectral data for distinguishing real and fake leaves. This implies that, in practical applications, FHSPI does not require extensive time to acquire data-redundant hypercubes but instead focuses on obtaining useful key spectral information efficiently. For example, at a sampling rate of 0.09%, only 32 structured light patterns need to be projected to acquire the spectral curves of two leaves in the same scene. With a projection frequency of 50 Hz, achievable by common commercial projectors, the time required to distinguish between real and fake leaves is reduced to just 0.6 seconds. This significantly enhances the efficiency of traditional fiber-optic spectrometers, which typically require multiple sequential spectral acquisitions. Moreover, the response of the fiber-optic spectrometer can be extended to the near-infrared (NIR) spectral region. Therefore, by introducing a NIR light source, we further validated the near-infrared spectral imaging and image recognition capabilities of the proposed FHSPI system, as shown in Fig. 5 f. At 880 nm, the system achieved 58% recognition accuracy for real leaves and 83% for fake leaves at 25% sampling rate, while reaching 40% (real) and 89% (fake) at 0.09% sampling rate. The near-infrared spectral imaging results and recognition accuracy (880 nm) of real and fake leaves under different sampling rates have been presented in the Fig. S6 and S8. We believe that target recognition under low sampling rates is not limited to specific conditions but rather has a certain degree of general applicability. We have validated the effectiveness of our approach across multiple experiments involving different targets and imaging backgrounds. As shown in Fig. 5 , we present imaging and recognition results under different sampling rates. Figures 5 h and 5 j display the imaging results at varying sampling rates, demonstrating that although image quality degrades at low sampling rates, proper wavelength selection can still effectively distinguish color differences between objects. Specifically, in Fig. 5 k, we successfully discriminated between two visually similar color patches on the chromaticity chart at an extremely low sampling rate of 0.7%, and further achieved enhanced performance by differentiating two distinctly colored gourds at an ultra-low sampling rate of 0.3%. These performance variations primarily originate from differences in the number of spectral regions present in the target objects (Fig. 5 l: 2 spectral regions in real/fake leaves, 7 in the multicolor gourd sample, and 24 in the standard color chart). The system's minimum required sampling rate decreases proportionally with increasing sparsity of the target spectral regions. This fundamental property enables scenario-specific projection pattern optimization, permitting rapid spectral acquisition while intrinsically avoiding the data redundancy inherent to conventional hyperspectral imaging systems. Table. The performance comparison between our FHSPI and commercial spectrometers. Encoding Device Detector Spectral Band Channels Resolution Max Resolution FOV Expense/ RMB Spectral Acquisition Speed W/O FS 350–1000 nm 2048 <1 nm 1 10° ~ 13000 — Our work Projector FS 350–1000 nm 2048 < 2 nm 1024×1024 10° ~ 16000 3.33 sp/s CE-FHSPI 350–1000 nm 2048 <2 nm 1024×1024 130° ~ 16650 3.33 sp/s DMD CE-FHSPI 350–1000 nm 2048 <2 nm 768×768 130° ~ 76650 1250 sp/s Commercial W/O Hyperspectral camera1 400–1000 nm 224 <5.5 nm 1024×1024 38° ~ 450000 80 sp/s W/O Hyperspectral camera 2 330–800 nm 255 <2.8 nm 1500×1000 8° ~ 640000 280 sp/s Discussions In order to demonstrate the advantages of FHSPI, we compared the key parameters before and after the upgrade, as well as with those of several typical commercial hyperspectral cameras. Simultaneously, we have created a radar chart that highlights the performance differences between the FHSPI system and commercial spectrometers, which is presented in Fig. S9. We designed a fiber-optic probe inspired by the compound eye, which significantly extended the field of view of the original single-fiber system (FS) with a mere 10° field of view to 130°, greatly enhancing the system's field-of-view range. The cost-effectiveness of the compound eye-like FHSPI (CE-FHSPI) system totals 16,650 RMB, which includes a projector (CB-U05 EPSON) at 3,000 RMB, a spectrometer (AMOS LBTEK) at 13,000 RMB, and a 1×16 fiber bundle at 650 RMB, offers a substantial cost advantage over existing typical commercial spectral cameras (e.g., Pika UV Resonon), which cost up to 600,000 RMB. Although low sampling rates inevitably degrade spatial resolution, our results demonstrate that spectral resolution remains largely unaffected. We tested the spectral resolution of FHSPI at different sampling rates, all of which reached a level below 2 nm as shown in Fig. S7. To more accurately assess spectral recognition performance, we introduced a novel parameter—the spectral acquisition rate—defined as the number of spectra acquired per unit time and expressed in “spectra per second” (sp/s). Using a commercial projector (CB-U05 EPSON), we generated 128×128 pixel Fourier fringes at 50 Hz. This enabled us to distinguish between the spectra of two leaves every 0.6 seconds, yielding a spectral acquisition speed of 3.33 sp/s. We plan to upgrade the encoding device to a digital micromirror device (DMD, V-7001VIS, ViALUX), which is expected to increase the projection frequency to 20,000 Hz. As a result, the system would be capable of distinguishing between two spectra every 0.0016 seconds, with the spectral acquisition speed reaching 1250 spectra per second. The most distinctive capability of our FHSPI system is its simultaneous multi-region spectral acquisition with ultra-low sampling rates. Our experimental results demonstrate that the system's minimum required sampling rate decreases with increasing sparsity of the target spectral regions. This fundamental property enables adaptive optimization of projection patterns for specific imaging scenarios, allowing the system to achieve rapid multi-region spectral collection while inherently avoiding the data redundancy that plagues conventional hyperspectral cameras. These improvements would significantly enhance the system's spectral imaging capabilities and provide robust support for high-efficiency spectral analysis. While commercial spectral cameras offer high spatial resolution, they often produce redundant data that hinder subsequent processing. In contrast, the FHSPI system addresses these limitations by balancing cost, efficiency, and performance, making it a promising solution for various applications. In summary, this work successfully upgraded the traditional fiber spectrometer to a fiber hyperspectral imager (FHSPI) through single-pixel imaging technology, thereby addressing the limitations of conventional fiber spectrometers. With the aid of a compound-eye structure design, the FHSPI significantly enhanced the imaging FOV and light utilization efficiency, and enabled rapid differentiation of complex samples at low sampling rates. Experimental results demonstrated the system's high efficiency in spectral imaging and target recognition, particularly under low sampling rate conditions. As a single-pixel imaging system, FHSPI inherits the intrinsic advantages of strong resistance to environmental interference and motion artifacts, ensuring robust performance in practical applications. The low cost and high performance of FHSPI have opened new avenues for the widespread application of spectral imaging technology, and are expected to promote technological breakthroughs and innovations in many fields. It should be noted that as a structured illumination system, FHSPI is particularly suitable for endoscopic and industrial applications where active light control is feasible, though its applicability in passive remote sensing scenarios is inherently limited by the requirement for active illumination. Future work will focus on developing a structured detection-based hyperspectral imaging system to overcome this limitation and expand the methodology's applicability to broader scenarios. Experimental Section In this experiment, we used a commercial digital projector (CB-U05 EPSON) with a brightness of 3400 lm and a contrast ratio of 15000:1. The projector has a resolution of 1920×1200 pixels and is capable of generating high-contrast binary or grayscale patterns. Its light source is a 210-watt white LED, which projects patterns onto the target object via an optical system. In the experimental setup, the projector sequentially displayed 8192 Fourier fringes with a resolution of 128 × 128 pixels every 0.02 seconds. The pattern size was matched to the imaging area of the target object to ensure sufficient accumulation of light signals. We also employed a fiber-optic spectrometer (AMOS LBTEK) with a spectral range ranging from 350 nm to 1000which each fiber has a core diameter of 200 µm, a total length of 1 meter, and a numerical aperture (NA) of 0.22. In this experiment, the imaging targets are a 24-color standard color card (TCCR-S Mennon), seven printed bottle gourd samples of different colors, as well as real and artificial leaves. The color card is made of materials such as ABS and PVC. It features a rich distribution of colors and stable physical properties and is commonly used for color correction in imaging devices. In this study, we utilized it to evaluate the imaging system's color reproduction and detail resolution capabilities, which are critical for assessing the accuracy and performance of the system. In addition to the color card, we also conducted a comparative imaging analysis of real and artificial leaves. The real leaf samples are from pothos (Epipremnum aureum), whose leaves exhibit typical characteristics of green plants, making them an ideal natural specimen for the study. The artificial leaf samples, made primarily of plastic, effectively mimic the appearance and texture of real leaves. This comparison enables us to assess the system's ability to distinguish between natural and synthetic materials, further demonstrating its versatility and precision. Declarations Acknowledgements Authors acknowledge the financial supports from the National Natural Science Foundation of China (62405236, 62375210, 62275210, 62105123); the National Young Talent Program; the Shaanxi Young Top-notch Talent Program; Xidian University Specially Funded Project for Interdisciplinary Exploration (TZJH2024059, TZJH2024060); Fundamental Research Funds for Central Universities (QTZX24079). Notes The manuscript was written through the contributions of all authors. All authors have given approval to the final version of the manuscript. Competing interests The authors declare no competing financial interests. Data availability Data available on request from the authors. Supporting Information Supporting Information is available from the Wiley Online Library or from the author. References Cadeado, A. N. & Machado, C. C.&Silva, S. G. Simultaneous reflectance and fluorescence measurements for portable formaldehyde determination in milk using a multi-channel spectrometer sensor. Food Chem. 464, 141583 (2025). Chen, L. & Lin, Q. & Yuan, D.&Gong, Z. Simultaneous measurement of trace levels of Hg, as, Sb, and Bi in coastal seawater with a multichannel chemical vapor generation atomic fluorescence spectrometer. Anal. Chem. 95, 15621–15627 (2023). Coburn, S. & Alden, C. B. & Wright, R., et al. Regional trace-gas source attribution using a field-deployed dual frequency comb spectrometer. Optica 5, 320–327 (2018). Yang, Z. & Yu, X. & Dedman, S., et al. UAV remote sensing applications in marine monitoring: Knowledge visualization and review. Sci. Total Environ. 838, 155939 (2022). El-Mashtoly, S. F.&Gerwert, K. Diagnostics and therapy assessment using label-free Raman imaging. Anal. Chem. 94, 120–142 (2021). Huang, L. & Sun, H. & Sun, L., et al. Rapid, label-free histopathological diagnosis of liver cancer based on Raman spectroscopy and deep learning. Nat. Commun. 14, 48 (2023). Olmanson, L. G. & Brezonik, P. L.&Bauer, M. E. Airborne hyperspectral remote sensing to assess spatial distribution of water quality characteristics in large rivers: The Mississippi River and its tributaries in Minnesota. Remote Sens. Environ. 130, 254–265 (2013). Zhong, Y. & Wang, X. & Wang, S.&Zhang, L. Advances in spaceborne hyperspectral remote sensing in China. Geo-spatial Inf. Sci. 24, 95–120 (2021). Romani, A. & Clementi, C. & Miliani, C.&Favaro, G. Fluorescence spectroscopy: a powerful technique for the noninvasive characterization of artwork. Acc. Chem. Res. 43, 837–846 (2010). Zeng, Z. & Liu, Y.&Wei, J. Recent advances in surface-enhanced raman spectroscopy (SERS): Finite-difference time-domain (FDTD) method for SERS and sensing applications. TrAC, Trends Anal. Chem. 75, 162–173 (2016). Pilling, M.&Gardner, P. Fundamental developments in infrared spectroscopic imaging for biomedical applications. Chem. Soc. Rev. 45, 1935–1957 (2016). Zhou, X. & Chen, C. & Zuo, E., et al. Cross Branch Co-Attention Network multimodal models based on Raman and FTIR spectroscopy for diagnosis of multiple selected cancers. Appl. Soft Comput. 166, 112204 (2024). Knez, D. & Toulson, B. W. & Chen, A., et al. Spectral imaging at high definition and high speed in the mid-infrared. Sci. Adv. 8, eade4247 (2022). Ram, B. G. & Oduor, P. & Igathinathane, C., et al. A systematic review of hyperspectral imaging in precision agriculture: Analysis of its current state and future prospects. Comput. Electron. Agric. 222, 109037 (2024). Wang, K. & Pu, H.&Sun, D. W. Emerging spectroscopic and spectral imaging techniques for the rapid detection of microorganisms: An overview. Compr. Rev. Food Sci. Food Saf. 17, 256–273 (2018). Hu, B.-L. & Hao, S.-J. & Sun, D.-X.&Liu, Y.-N. A novel scene-based non-uniformity correction method for SWIR push-broom hyperspectral sensors. ISPRS J. Photogramm. Remote Sens. 131, 160–169 (2017). Jemec, J. & Pernuš, F. & Likar, B.&Bürmen, M. Deconvolution-based restoration of SWIR pushbroom imaging spectrometer images. Opt. Express 24, 24704–24718 (2016). Lv, S.&Liu, Y.-N. Anomaly detection in satellite-borne push-broom hyperspectral imagery based on joint low-rank tensor approximation. IEEE Trans. Geosci. Remote Sens. 62, 1–20 (2023). Pham, Q. T.&Liou, N.-S. The development of on-line surface defect detection system for jujubes based on hyperspectral images. Comput. Electron. Agric. 194, 106743 (2022). Sarić, R. & Nguyen, V. D. & Burge, T., et al. Applications of hyperspectral imaging in plant phenotyping. Trends Plant Sci. 27, 301–315 (2022). Yeh, K. & Sharma, I. & Falahkheirkhah, K., et al. Infrared spectroscopic laser scanning confocal microscopy for whole-slide chemical imaging. Nat. Commun. 14, 5215 (2023). Hua, X. & Wang, Y. & Wang, S., et al. Ultra-compact snapshot spectral light-field imaging. Nat. Commun. 13, 2732 (2022). McClung, A. & Samudrala, S. & Torfeh, M., et al. Snapshot spectral imaging with parallel metasystems. Sci. Adv. 6, eabc7646 (2020). Zheng, S. & Liu, Y. & Meng, Z., et al. Deep plug-and-play priors for spectral snapshot compressive imaging. Photonics Res. 9, B18-B29 (2021). Glaser, A. K. & Reder, N. P. & Chen, Y., et al. Multi-immersion open-top light-sheet microscope for high-throughput imaging of cleared tissues. Nat. Commun. 10, 2781 (2019). Wang, Y. & Peng, H. & Liu, K., et al. Multi-point scanning confocal Raman spectroscopy for accurate identification of microorganisms at the single-cell level. Talanta 254, 124112 (2023). Xintian, W. & Xuesong, M. & Xiaodong, W., et al. Motion position prediction and machining accuracy compensation of galvanometer scanner based on BWO-GRU model. Mech. Syst. Sig. Process. 210, 111081 (2024). Edgar, M. P. & Gibson, G. M.&Padgett, M. J. Principles and prospects for single-pixel imaging. Nat. Photonics 13, 13–20 (2019). Gibson, G. M. & Johnson, S. D.&Padgett, M. J. Single-pixel imaging 12 years on: a review. Opt. Express 28, 28190–28208 (2020). Zhang, Z. & Liu, S. & Peng, J., et al. Simultaneous spatial, spectral, and 3D compressive imaging via efficient Fourier single-pixel measurements. Optica 5, 315–319 (2018). Zhang, Z. & Ma, X.&Zhong, J. Single-pixel imaging by means of Fourier spectrum acquisition. Nat. Commun. 6, 6225 (2015). Liu, Y. & Ji, Z. & Cen, G., et al. Perovskite-based color camera inspired by human visual cells. Light Sci. Appl. 12, 43 (2023). Liu, Y. & Lin, D. & Xing, J., et al. Mode-switching single-pixel imaging via a high-performance perovskite-Si dual-mode photodetector. Photonics Res. 12, 2873–2880 (2024). Xu, Y. & Lu, L. & Saragadam, V.&Kelly, K. F. A compressive hyperspectral video imaging system using a single-pixel detector. Nat. Commun. 15, 1456 (2024). Tao, C. & Zhu, H. & Wang, X., et al. Compressive single-pixel hyperspectral imaging using RGB sensors. Opt. Express 29, 11207–11220 (2021). Jin, S. & Hui, W. & Wang, Y., et al. Hyperspectral imaging using the single-pixel Fourier transform technique. Sci. Rep. 7, 45209 (2017). Chen, W. & Sun, M.-J. & Deng, W.-J., et al. Hyperspectral imaging via a multiplexing digital micromirror device. Opt. Lasers Eng. 151, 106889 (2022). Jiang, X. & Li, Z. & Du, G., et al. Fast hyperspectral single-pixel imaging via frequency-division multiplexed illumination. Opt. Express 30, 25995–26005 (2022). Meng, H. & Gao, Y. & Wang, X., et al. Quantum dot-enabled infrared hyperspectral imaging with single-pixel detection. Light Sci. Appl. 13, 121 (2024). Ochoa, M. & Rudkouskaya, A. & Yao, R., et al. High compression deep learning based single-pixel hyperspectral macroscopic fluorescence lifetime imaging in vivo. Biomed. Opt. Express 11, 5401–5424 (2020). Tian, Y. & Ding, Z. & Feng, H., et al. Single-pixel imaging based on optical fibers. IEEE Photonics J. 12, 1–7 (2020). Ji, Z. & Liu, Y. & Zhao, C., et al. Perovskite Wide-Angle Field‐Of‐View Camera. Adv. Mater. 34, 2206957 (2022). Jiang, H. & Tsoi, C. C. & Yu, W., et al. Optical fibre based artificial compound eyes for direct static imaging and ultrafast motion detection. Light Sci. Appl. 13, 256 (2024). Ma, M. & Zhang, Y. & Deng, H., et al. Super-resolution and super-robust single-pixel superposition compound eye. Opt. Lasers Eng. 146, 106699 (2021). Additional Declarations There is no conflict of interest Supplementary Files SI.docx SI Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7333525","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":514119829,"identity":"feaafc16-5b22-4e02-90dd-cc1dea4f88af","order_by":0,"name":"Xueli Chen","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA1UlEQVRIiWNgGAWjYDACdgY2IGkDJBMbQHzGBoJamMFa0kjXchiIExiI02JwmPnZg487zif2sSc3fi5gsJHdcAAogk+LZDObueHMM7eN2XgeNkvPYEgz3nCAzdwAnxZ+Zh42ad6223JsEokN0jwMhxM3HOBhk8CnhQ2k5W/bOaCyxObfPAz/CWsB28LYdgBkSxvQlgOEtQD9YibZ25YM8kubNY9BsvHMw2xmeLUYHG9+JvGzzS5xfnv649s8FXayfSARfFrQTWAAxdQoGAWjYBSMAkoBABu7Pr5WpQnXAAAAAElFTkSuQmCC","orcid":"https://orcid.org/0000-0002-3898-9892","institution":"Xidian University","correspondingAuthor":true,"prefix":"","firstName":"Xueli","middleName":"","lastName":"Chen","suffix":""},{"id":514119830,"identity":"bd912fb8-4921-43e1-871f-e642a7bfae68","order_by":1,"name":"Zhong Ji","email":"","orcid":"","institution":"Jinan University","correspondingAuthor":false,"prefix":"","firstName":"Zhong","middleName":"","lastName":"Ji","suffix":""},{"id":514119831,"identity":"d04e6a4d-f967-4f81-aa18-3c9746a111f0","order_by":2,"name":"Yujin Liu","email":"","orcid":"","institution":"XIDIAN UNIVERSITY","correspondingAuthor":false,"prefix":"","firstName":"Yujin","middleName":"","lastName":"Liu","suffix":""},{"id":514119832,"identity":"0245c307-f73f-47d4-b21e-0763c16886b5","order_by":3,"name":"Jingyang Xing","email":"","orcid":"","institution":"Southern University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Jingyang","middleName":"","lastName":"Xing","suffix":""},{"id":514119833,"identity":"91b46841-6eeb-4720-b6a1-182d7c2611f9","order_by":4,"name":"Hanyan Zhao","email":"","orcid":"","institution":"Xidian University","correspondingAuthor":false,"prefix":"","firstName":"Hanyan","middleName":"","lastName":"Zhao","suffix":""},{"id":514119834,"identity":"727732a7-e0f8-42e3-a56e-ca4566972371","order_by":5,"name":"hongling wan","email":"","orcid":"","institution":"Xidian University","correspondingAuthor":false,"prefix":"","firstName":"hongling","middleName":"","lastName":"wan","suffix":""},{"id":514119835,"identity":"f40ad44b-76db-4b7e-a397-6f75b8bd1ec4","order_by":6,"name":"Hao Wang","email":"","orcid":"","institution":"Xidian University","correspondingAuthor":false,"prefix":"","firstName":"Hao","middleName":"","lastName":"Wang","suffix":""},{"id":514119836,"identity":"7392a4e2-06b2-4777-acb8-0cfc277a98f7","order_by":7,"name":"Yuncheng Yuncheng Xu","email":"","orcid":"","institution":"Xidian University","correspondingAuthor":false,"prefix":"","firstName":"Yuncheng","middleName":"Yuncheng","lastName":"Xu","suffix":""}],"badges":[],"createdAt":"2025-08-09 11:15:13","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7333525/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7333525/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":91930756,"identity":"5d0eb3ba-3967-4e43-83b1-4f2d9d901070","added_by":"auto","created_at":"2025-09-23 02:27:14","extension":"doc","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":5066240,"visible":true,"origin":"","legend":"","description":"","filename":"ManuscriptLSA20250745.doc","url":"https://assets-eu.researchsquare.com/files/rs-7333525/v1/4afc9b5342129c7ba05de1cc.doc"},{"id":91930752,"identity":"1c381fa2-3bff-41e7-96a0-f34c4609627d","added_by":"auto","created_at":"2025-09-23 02:27:13","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":9338,"visible":true,"origin":"","legend":"","description":"","filename":"LSA20251866.json","url":"https://assets-eu.researchsquare.com/files/rs-7333525/v1/4adddb16e632b01e9150e2f2.json"},{"id":91930753,"identity":"5cc74e64-dcec-43e9-86bf-921d7866c61e","added_by":"auto","created_at":"2025-09-23 02:27:13","extension":"docx","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":2177635,"visible":true,"origin":"","legend":"","description":"","filename":"SI.docx","url":"https://assets-eu.researchsquare.com/files/rs-7333525/v1/8644dc6e075ec200e0a8681a.docx"},{"id":91932407,"identity":"19660320-3572-4106-9126-66dd923c6a6d","added_by":"auto","created_at":"2025-09-23 02:35:14","extension":"xml","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":104117,"visible":true,"origin":"","legend":"","description":"","filename":"LSA202518660enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7333525/v1/61f4e082dbe45e3dc5900db6.xml"},{"id":91930764,"identity":"5d5e40fd-71f0-4620-b48f-fe987a21c9da","added_by":"auto","created_at":"2025-09-23 02:27:14","extension":"png","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":585286,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7333525/v1/d6af0527f1d4d2ac0e1d4b39.png"},{"id":91932405,"identity":"1fd5d8b0-07dd-4053-8e5b-20d1b6c1367e","added_by":"auto","created_at":"2025-09-23 02:35:14","extension":"png","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":600040,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7333525/v1/0e29af2d305cfcf418004f7d.png"},{"id":91932411,"identity":"bb4aa72a-7934-412a-a3bc-0ec41ec336a9","added_by":"auto","created_at":"2025-09-23 02:35:14","extension":"png","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":683106,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7333525/v1/067703d582c15927010f07a0.png"},{"id":91930773,"identity":"169dbc07-3203-4943-bd52-507fd3b3c0e6","added_by":"auto","created_at":"2025-09-23 02:27:14","extension":"png","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":2574762,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7333525/v1/c47e089e1e25a49559c894d5.png"},{"id":91930760,"identity":"2f27720c-e47a-48d7-b184-e990800b2a09","added_by":"auto","created_at":"2025-09-23 02:27:14","extension":"png","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":429692,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7333525/v1/3149bd2f74d8645f3abca9a5.png"},{"id":91930761,"identity":"e78875bd-63b4-44d4-a524-656581b8817a","added_by":"auto","created_at":"2025-09-23 02:27:14","extension":"png","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":105122,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7333525/v1/c262e34329d0bc77bebbf79b.png"},{"id":91930770,"identity":"aea8427d-5d8c-4ed5-aea9-90a151fa171c","added_by":"auto","created_at":"2025-09-23 02:27:14","extension":"png","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":124936,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7333525/v1/72e2628d2c18481dba076a6d.png"},{"id":91932408,"identity":"c96bbcec-6394-42ec-8b54-3e3aa68aa789","added_by":"auto","created_at":"2025-09-23 02:35:14","extension":"png","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":64025,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7333525/v1/a36f70ccc88df8772ef15aee.png"},{"id":91930768,"identity":"f165afe5-f2a5-4e6e-a47b-e15e96ab4a7e","added_by":"auto","created_at":"2025-09-23 02:27:14","extension":"png","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":316081,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7333525/v1/a6dd933a1b76a17db877dce2.png"},{"id":91932412,"identity":"e090a9b8-6081-4162-87e0-899b5b7cbd3c","added_by":"auto","created_at":"2025-09-23 02:35:14","extension":"png","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":66134,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7333525/v1/6cb5289c83e3a2586b1781c3.png"},{"id":91932410,"identity":"882e632f-7770-4894-a762-70e7c6899880","added_by":"auto","created_at":"2025-09-23 02:35:14","extension":"xml","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":103003,"visible":true,"origin":"","legend":"","description":"","filename":"LSA202518660structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7333525/v1/c3a024b6a39e3221629c4f21.xml"},{"id":91930766,"identity":"9affeffe-1eb8-4905-9a09-061c21c6f54b","added_by":"auto","created_at":"2025-09-23 02:27:14","extension":"html","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":111866,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7333525/v1/0b806fac321829b536b3de0b.html"},{"id":91932404,"identity":"ad2f56b1-dc0f-44ea-8169-73a41d91f78c","added_by":"auto","created_at":"2025-09-23 02:35:13","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":318500,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSingle-pixel hyperspectral imaging based on fiber-optic spectrometer\u003c/strong\u003e (a) Traditional fiber-optic spectrometers can only obtain a single point spectrum. (b) Concept of a fiber-optic hyperspectral imager capable of acquiring spectral images data cube. (c) The architecture and principle of the fiber-optic hyperspectral camera via the single-pixel imaging method.\u003c/p\u003e","description":"","filename":"image1.png","url":"https://assets-eu.researchsquare.com/files/rs-7333525/v1/d06b4e2653d08e9561e27c5d.png"},{"id":91930757,"identity":"a26d44de-2753-4983-9f36-4c11525eacd5","added_by":"auto","created_at":"2025-09-23 02:27:14","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":329883,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCompound eye-like fiber-optic probe improves imaging\u003c/strong\u003e \u003cstrong\u003eFOV \u003c/strong\u003e(a-c) FOV diagram (a), Fourier spatial spectra (c) and imaging results (c) of the single fiber-optics SPI. (d-f) FOV diagram (d), Fourier spatial spectra (e) and imaging results (f) of the compound eye fiber-optics SPI. (g, h) 3D Mapping of c, f shows the FOV area. (i) Normalized grayscale of line in c, f. (j) Geometric diagram of the FOV of a, d. (k) Theoretical FOV area estimation of compound eye fiber-optic single-pixel imaging system.\u003c/p\u003e","description":"","filename":"image2.png","url":"https://assets-eu.researchsquare.com/files/rs-7333525/v1/c7f550fe275cda71b349aa95.png"},{"id":91930758,"identity":"d9ae8a0c-e46c-4519-ab63-5be3b2952d16","added_by":"auto","created_at":"2025-09-23 02:27:14","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":644445,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSpectral imaging demonstration of FHSPI\u003c/strong\u003e (a) Photograph of a standard color card for spectral imaging. (b) Spectral data cube that demonstrates 200 spectral images, Δλ=1 nm. (c) Spectral images from 450 to 675 nm (15 nm interval) showing the spectral specificity of the color block. (d) Spectral curves of each color block on the color card extracted from the data cube.\u003c/p\u003e","description":"","filename":"image3.png","url":"https://assets-eu.researchsquare.com/files/rs-7333525/v1/93f55ddd74e8128525f3b102.png"},{"id":91932406,"identity":"91bba36c-70f7-46c4-96a7-147391288495","added_by":"auto","created_at":"2025-09-23 02:35:14","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":913070,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eEffect of spectral throughput on imaging quality \u003c/strong\u003e(a) The spectrum curve of the white light source used in this experiment. (b) Spectral imaging results (FWHM=1 nm), highlighting the poor imaging quality in wavelength (WL) with weaker light intensity. (c) Two spectral sampling methods, one is a constant FWHM of 1 nm, and the other is an increasing FWHM (2 nm increases to 10 nm). (d) Spectral images corresponding to the two spectral sampling methods. (e) SNR of spectral images. (f) Photograph of imaging object. (g, h) RGB images obtained by two sampling methods and their fused color images.\u003c/p\u003e","description":"","filename":"image4.png","url":"https://assets-eu.researchsquare.com/files/rs-7333525/v1/51cf7b1076902ed2b21d2fae.png"},{"id":91932409,"identity":"8c0c9783-36aa-4386-a707-d738191770b5","added_by":"auto","created_at":"2025-09-23 02:35:14","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":412584,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSpectral imaging and analysis at different sampling rates based on FHSPI \u003c/strong\u003e(a) Photograph of real and fake leaves. (b) Spectral curves of real and fake leaves. (c, d) Image results at 580 nm (c) and 670 nm (d) under different sampling rates. (e) Spectral mapping of real and fake leaves obtained by FHSPI with different sampling rates. (f) Near-infrared spectral imaging (880 nm) of real and fake leaves.\u003cstrong\u003e \u003c/strong\u003e(g) Photograph of a standard color card for spectral imaging with 24 spectral region numbers. (h) Spectral plots of standard color chart at 25% and 0.7% sampling rates (600 nm) (i) Photograph of a multicolor gourd sample for spectral imagingwith 7 spectral region numbers. (j) Spectral plots of multicolor gourd at 25% and 0.7% sampling rates (500 nm) (k) Spectral curves from standard color chart (25%, 0.7% sampling) and multicolor gourd (25%, 0.3% sampling) (l) Relationship between the number of spectral regions and required projection patterns.\u003c/p\u003e","description":"","filename":"image5.png","url":"https://assets-eu.researchsquare.com/files/rs-7333525/v1/48854b6553453f09e3fd9946.png"},{"id":94012955,"identity":"47b32825-ebe8-43da-b094-d7983717bd65","added_by":"auto","created_at":"2025-10-21 10:28:42","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":3338022,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7333525/v1/abd940d9-3064-4dfa-883d-650b4123c350.pdf"},{"id":91932403,"identity":"a1f0b513-0f71-4634-a27f-7b4148aeceee","added_by":"auto","created_at":"2025-09-23 02:35:13","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":2177635,"visible":true,"origin":"","legend":"SI","description":"","filename":"SI.docx","url":"https://assets-eu.researchsquare.com/files/rs-7333525/v1/426db7741348db21f60c27ea.docx"}],"financialInterests":"There is no conflict of interest","formattedTitle":"Upgrading fiber-optic spectrometers to hyperspectral imagers","fulltext":[{"header":"Introduction","content":"\u003cp\u003eFiber-optic spectrometers are widely recognized for their compact design, high sensitivity, and ability to deliver rapid and precise spectral analysis across a broad wavelength range. These advantages have made them indispensable in fields such as chemical analysis\u003csup\u003e[\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]\u003c/sup\u003e, \u003cem\u003ein-situ\u003c/em\u003e environmental monitoring\u003csup\u003e[\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]\u003c/sup\u003e, and label-free biomedical diagnostics\u003csup\u003e[\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]\u003c/sup\u003e. However, a significant drawback of fiber-optic spectrometers is their inability to capture spatial image information, as they are limited to providing spectral data from a single point or an integrated area. This limitation severely restricts their application in scenarios where spatial context is critical, such as remote sensing\u003csup\u003e[\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]\u003c/sup\u003e, material characterization\u003csup\u003e[\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]\u003c/sup\u003e, and biological imaging\u003csup\u003e[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]\u003c/sup\u003e. Without spatial resolution, the interpretation of data from heterogeneous samples or dynamic processes becomes incomplete, potentially leading to inaccurate conclusions.\u003c/p\u003e\u003cp\u003eIn contrast, commercial spectral imaging systems overcome the dimensionality limitation of fiber-optic spectrometers by capturing three-dimensional data (two spatial dimensions and one spectral dimension), enabling spatially resolved spectral analysis\u003csup\u003e[\u003cspan additionalcitationids=\"CR14\" citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]\u003c/sup\u003e. These systems typically employ methodologies such as push-broom scanning\u003csup\u003e[\u003cspan additionalcitationids=\"CR17 CR18\" citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]\u003c/sup\u003e, point-scanning\u003csup\u003e[\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]\u003c/sup\u003e, or snapshot imaging\u003csup\u003e[\u003cspan additionalcitationids=\"CR23\" citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]\u003c/sup\u003e. For instance, push-broom systems utilize a line-scan approach combined with a moving platform to build spatial-spectral data cubes, while snapshot systems rely on optical filters or coded apertures to simultaneously acquire spatial and spectral information. However, such implementations often require complex optical components (e.g., high-resolution detector arrays, precision scanning mechanisms, or specialized filters), which drastically increase manufacturing costs. Given the limitations of fiber-optic spectrometers and the high cost of commercial spectral imaging systems, developing a low-cost solution to transform fiber-optic spectrometers into a spectral imager holds profound significance.\u003c/p\u003e\u003cp\u003eScanning-based methods, such as motorized stage\u003csup\u003e[\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e]\u003c/sup\u003e and galvanometer mirrors\u003csup\u003e[\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e]\u003c/sup\u003e, have been developed to upgrade fiber-optic spectrometers for spatial-spectral imaging. However, these approaches face challenges such as slow acquisition, mechanical complexity, and high costs, limiting their practicality. In recent decades, single-pixel computational imaging\u003csup\u003e[\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e, \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e]\u003c/sup\u003e emerges as a promising alternative to enhance fiber-optic spectrometers into spectral imaging systems. By leveraging compressed sensing and computational algorithms\u003csup\u003e[\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e, \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e]\u003c/sup\u003e, single-pixel imaging eliminates the need for complex scanning mechanisms or expensive detector arrays, significantly reducing system cost and complexity\u003csup\u003e[\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e, \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e]\u003c/sup\u003e. Notably, advanced single-pixel hyperspectral imaging systems have demonstrated exceptional capabilities, such as compressive hyperspectral video imaging systems based on single-pixel detectors\u003csup\u003e[\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e]\u003c/sup\u003e, compressed single-pixel hyperspectral imaging systems using RGB sensors\u003csup\u003e[\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e]\u003c/sup\u003e, Fourier single-pixel hyperspectral imaging systems\u003csup\u003e[\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e]\u003c/sup\u003e, hyperspectral imaging systems achieving spectral multiplexing through digital micromirror devices (DMD)\u003csup\u003e[\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e]\u003c/sup\u003e, rapid hyperspectral single-pixel imaging systems based on frequency-division multiplexed illumination\u003csup\u003e[\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e]\u003c/sup\u003e, quantum dot-enabled infrared hyperspectral imaging technologies\u003csup\u003e[\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e]\u003c/sup\u003e, and deep learning-based highly compressed single-pixel macroscopic fluorescence lifetime imaging techniques\u003csup\u003e[\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e]\u003c/sup\u003e. However, our approach specifically focuses on enhancing existing fiber-optic spectrometer architectures. This approach uses a single-pixel detector combined with spatially modulated light patterns to reconstruct high-resolution images, offering advantages such as faster data acquisition, robustness to misalignment, and compatibility with existing fiber-optic spectrometer hardware. Therefore, by combining advanced single-pixel imaging technology with commercial fiber-optic spectrometers, it is expected to develop a cost-effective fiber-optic spectral imager.\u003c/p\u003e\u003cp\u003eThis paper integrates state-of-the-art single-pixel imaging technology based on Fourier orthogonal matrix patterns with a commercial fiber-optic spectrometer, developing a novel fiber-optic hyperspectral single-pixel imager (FHSPI). This system effectively acquires spectral hypercubes, enabling the retrieval of spectral information for each pixel in the image. In fact, the small aperture of fiber-optic spectrometers results in a limited field of view (FOV) for the FHSPI\u003csup\u003e[\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e]\u003c/sup\u003e. To address this, inspired by the wide FOV of insect compound eyes\u003csup\u003e[\u003cspan additionalcitationids=\"CR43\" citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e]\u003c/sup\u003e, we developed a compound eye-like FHSPI, effectively increasing the imaging FOV area by nearly 164.8 times. This compound-eye optic fiber probe not only significantly enhances the imaging field of view but also improves the system's light utilization efficiency. Using compound eye-like FHSPI, we successfully acquired three-dimensional hypercube data of a color chart and resolved the spectral curves of each color patch. Additionally, leveraging the tunable spectral resolution of fiber-optic spectrometers, we can enhance the effective light throughput by adjusting the spectral bandwidth, significantly improving the imaging quality for specific spectral bands. More importantly, we found that the FHSPI can distinguish between real and artificial leaves at an ultra-low sampling rate, which significantly enhances the system's imaging speed and enables target spectral identification while reducing data redundancy exponentially. Our proposed FHSPI system holds great promise for advancing high-speed, low-cost spectral imaging applications, with potential breakthroughs in fields such as precision agriculture, environmental monitoring, and biomedical diagnostics, where rapid and accurate spectral analysis is critical.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003e\u003c/p\u003e\u003cp\u003eAs a mature commercial product, fiber-optic spectrometers offer significant advantages, including compact size, high sensitivity, and the ability to perform real-time, cost-effective spectral measurements across a wide wavelength range. Their working principle, illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea, involves transmitting light through optical fibers to a diffraction grating, which disperses the light into its spectral components for detection by a linear array detector, enabling precise spectral analysis. Unfortunately, its capability to acquire only a single-point spectral curve at a time restricts its range of applications. Therefore, if it could be upgraded to a fiber-optic hyperspectral imager, it would significantly enhance its spectral information density (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eb).\u003c/p\u003e\u003cp\u003eBased on this concept, we propose a single-pixel computational imaging approach for fiber-based spectral imaging, as illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ec. This method requires no modification to the existing fiber-optic spectrometer and achieves spatial-spectral information by simply encoding the illumination source using a structured light modulator (e.g., DMD or a projector). The system achieves three-dimensional spectral imaging by algorithmically reconstructing the temporal signals of structured light at different wavelengths captured by the fiber-optic spectrometer. The single-pixel imaging method assigns the task of acquiring spatial information to light source encoding and reconstruction algorithms, eliminating the need for detectors to have spatial resolution capabilities. This bypasses the highly challenging manufacturing processes required for spectral imaging detectors and offers a new possibility for developing low-cost hyperspectral imagers. The detailed architecture and imaging method principles of the FHSPI system are presented in Fig. \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e and S2.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eIt's well known that the small aperture of a single optical fiber leads to a very limited FOV. As a result, the single-pixel fiber imaging system developed using a single optical fiber, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ea, also faces the issue of a narrow FOV. Figure\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eb and \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ec show the spectrum and reconstruction images obtained using the single-fiber system at a sampling rate of 15.2%, respectively. These experimental results further confirm the issue of the narrow FOV. The imaging area has a diameter of 10 cm when the fiber is positioned 30 cm away from the scene. Based on geometric calculations, the imaging FOV angle (\u003cem\u003eθ\u003c/em\u003e) of the single fiber system is determined to be 10\u0026deg; (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ej), which matches the fiber's acceptance angle. To address the issue of limited FOV in single-fiber imaging systems, we took inspiration from the compound eyes of insects and designed a hemispherical fiber-optic port with a compound-eye structure. The probe was fabricated using high-precision 3D printing technology, featuring 16 multimode optical fibers (NA\u0026thinsp;=\u0026thinsp;0.22) arranged in a concentric 1-6-9 configuration (1 central fiber, 6 middle fibers, and 9 outer fibers) within a 20-mm-diameter housing. This fiber-optic port consists of a 1-to-16 fiber optic interface and a hemispherical shell, with 16 fibers evenly distributed across the shell, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ed. This design effectively expands the fiber's FOV.\u003c/p\u003e\u003cp\u003eTo address the issue of limited FOV in single-fiber imaging systems, we took inspiration from the compound eyes of insects and designed a hemispherical fiber-optic port with a compound-eye structure. This fiber-optic port consists of a 1-to-16 fiber optic interface and a hemispherical shell, with 16 fibers evenly distributed across the shell, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ed. This design effectively expands the fiber\u0026rsquo;s FOV. Figures\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ee and \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ef show the spectrum and reconstruction results obtained by the compound-eye-type FHSPI system at a sampling rate of 15.2%, with the entire image being well captured. Figures\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eg and \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eh present the 3D mapping of the two images (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ec, f), more clearly demonstrating the wide FOV advantage of the compound-eye system. Moreover, based on the compound-eye structure and the calculated field of view of a single fiber, we estimate the field of view angle (β) of the compound-eye system to be 130 degrees (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ej). Therefore, it can be calculated that the theoretical imaging FOV area of the compound-eye system can reach 12941.9 square centimeters (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ek), which is 164.8 times that of the single-fiber system. The photograph of the compound-eye fiber-optics has been presented in Fig. S3.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eTo verify the spectral imaging and color spectral recognition capabilities of the FHSPI system, we performed imaging on a standard color card (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea). Figure\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eb shows the spectral-spatial data cube of the standard color card obtained using the FHSPI system, which records spectral images within the wavelength range of 450 nm to 650 nm, with a spectral resolution of 1 nm. This covers almost the entire visible light spectrum. The cross-sectional view of data cube clearly demonstrates that the intensity of different color blocks varies significantly with wavelength changes. To better illustrate the intensity differences of different color blocks at various wavelengths, we extracted 16 spectral images at equal intervals of 15 nm, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ec. In addition, we extracted the spectral curves of each color block on the color card from the data cube, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ed. These experiments demonstrate that the FHSPI system has a powerful capability for acquiring spectral information of scene. It can obtain the spectral curves of each point in a scene with just a single imaging process. This significantly enhances the spectral analysis capabilities of traditional fiber-optic spectrometers. The voltage curves (V-m curves) of the spectrometer at different wavelengths have been presented in Fig. S4, where m refers to the sequence of patterns.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eIn the FHSPI system, in addition to the signal sensitivity of the spectral detector, the spectral throughput also significantly impacts imaging quality. For example, the spectral curve of the UHE-type light source (ELPLP96) we used, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ea, reveals that the light signal intensity is significantly lower near the cutoff wavelengths around 420 nm and 680 nm compared to the mid-wavelength region. This results in noticeably poorer imaging quality in these spectral bands, as demonstrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eb. Thanks to the ultra-high spectral resolution (less than 1 nm) of the fiber-optic spectrometer (AMOS LBTEK), the effective light throughput can be determined by spectral integration\u0026mdash;summing the intensity values over progressively increased FWHM while maintaining central wavelength stability. As a result, we can perform spectral integration in cutoff wavelengths regions with weaker light intensity to effectively enhance the light throughput in specific wavelength ranges, thereby potentially improving imaging quality. We compared two spectral sampling methods for the spectral images in the 425 to 440 nm wavelength range. As shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ec, the blue curve represents the uniform FWHM interval sampling method, while the gray curve corresponds to the increasing FWHM interval sampling method. Figure\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ed displays the spectral images obtained using the two sampling methods, revealing that image quality improves significantly as the FWHM increases. Taking the 430 nm spectral image as an example, when the FWHM increases from 1 nm to 10 nm, the signal-to-noise ratio (SNR) of the image rises from 10.8 to 26.5. The method for calculating the image signal-to-noise ratio (SNR) is presented in Fig. S5. Therefore, in practical applications where the focus is on acquiring spatial information, the system can significantly enhance image quality by increasing spectral throughput. Additionally, we compared the color image synthesis capabilities of the spectral sampling methods. Figure\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ef shows the original image used for imaging, while Figs.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eg and \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eh demonstrate that the wide-spectrum overlapping sampling method achieves a slight improvement in color accuracy in the synthesized images. Notably, under conditions of spectral intensity imbalance, this improvement may become more pronounced. As demonstrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ee, the adaptive sampling strategy of this method effectively compensates for intensity deficiencies in specific spectral bands, potentially reducing color distortion compared to conventional methods in controlled low-intensity scenarios. This characteristic suggests the method's particular value for applications involving non-uniform illumination conditions.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eTo validate the performance of the FHSPI system in practical applications, we conducted imaging tests on real leaf and fake leaf (Figure. 5a), aiming to distinguish between them based on spectral differences. Figure\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eb presents the spectral images of real and fake leaves extracted using FHSPI. As shown, the spectral curves of real and fake leaves exhibit significant differences beyond 650 nm, which can be effectively utilized to distinguish between them. More importantly, we found that by selecting appropriate imaging wavelengths, here is 670 nm, even low-quality imaging results with low sampling rates can effectively distinguish between real and fake leaves, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ec, d. Figure\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ee shows spectral mapping of real and fake leaves obtained by FHSPI with different sampling rates. This further indicates that the sampling rate has no impact on the extraction of spectral data for distinguishing real and fake leaves. This implies that, in practical applications, FHSPI does not require extensive time to acquire data-redundant hypercubes but instead focuses on obtaining useful key spectral information efficiently. For example, at a sampling rate of 0.09%, only 32 structured light patterns need to be projected to acquire the spectral curves of two leaves in the same scene. With a projection frequency of 50 Hz, achievable by common commercial projectors, the time required to distinguish between real and fake leaves is reduced to just 0.6 seconds. This significantly enhances the efficiency of traditional fiber-optic spectrometers, which typically require multiple sequential spectral acquisitions. Moreover, the response of the fiber-optic spectrometer can be extended to the near-infrared (NIR) spectral region. Therefore, by introducing a NIR light source, we further validated the near-infrared spectral imaging and image recognition capabilities of the proposed FHSPI system, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ef. At 880 nm, the system achieved 58% recognition accuracy for real leaves and 83% for fake leaves at 25% sampling rate, while reaching 40% (real) and 89% (fake) at 0.09% sampling rate. The near-infrared spectral imaging results and recognition accuracy (880 nm) of real and fake leaves under different sampling rates have been presented in the Fig. S6 and S8.\u003c/p\u003e\u003cp\u003eWe believe that target recognition under low sampling rates is not limited to specific conditions but rather has a certain degree of general applicability. We have validated the effectiveness of our approach across multiple experiments involving different targets and imaging backgrounds. As shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e, we present imaging and recognition results under different sampling rates. Figures\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eh and \u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ej display the imaging results at varying sampling rates, demonstrating that although image quality degrades at low sampling rates, proper wavelength selection can still effectively distinguish color differences between objects. Specifically, in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ek, we successfully discriminated between two visually similar color patches on the chromaticity chart at an extremely low sampling rate of 0.7%, and further achieved enhanced performance by differentiating two distinctly colored gourds at an ultra-low sampling rate of 0.3%.\u003c/p\u003e\u003cp\u003eThese performance variations primarily originate from differences in the number of spectral regions present in the target objects (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003el: 2 spectral regions in real/fake leaves, 7 in the multicolor gourd sample, and 24 in the standard color chart). The system's minimum required sampling rate decreases proportionally with increasing sparsity of the target spectral regions. This fundamental property enables scenario-specific projection pattern optimization, permitting rapid spectral acquisition while intrinsically avoiding the data redundancy inherent to conventional hyperspectral imaging systems.\u003c/p\u003e\u003cp\u003eTable. The performance comparison between our FHSPI and commercial spectrometers.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Taba\" border=\"1\"\u003e\u003ccolgroup cols=\"8\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eEncoding Device\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eDetector\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eSpectral Band Channels\u003c/p\u003e\u003cp\u003eResolution\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eMax Resolution\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c6\"\u003e\u003cp\u003eFOV\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c7\"\u003e\u003cp\u003eExpense/\u003c/p\u003e\u003cp\u003eRMB\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c8\"\u003e\u003cp\u003eSpectral Acquisition Speed\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eW/O\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eFS\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e350\u0026ndash;1000 nm\u003c/p\u003e\u003cp\u003e2048\u003c/p\u003e\u003cp\u003e\u0026lt;1 nm\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e10\u0026deg;\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e~\u0026thinsp;13000\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e\u0026mdash;\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e\u003cp\u003e\u003cb\u003eOur work\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003eProjector\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eFS\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e350\u0026ndash;1000 nm\u003c/p\u003e\u003cp\u003e2048\u003c/p\u003e\u003cp\u003e\u0026lt; 2 nm\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e1024\u0026times;1024\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e10\u0026deg;\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e~\u0026thinsp;16000\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e3.33 sp/s\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eCE-FHSPI\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e350\u0026ndash;1000 nm\u003c/p\u003e\u003cp\u003e2048\u003c/p\u003e\u003cp\u003e\u0026lt;2 nm\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e1024\u0026times;1024\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e130\u0026deg;\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e~\u0026thinsp;16650\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e3.33 sp/s\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eDMD\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eCE-FHSPI\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e350\u0026ndash;1000 nm\u003c/p\u003e\u003cp\u003e2048\u003c/p\u003e\u003cp\u003e\u0026lt;2 nm\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e768\u0026times;768\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e130\u0026deg;\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e~\u0026thinsp;76650\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e1250 sp/s\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e\u003cb\u003eCommercial\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eW/O\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eHyperspectral camera1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e400\u0026ndash;1000 nm\u003c/p\u003e\u003cp\u003e224\u003c/p\u003e\u003cp\u003e\u0026lt;5.5 nm\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e1024\u0026times;1024\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e38\u0026deg;\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e~\u0026thinsp;450000\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e80 sp/s\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eW/O\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eHyperspectral camera 2\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e330\u0026ndash;800 nm\u003c/p\u003e\u003cp\u003e255\u003c/p\u003e\u003cp\u003e\u0026lt;2.8 nm\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e1500\u0026times;1000\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e8\u0026deg;\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e~\u0026thinsp;640000\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e280 sp/s\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e"},{"header":"Discussions","content":"\u003cp\u003eIn order to demonstrate the advantages of FHSPI, we compared the key parameters before and after the upgrade, as well as with those of several typical commercial hyperspectral cameras. Simultaneously, we have created a radar chart that highlights the performance differences between the FHSPI system and commercial spectrometers, which is presented in Fig. S9. We designed a fiber-optic probe inspired by the compound eye, which significantly extended the field of view of the original single-fiber system (FS) with a mere 10\u0026deg; field of view to 130\u0026deg;, greatly enhancing the system's field-of-view range. The cost-effectiveness of the compound eye-like FHSPI (CE-FHSPI) system totals 16,650 RMB, which includes a projector (CB-U05 EPSON) at 3,000 RMB, a spectrometer (AMOS LBTEK) at 13,000 RMB, and a 1\u0026times;16 fiber bundle at 650 RMB, offers a substantial cost advantage over existing typical commercial spectral cameras (e.g., Pika UV Resonon), which cost up to 600,000 RMB. Although low sampling rates inevitably degrade spatial resolution, our results demonstrate that spectral resolution remains largely unaffected. We tested the spectral resolution of FHSPI at different sampling rates, all of which reached a level below 2 nm as shown in Fig. S7. To more accurately assess spectral recognition performance, we introduced a novel parameter\u0026mdash;the spectral acquisition rate\u0026mdash;defined as the number of spectra acquired per unit time and expressed in \u0026ldquo;spectra per second\u0026rdquo; (sp/s). Using a commercial projector (CB-U05 EPSON), we generated 128\u0026times;128 pixel Fourier fringes at 50 Hz. This enabled us to distinguish between the spectra of two leaves every 0.6 seconds, yielding a spectral acquisition speed of 3.33 sp/s. We plan to upgrade the encoding device to a digital micromirror device (DMD, V-7001VIS, ViALUX), which is expected to increase the projection frequency to 20,000 Hz. As a result, the system would be capable of distinguishing between two spectra every 0.0016 seconds, with the spectral acquisition speed reaching 1250 spectra per second. The most distinctive capability of our FHSPI system is its simultaneous multi-region spectral acquisition with ultra-low sampling rates. Our experimental results demonstrate that the system's minimum required sampling rate decreases with increasing sparsity of the target spectral regions. This fundamental property enables adaptive optimization of projection patterns for specific imaging scenarios, allowing the system to achieve rapid multi-region spectral collection while inherently avoiding the data redundancy that plagues conventional hyperspectral cameras. These improvements would significantly enhance the system's spectral imaging capabilities and provide robust support for high-efficiency spectral analysis. While commercial spectral cameras offer high spatial resolution, they often produce redundant data that hinder subsequent processing. In contrast, the FHSPI system addresses these limitations by balancing cost, efficiency, and performance, making it a promising solution for various applications.\u003c/p\u003e\u003cp\u003eIn summary, this work successfully upgraded the traditional fiber spectrometer to a fiber hyperspectral imager (FHSPI) through single-pixel imaging technology, thereby addressing the limitations of conventional fiber spectrometers. With the aid of a compound-eye structure design, the FHSPI significantly enhanced the imaging FOV and light utilization efficiency, and enabled rapid differentiation of complex samples at low sampling rates. Experimental results demonstrated the system's high efficiency in spectral imaging and target recognition, particularly under low sampling rate conditions. As a single-pixel imaging system, FHSPI inherits the intrinsic advantages of strong resistance to environmental interference and motion artifacts, ensuring robust performance in practical applications. The low cost and high performance of FHSPI have opened new avenues for the widespread application of spectral imaging technology, and are expected to promote technological breakthroughs and innovations in many fields. It should be noted that as a structured illumination system, FHSPI is particularly suitable for endoscopic and industrial applications where active light control is feasible, though its applicability in passive remote sensing scenarios is inherently limited by the requirement for active illumination. Future work will focus on developing a structured detection-based hyperspectral imaging system to overcome this limitation and expand the methodology's applicability to broader scenarios.\u003c/p\u003e"},{"header":"Experimental Section","content":"\u003cp\u003eIn this experiment, we used a commercial digital projector (CB-U05 EPSON) with a brightness of 3400 lm and a contrast ratio of 15000:1. The projector has a resolution of 1920\u0026times;1200 pixels and is capable of generating high-contrast binary or grayscale patterns. Its light source is a 210-watt white LED, which projects patterns onto the target object via an optical system. In the experimental setup, the projector sequentially displayed 8192 Fourier fringes with a resolution of 128 \u0026times; 128 pixels every 0.02 seconds. The pattern size was matched to the imaging area of the target object to ensure sufficient accumulation of light signals. We also employed a fiber-optic spectrometer (AMOS LBTEK) with a spectral range ranging from 350 nm to 1000which each fiber has a core diameter of 200 \u0026micro;m, a total length of 1 meter, and a numerical aperture (NA) of 0.22.\u003c/p\u003e\u003cp\u003eIn this experiment, the imaging targets are a 24-color standard color card (TCCR-S Mennon), seven printed bottle gourd samples of different colors, as well as real and artificial leaves. The color card is made of materials such as ABS and PVC. It features a rich distribution of colors and stable physical properties and is commonly used for color correction in imaging devices. In this study, we utilized it to evaluate the imaging system's color reproduction and detail resolution capabilities, which are critical for assessing the accuracy and performance of the system. In addition to the color card, we also conducted a comparative imaging analysis of real and artificial leaves. The real leaf samples are from pothos (Epipremnum aureum), whose leaves exhibit typical characteristics of green plants, making them an ideal natural specimen for the study. The artificial leaf samples, made primarily of plastic, effectively mimic the appearance and texture of real leaves. This comparison enables us to assess the system's ability to distinguish between natural and synthetic materials, further demonstrating its versatility and precision.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAuthors acknowledge the financial supports from the National Natural Science Foundation of China (62405236, 62375210, 62275210, 62105123); the National Young Talent Program; the Shaanxi Young Top-notch Talent Program; Xidian University Specially Funded Project for Interdisciplinary Exploration (TZJH2024059, TZJH2024060); Fundamental Research Funds for Central Universities (QTZX24079).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eNotes\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe manuscript was written through the contributions of all authors. All authors have given approval to the final version of the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare no competing financial interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eData available on request from the authors.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSupporting Information\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSupporting Information is available from the Wiley Online Library or from the author.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eCadeado, A. N. \u0026amp; Machado, C. C.\u0026amp;Silva, S. G. Simultaneous reflectance and fluorescence measurements for portable formaldehyde determination in milk using a multi-channel spectrometer sensor. \u003cem\u003eFood Chem.\u003c/em\u003e 464, 141583 (2025).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eChen, L. \u0026amp; Lin, Q. \u0026amp; Yuan, D.\u0026amp;Gong, Z. Simultaneous measurement of trace levels of Hg, as, Sb, and Bi in coastal seawater with a multichannel chemical vapor generation atomic fluorescence spectrometer. \u003cem\u003eAnal. Chem.\u003c/em\u003e 95, 15621\u0026ndash;15627 (2023).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCoburn, S. \u0026amp; Alden, C. B. \u0026amp; Wright, R., et al. Regional trace-gas source attribution using a field-deployed dual frequency comb spectrometer. \u003cem\u003eOptica\u003c/em\u003e 5, 320\u0026ndash;327 (2018).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYang, Z. \u0026amp; Yu, X. \u0026amp; Dedman, S., et al. UAV remote sensing applications in marine monitoring: Knowledge visualization and review. \u003cem\u003eSci. Total Environ.\u003c/em\u003e 838, 155939 (2022).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eEl-Mashtoly, S. F.\u0026amp;Gerwert, K. Diagnostics and therapy assessment using label-free Raman imaging. \u003cem\u003eAnal. Chem.\u003c/em\u003e 94, 120\u0026ndash;142 (2021).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHuang, L. \u0026amp; Sun, H. \u0026amp; Sun, L., et al. Rapid, label-free histopathological diagnosis of liver cancer based on Raman spectroscopy and deep learning. \u003cem\u003eNat. Commun.\u003c/em\u003e 14, 48 (2023).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eOlmanson, L. G. \u0026amp; Brezonik, P. L.\u0026amp;Bauer, M. E. Airborne hyperspectral remote sensing to assess spatial distribution of water quality characteristics in large rivers: The Mississippi River and its tributaries in Minnesota. \u003cem\u003eRemote Sens. Environ.\u003c/em\u003e 130, 254\u0026ndash;265 (2013).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZhong, Y. \u0026amp; Wang, X. \u0026amp; Wang, S.\u0026amp;Zhang, L. Advances in spaceborne hyperspectral remote sensing in China. \u003cem\u003eGeo-spatial Inf. Sci.\u003c/em\u003e 24, 95\u0026ndash;120 (2021).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRomani, A. \u0026amp; Clementi, C. \u0026amp; Miliani, C.\u0026amp;Favaro, G. Fluorescence spectroscopy: a powerful technique for the noninvasive characterization of artwork. \u003cem\u003eAcc. Chem. Res.\u003c/em\u003e 43, 837\u0026ndash;846 (2010).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZeng, Z. \u0026amp; Liu, Y.\u0026amp;Wei, J. Recent advances in surface-enhanced raman spectroscopy (SERS): Finite-difference time-domain (FDTD) method for SERS and sensing applications. \u003cem\u003eTrAC, Trends Anal. Chem.\u003c/em\u003e 75, 162\u0026ndash;173 (2016).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ePilling, M.\u0026amp;Gardner, P. Fundamental developments in infrared spectroscopic imaging for biomedical applications. \u003cem\u003eChem. Soc. Rev.\u003c/em\u003e 45, 1935\u0026ndash;1957 (2016).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZhou, X. \u0026amp; Chen, C. \u0026amp; Zuo, E., et al. Cross Branch Co-Attention Network multimodal models based on Raman and FTIR spectroscopy for diagnosis of multiple selected cancers. \u003cem\u003eAppl. Soft Comput.\u003c/em\u003e 166, 112204 (2024).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKnez, D. \u0026amp; Toulson, B. W. \u0026amp; Chen, A., et al. Spectral imaging at high definition and high speed in the mid-infrared. \u003cem\u003eSci. Adv.\u003c/em\u003e 8, eade4247 (2022).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRam, B. G. \u0026amp; Oduor, P. \u0026amp; Igathinathane, C., et al. A systematic review of hyperspectral imaging in precision agriculture: Analysis of its current state and future prospects. \u003cem\u003eComput. Electron. Agric.\u003c/em\u003e 222, 109037 (2024).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWang, K. \u0026amp; Pu, H.\u0026amp;Sun, D. W. Emerging spectroscopic and spectral imaging techniques for the rapid detection of microorganisms: An overview. \u003cem\u003eCompr. Rev. Food Sci. Food Saf.\u003c/em\u003e 17, 256\u0026ndash;273 (2018).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHu, B.-L. \u0026amp; Hao, S.-J. \u0026amp; Sun, D.-X.\u0026amp;Liu, Y.-N. A novel scene-based non-uniformity correction method for SWIR push-broom hyperspectral sensors. \u003cem\u003eISPRS J. Photogramm. Remote Sens.\u003c/em\u003e 131, 160\u0026ndash;169 (2017).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eJemec, J. \u0026amp; Pernuš, F. \u0026amp; Likar, B.\u0026amp;B\u0026uuml;rmen, M. Deconvolution-based restoration of SWIR pushbroom imaging spectrometer images. \u003cem\u003eOpt. Express\u003c/em\u003e 24, 24704\u0026ndash;24718 (2016).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLv, S.\u0026amp;Liu, Y.-N. Anomaly detection in satellite-borne push-broom hyperspectral imagery based on joint low-rank tensor approximation. \u003cem\u003eIEEE Trans. Geosci. Remote Sens.\u003c/em\u003e 62, 1\u0026ndash;20 (2023).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ePham, Q. T.\u0026amp;Liou, N.-S. The development of on-line surface defect detection system for jujubes based on hyperspectral images. \u003cem\u003eComput. Electron. Agric.\u003c/em\u003e 194, 106743 (2022).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSarić, R. \u0026amp; Nguyen, V. D. \u0026amp; Burge, T., et al. Applications of hyperspectral imaging in plant phenotyping. \u003cem\u003eTrends Plant Sci.\u003c/em\u003e 27, 301\u0026ndash;315 (2022).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYeh, K. \u0026amp; Sharma, I. \u0026amp; Falahkheirkhah, K., et al. Infrared spectroscopic laser scanning confocal microscopy for whole-slide chemical imaging. \u003cem\u003eNat. Commun.\u003c/em\u003e 14, 5215 (2023).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHua, X. \u0026amp; Wang, Y. \u0026amp; Wang, S., et al. Ultra-compact snapshot spectral light-field imaging. \u003cem\u003eNat. Commun.\u003c/em\u003e 13, 2732 (2022).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMcClung, A. \u0026amp; Samudrala, S. \u0026amp; Torfeh, M., et al. Snapshot spectral imaging with parallel metasystems. \u003cem\u003eSci. Adv.\u003c/em\u003e 6, eabc7646 (2020).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZheng, S. \u0026amp; Liu, Y. \u0026amp; Meng, Z., et al. Deep plug-and-play priors for spectral snapshot compressive imaging. \u003cem\u003ePhotonics Res.\u003c/em\u003e 9, B18-B29 (2021).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGlaser, A. K. \u0026amp; Reder, N. P. \u0026amp; Chen, Y., et al. Multi-immersion open-top light-sheet microscope for high-throughput imaging of cleared tissues. \u003cem\u003eNat. Commun.\u003c/em\u003e 10, 2781 (2019).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWang, Y. \u0026amp; Peng, H. \u0026amp; Liu, K., et al. Multi-point scanning confocal Raman spectroscopy for accurate identification of microorganisms at the single-cell level. \u003cem\u003eTalanta\u003c/em\u003e 254, 124112 (2023).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eXintian, W. \u0026amp; Xuesong, M. \u0026amp; Xiaodong, W., et al. Motion position prediction and machining accuracy compensation of galvanometer scanner based on BWO-GRU model. \u003cem\u003eMech. Syst. Sig. Process.\u003c/em\u003e 210, 111081 (2024).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eEdgar, M. P. \u0026amp; Gibson, G. M.\u0026amp;Padgett, M. J. Principles and prospects for single-pixel imaging. \u003cem\u003eNat. Photonics\u003c/em\u003e 13, 13\u0026ndash;20 (2019).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGibson, G. M. \u0026amp; Johnson, S. D.\u0026amp;Padgett, M. J. Single-pixel imaging 12 years on: a review. \u003cem\u003eOpt. Express\u003c/em\u003e 28, 28190\u0026ndash;28208 (2020).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZhang, Z. \u0026amp; Liu, S. \u0026amp; Peng, J., et al. Simultaneous spatial, spectral, and 3D compressive imaging via efficient Fourier single-pixel measurements. \u003cem\u003eOptica\u003c/em\u003e 5, 315\u0026ndash;319 (2018).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZhang, Z. \u0026amp; Ma, X.\u0026amp;Zhong, J. Single-pixel imaging by means of Fourier spectrum acquisition. \u003cem\u003eNat. Commun.\u003c/em\u003e 6, 6225 (2015).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLiu, Y. \u0026amp; Ji, Z. \u0026amp; Cen, G., et al. Perovskite-based color camera inspired by human visual cells. \u003cem\u003eLight Sci. Appl.\u003c/em\u003e 12, 43 (2023).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLiu, Y. \u0026amp; Lin, D. \u0026amp; Xing, J., et al. Mode-switching single-pixel imaging via a high-performance perovskite-Si dual-mode photodetector. \u003cem\u003ePhotonics Res.\u003c/em\u003e 12, 2873\u0026ndash;2880 (2024).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eXu, Y. \u0026amp; Lu, L. \u0026amp; Saragadam, V.\u0026amp;Kelly, K. F. A compressive hyperspectral video imaging system using a single-pixel detector. \u003cem\u003eNat. Commun.\u003c/em\u003e 15, 1456 (2024).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTao, C. \u0026amp; Zhu, H. \u0026amp; Wang, X., et al. Compressive single-pixel hyperspectral imaging using RGB sensors. \u003cem\u003eOpt. Express\u003c/em\u003e 29, 11207\u0026ndash;11220 (2021).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eJin, S. \u0026amp; Hui, W. \u0026amp; Wang, Y., et al. Hyperspectral imaging using the single-pixel Fourier transform technique. \u003cem\u003eSci. Rep.\u003c/em\u003e 7, 45209 (2017).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eChen, W. \u0026amp; Sun, M.-J. \u0026amp; Deng, W.-J., et al. Hyperspectral imaging via a multiplexing digital micromirror device. \u003cem\u003eOpt. Lasers Eng.\u003c/em\u003e 151, 106889 (2022).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eJiang, X. \u0026amp; Li, Z. \u0026amp; Du, G., et al. Fast hyperspectral single-pixel imaging via frequency-division multiplexed illumination. \u003cem\u003eOpt. Express\u003c/em\u003e 30, 25995\u0026ndash;26005 (2022).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMeng, H. \u0026amp; Gao, Y. \u0026amp; Wang, X., et al. Quantum dot-enabled infrared hyperspectral imaging with single-pixel detection. \u003cem\u003eLight Sci. Appl.\u003c/em\u003e 13, 121 (2024).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eOchoa, M. \u0026amp; Rudkouskaya, A. \u0026amp; Yao, R., et al. High compression deep learning based single-pixel hyperspectral macroscopic fluorescence lifetime imaging in vivo. \u003cem\u003eBiomed. Opt. Express\u003c/em\u003e 11, 5401\u0026ndash;5424 (2020).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTian, Y. \u0026amp; Ding, Z. \u0026amp; Feng, H., et al. Single-pixel imaging based on optical fibers. \u003cem\u003eIEEE Photonics J.\u003c/em\u003e 12, 1\u0026ndash;7 (2020).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eJi, Z. \u0026amp; Liu, Y. \u0026amp; Zhao, C., et al. Perovskite Wide-Angle Field‐Of‐View Camera. \u003cem\u003eAdv. Mater.\u003c/em\u003e 34, 2206957 (2022).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eJiang, H. \u0026amp; Tsoi, C. C. \u0026amp; Yu, W., et al. Optical fibre based artificial compound eyes for direct static imaging and ultrafast motion detection. \u003cem\u003eLight Sci. Appl.\u003c/em\u003e 13, 256 (2024).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMa, M. \u0026amp; Zhang, Y. \u0026amp; Deng, H., et al. Super-resolution and super-robust single-pixel superposition compound eye. \u003cem\u003eOpt. Lasers Eng.\u003c/em\u003e 146, 106699 (2021).\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-7333525/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7333525/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eTraditional fiber-optic spectrometers, while featuring compact designs and high sensitivity, are constrained to single-point spectral acquisition with limited spatial resolution, thereby restricting their broder applications. Although existing commercial spectral imaging systems can provide three-dimensional spectral cubes, they often require complex optical configurations, resulting in high costs and operational complexity. To overcome these limitations, this paper proposes a fiber-optic hyperspectral single-pixel imager (FHSPI) based on single-pixel computational imaging technology. In addition, by integrating a fiber-optic probe inspired by the compound eyes of insects, the FHSPI significantly expands the imaging field of view (FOV) to 130 degrees, representing a substantial increase from the 10-degree FOV of a single fiber. The FHSPI we proposed can achieve spectral image data cubes with a resolution of 1 nm, and by developing a spectral flux integration method, it enables high-quality spectral imaging under low-light conditions. Operating at a sub-0.1% sampling rate (0.09%), the FHSPI enables rapid distinguishing between authentic from artificial green leaves with a theoretical spectral identification speed of up to 1250 spectra per second (sp/s). We present a novel approach for developing cost-effective, high-performance fiber-optic spectral imaging technology that harnesses the potential advantages of FHSPI in addressing the data redundancy issues faced by conventional hyperspectral imagers. This advancement holds significant potential for fostering innovations in various applications, including precision agriculture, environmental monitoring, and biomedical diagnostics.\u003c/p\u003e","manuscriptTitle":"Upgrading fiber-optic spectrometers to hyperspectral imagers","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-09-23 02:27:09","doi":"10.21203/rs.3.rs-7333525/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"2dc4b700-88ab-44cc-8279-f93f4990948f","owner":[],"postedDate":"September 23rd, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":54619998,"name":"Physical sciences/Optics and photonics/Optical techniques/Optical spectroscopy"},{"id":54619999,"name":"Physical sciences/Optics and photonics/Optical techniques/Imaging and sensing"}],"tags":[],"updatedAt":"2025-10-21T10:19:01+00:00","versionOfRecord":[],"versionCreatedAt":"2025-09-23 02:27:09","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7333525","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7333525","identity":"rs-7333525","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}