Open Access
Add to BibSonomy
Abstract
The function of a mask in the integral field imaging spectrometer (IFIS), which segments image and samples, leads to the drawback of low spectral energy transmittance. Here, we improve field-of-view segmentation method and propose a dual micro-lens array imaging spectrometer (DMAIS). DMAIS comprises a projection lens (PL), a segmentation collimation module (SCM), and a telecentric lens (TL). And SCM, based on a dual micro-lens array, is the core component of it. By employing a lens array focusing approach instead of aperture sampling, DMAIS effectively enhances energy transmittance and reduces spectral bending. The ZEMAX simulation results indicate that compared to IFIS, DMAIS demonstrates a 109.2% increase in energy transmittance and a 32.9% reduction in spectral bending.
© 2024 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Optical spectrometers have found diverse applications across multiple domains [1–9]. Among these, dispersive spectrometers have long been a focus of international research due to their straightforward design. Evolving over time, spectral imaging techniques are classified into swing-scan, push-broom, and stare types based on the scanning approach for acquiring three-dimensional data cubes [10]. Stare-type spectral imaging systems, employing area array detectors, are suitable for tunable filter and innovative snapshot imaging spectrometers. Snapshot imaging spectrometers possess the capability to concurrently capture spatial and spectral information, boasting a more precise geometric structure and representing a focal area in imaging spectrometry research. However, these snapshot systems encounter challenges related to spectral bending and diminished energy throughput, which impede their application and advancement.
Recently, numerous researchers have focused on employing algorithms for post-correction of spectral bending to enhance spectral image quality. Qian et al. analyzed the influence of different spectral line distortions on spectral imaging and highlighted that, to accurately reconstruct target scenes, the spectral offset caused by line distortion should be controlled within half a unit pixel [11]. Spectral bending often leads to spectral overlap. To address this issue, Zheng et al. proposed a spectral bending correction algorithm based on curve distance, aiming to rectify the distortion by calculating imaging methods [12]. However, this algorithm is suitable only for fiber-optic spectral images and demonstrates lower accuracy for broader spectral curves. Scholars have tackled spectral bending by enhancing the optical structure from the imaging module. Zhang et al. introduced a method that utilizes off-axis lenses to correct spectral and trapezoidal distortions [13]. By employing optical correction methods, they effectively constrained spectral bending to within half the size of a unit pixel. Therefore, compared to algorithmic correction, optical correction appears to fundamentally mitigate spectral bending and holds greater universality.
Snapshot imaging spectrometers aim to capture spatial and spectral information simultaneously in a single shot. In pursuit of it, Cao et al. introduced a prism-mask multispectral video imaging system (PMVIS) where incident light from the scene is sampled by a mask and dispersed into spectra using a prism [14]. This system replaces a slit with a mask to control spectral imaging on the CCD, enhancing spatial resolution and reducing spectral bending. Prisms offer advantages such as low cost, no spectral overlap, and good stability, thereby reducing the optical system's expenses [15–18]. Liu et al. improved PMVIS by adopting a traditional prism configuration, achieving a design with an average spectral resolution of 10 nm, thereby enhancing the spectral resolution. [19]. However, this system becomes cumbersome. To address this, Li et al. proposed an objective lens and a collimating module, achieving a lightweight transformation of the snapshot spectral optical system [20]. Yet, the limited field of view of the objective lens hinders the acquisition of extensive visual field information. Michal E. Pawlowski et al. proposed an Image Mapping Spectrometer (IMS) with a working wavelength range of nm $515\sim 842\textrm{nm}$, achieving high-precision spectral imaging [21]. However, IMS also faces the challenge of a small field of view, limiting the development and application of IMS. To expand the field of view, Zeng et al. proposed improvements to the objective lens, introducing the IFIS [22]. IFIS features a wide-field spatial telecentric lens that captures large scene images. It utilizes micron-level mask sampling and disperses light through a prism-grating combination to achieve large field-of-view and high spectral resolution imaging. Conventional snapshot imaging spectrometers, commonly using mask sampling and segmenting scene information, isolate a significant portion of light, leaving space for spectral dispersion, thus enabling the simultaneous capture of spatial and spectral information. However, the use of masks blocks a considerable amount of light from entering the dispersion module, resulting in a low system energy transmittance. The dispersion module in IFIS consists of a prism and a grating. Due to the width and length of the mask, only the light emitted from the central aperture of the mask refracts on the main plane of the prism after collimation. Light emitted from the remaining apertures of the mask enters the prism at different angles, causing spectral bending.
Addressing the aforementioned issues, the image segmentation module by a dual-micro-lens array instead of the mask sampling is improved. We propose DMAIS and have conducted optical design for its core module, SCM. SCM is composed of five optical components: a collimating lens, the first micro-lens array, a planar aperture, the second micro-lens array, and a prism. Within DMAIS, the collimating lens aligns the beam focused by the PL to enter the first micro-lens array in a parallel manner. Sub-lenses with identical diameter and aligned center points in the first and second micro-lens arrays segment and collimate parallel beams, achieving image sampling with minimal energy loss. The planar aperture isolates improperly focused light, reducing stray light interference and spectral bending. The prism, in conjunction with the second micro-lens array, regulates the size of dispersed beams to prevent spectral overlap.
2. Optical system design and principles
2.1 Structural design of DMAIS
A schematic of DMAIS is shown in Fig. 1. The system comprises the PL, SCM, and TL. The SCM, serving as the central component, undertakes the tasks of segmenting scenes, narrowing beams, and dispersing light. The PL captures extensive scene details and light rays through the collimating lens to parallelize them slightly smaller than the micro-lens array. The light enters the first micro-lens array at a size slightly smaller than it. Both the first and second micro-lens arrays share uniform sub-lens diameters and arrangements. In a microlens array, the primary purpose of view reuse is to enable multi-angle acquisition. When light rays enter the microlens array from different directions, they form different focal spots or imaging points on the sensor. By analyzing the positions and characteristics of these focal spots or imaging points, image information from different perspectives can be obtained. However, DMAIS collimates the light rays into parallel rays before they enter the microlens array, causing the light to enter in the same direction and preventing the sub-lenses on the microlens array from producing multi-angle images. Instead, DMAIS only performs the task of field segmentation. Therefore, the DMAIS system does not involve view reuse.
The dual micro-lens array model, featuring a one-to-one correspondence among sub-lenses, allows the segmentation and subsequent individual collimation of acquired light. It ensures that each lens emits light refracted on the prism's main plane. Positioned at the rear focal length of the first micro-lens array and the front focal length of the second micro-lens array, the mask no longer serves as an image segmentation sampling element but functions as a planar aperture. Planar aperture isolates inappropriately focused light to mitigate stray light interference and spectral bending. As the exclusive dispersion element, the prism cooperates with the second micro-lens array in controlling spectral dispersion. The TL boasts stable magnification and eliminates perspective distortions, resulting in clear and undistorted image acquisition, substantially reducing potential errors. Hence, the selection of the TL as the spectral image capture device facilitates the transmission of spectral images to the CCD.
Through dual micro-lens array focusing, DMAIS achieves high-energy throughput image segmentation. Additionally, it ensures the consistent and parallel incidence of light on the prism based on the one-to-one correspondence pattern among sub-lenses. Thus, spectral bending caused by incident light entering the prism at varying angles during dispersion is prevented.
2.2 Optical principles of SCM
The schematic diagram of SCM is shown in Fig. 2. The image side aperture angle of PL ($u^{\prime}$) can be obtained from the relative aperture ($1/\textrm{F}$), thus allowing $u^{\prime}$ to be determined by Eq. (1).
Given the known dimension of the micro lens array ($d$), the diameter of the collimated light after passing through the lens should be less than d yet greater than the unit micro lens diameter (${d_m}$). Owing to the minute size of the individual micro lenses, errors can occur during the adjustment process. Hence, the light's radius should exceed ${d_m}$. The focal length ($f^{\prime}$) of the collimating lens ($L$) adheres to the relationship described by Eq. (2).
Fig. 2. Schematic diagram of SCM. (a) Optical path diagram of SCM; (b) Optical path diagram of prism.
Diffraction effects can be neglected only when the source size is significantly larger than the wavelength by a factor of 100. This system is designed for visible light imaging, with an operational wavelength of $420\sim 700\textrm{nm}$. To avoid spatial wastage, the unit lens size of the micro-lens array should be smaller than $0.7\textrm{mm}$. Consequently, diffraction effects in the system cannot be disregarded and require analysis.
According to the Fresnel approximation conditions,
where z represents the distance between the sampling plane and the diffraction plane; $(x,y)$ denotes the coordinates in the sampling plane; $({x_1},{y_1})$ represents the coordinates in the diffraction plane; $\lambda $ stands for the wavelength. From Eq. (3), we obtain $z \gg 2.109mm$.According to the Fraunhofer approximation conditions,
we can obtain $z \gg 175mm$. In this paper, z falls within the range of $(2.109,175)$. Hence, the system falls under the Fresnel approximation.For a lens array consisting of $N \times N$ lenses, each with identical physical structure and dimensions, the transmission function $t({x_1},{y_1})$ of the entire diffraction screen can be expressed as a combination of the transmission function ${t_0}({x_1},{y_1})$ of a single lens repeated $N \times N$ times:
Assuming ${\tilde{{\textbf E}}_0}({x_1},{y_1})$ is the complex amplitude distribution of the illuminating light field before passing through the diffraction screen, and $\tilde{{\textbf E}}({x_1},{y_1})$ is the complex amplitude distribution immediately after passing through the diffraction screen, their relationship is given by
Let $\tilde{{\textbf E}}(x,y) = 0$ and determine the positions on the sampling plane where the light intensity is zero. Additionally, employ Eq. (7) as a constraint to ensure that each aperture only emits rays from its corresponding micro-lens unit.
In order to ensure a one-to-one correspondence between sub-apertures of the dual micro-lens array, the unit lens diameters of the first and second micro-lens arrays are identical. As depicted in Fig. 2, there exists a constraint among the focal lengths of the first micro-lens array (${f_1}$), the second micro-lens array (${f_2}$), the unit lens diameter (${d_m}$), and the width of the micro-beam (${d_l}$), which follows the principle of similar triangles
Figure 2(b) illustrates the dispersion characteristics of a dispersive prism when a single parallel micro-beam undergoes dispersion. The width of the micro-beam is denoted by ${d_l}$. After dispersion within the prism, it forms spectral lines. The projection of these spectral lines onto an optical screen at a distance l from the prism results in a length $d^{\prime}$. Here, the prism is denoted by $\Delta ABC$, with an apex angle $\alpha $ and a rotation angle $\gamma $. Parallel segments BE and FM intersect the perpendicular line within the prism at points E and intersect the optical screen at points M, respectively. Additionally, line FK is drawn parallel to GH, intersecting the optical screen at point K.
The angular dispersion of the prism controls the length of the spectral lines. To prevent spectral overlap, the light should pass through the prism at the minimum incident angle for the initial wavelength ${\lambda _1}$. Additionally, $d^{\prime}$ needs to satisfy the following condition:
Then $\gamma$ can be expressed as
where the refractive index of the prism for the ${\lambda _1}$ is denoted as ${n_{\lambda 1}}$. For an equilateral dispersing prism, $\alpha = 60^\circ $.The range of light wavelength is constrained by ${\lambda _1}$ and ${\lambda _2}$, and the range of values for l can be determined by Eq. (11).
3. Structure design and simulation of DMAIS
3.1 System design metrics
To validate the feasibility of the spectrometer design based on the dual micro-lens array segmentation model, the system was designed and optimized according to the design criteria presented in Table 1.
Table 1. Design Indicators of Spectrometer.
3.2 Parameter calculation and system simulation
Smaller micro-lens unit diameters lead to a larger number of micro-beams. However, excessively small sizes result in diffraction will cause the light to deviate from the geometric optical path. To determine the micro-lens unit size, we used ZEMAX simulations based on Eq. (10) and Eq. (11) under conditions of F = 3.5 for different micro-lens unit diameters. As shown in Fig. 3, at 0.4 mm diameter, adjacent apertures exhibit diffraction; at 0.5 mm diameter, adjacent apertures show minimal diffraction and appropriate spacing; at 0.6 mm diameter, adjacent apertures have minimal diffraction but excessive spacing, leading to spatial wastage. Therefore, the chosen micro-lens array unit diameter for this system fulfills the following criteria:
For any single-wavelength calibration image, comprising $15 \times 15$ points, with an effective pixel count of $208 \times 156$, the spatial resolution is $208 \times 156$. In the same optical system, the module numerical aperture remains consistent. Hence, the numerical aperture value for PL is taken as 3.5, with its key parameters shown in Table 2.
Table 2. Main parameters of PL.
where FOV represents the field of view, f is the focal length, and VL stands for volume length.
The focal length range of the collimating lens can be obtained based on Eqs. (1), (2), and (12).
The refractive apex angle of the equilateral dispersive prism is denoted as $\alpha = 60^\circ $, with material specified as ${\textrm{F}_2}$. Based on Eq. (18), the rotational angle can be calculated as
Substituting Eq. (11), the range of values for l can be determined
According to the mentioned parameters, the ZEMAX simulation diagram of the complete optical structure for DMAIS is illustrated in Fig. 4.
In accordance with the design specifications of the spectrometer, the operational wavelength range for this optical system is $420\sim 700\textrm{nm}$, with a central wavelength of 500 nm, and an energy transmittance exceeding 50%. In the IFIS, the mask serves as the image segmentation component, influencing the transmittance of light and the spectral bending. In contrast, the DMAIS utilizes a dual micro-lens array to segment the scene, cutting the light into multiple micro-beams and collimating them. All these micro-beams exit in parallel in the same direction and undergo refraction at the main section of the prism. To validate this structure, we compare the energy transmittance of IFIS and DMAIS under different mask apertures, as depicted in Fig. 5.
Fig. 5. Relationship diagram between energy transmittance and mask aperture of IFIS and DMAIS. (a) Energy transmittance of different wavelengths at different apertures; (b) Total energy transmittance at different apertures; (c) Partial spectrum image of IFIS under 0.05 mm aperture mask; (d) Partial spectrum image of IFIS under 0.1 mm aperture mask; (e) Partial spectrum image of DMAIS under 0.05 mm aperture mask; (f) Partial spectrum image of DMAIS under 0.1 mm aperture mask.
As shown in Fig. 5(a), the aperture sizes of mark vary from 0.01 mm to 0.1 mm with a 0.01 mm interval. The energy transmittance of different wavelength light in both IFIS and DMAIS optical systems is observed. In IFIS, the energy transmittance is not significantly influenced by wavelength for aperture sizes in the range of 0.01 mm to 0.1 mm. In DMAIS, the transmittance of shorter-wavelength light is more affected by aperture sizes, while the transmittance of light with a wavelength greater than 660 nm is almost unaffected by the aperture. Figure 5(b) illustrates the relationship between the total energy transmittance of IFIS and DMAIS optical systems and the aperture sizes of mark, ranging from 0.01 mm to 0.1 mm with a 0.01 mm interval. It can be observed that the energy transmittance increases with the enlargement of the mask aperture. When the aperture size of mark is 0.01 mm, the light is almost unable to pass through IFIS, while DMAIS still achieves a transmittance of 38.5%. Moreover, in the DMAIS system, the transmittance reaches 100% when the aperture increases to 0.08 mm.
When the mask aperture is 0.1 mm, the energy transmittance of IFIS increases to its maximum at 47.8%, which is only 47.8% of the transmittance achieved by the DMAIS system. Therefore, compared to IFIS, DMAIS shows a remarkable 109.2% increase in energy transmittance, addressing the issue of low light flux caused by the small mask aperture in IFIS. It significantly enhances the brightness and clarity of spectral images, making them easier to capture.
To validate DMAIS's improvement in energy transmittance by experiment, we alternately inserted aperture masks with diameters of 0.05 mm and 0.1 mm into both IFIS and DMAIS systems, recording spectral images under these two types of masks. As illustrated in Fig. 5(c), (d), (e), and (f), comparison of images from the same optical system under different apertures reveals a positive correlation between mask aperture diameter and energy transmittance. Furthermore, comparing imaging results of IFIS and DMAIS under the same aperture, it was observed that spectral images obtained by DMAIS exhibited superior brightness and imaging quality compared to IFIS. The imaging results of IFIS displayed significant spectral bending and distortion. In contrast, DMAIS imaging results showed significant improvements in spectral bending and distortion. Therefore, we conclude that DMAIS effectively enhances the system's energy transmittance, reduces spectral bending, and mitigates system distortion.
Spectral bending is a crucial parameter influencing imaging results. We tracked and compared the spectral bending of IFIS and DMAIS at seven different wavelengths, and the results are presented in Fig. 6. It can be observed that longer-wavelength light exhibits more significant spectral bending. The maximum spectral curvature for IFIS is 31% per pixel, while for DMAIS, it is reduced to 20.8% per pixel. Compared to IFIS, DMAIS shows a 32.9% reduction in spectral curvature.
Distortion can result in changes to the position, shape, size and information of objects in an image. Thereby, distortion causes image warping and impacts the accuracy and quality of optical system imaging. In image processing, distortion may manifest as either positive or negative distortion, depending on the direction of the distortion. When distortion causes objects in the image to expand outward relative to their actual positions, it is called barrel distortion and is represented by positive values. Conversely, when distortion causes objects in the image to contract inward relative to their actual positions, it is called pincushion distortion and is represented by negative values. Within the wavelength range of $400\sim 550\textrm{nm}$, pincushion distortion is observed. Hence the distortion is negative. The sign of the distortion depends on the direction of the distortion, while the numerical value quantifies the degree of distortion, which is influenced by the wavelength. The distortion is minimal at the central wavelength, with greater distortion observed as the wavelength deviates from the central wavelength. The X-axis represents the position of objects in the field of view of the optical system. 0 denotes the center of the field of view, while 1 represents the field of view boundary. We plotted the distortion of DMAIS at seven different wavelengths, as shown in Fig. 7. According to Fig. 7, it can be observed that the distortion is minimal at 500 nm, at only 0.2%, while the distortion is maximal at 420 nm, at 4.3%. According to the design specifications, the distortion should be less than 5%. Therefore, DMAIS meets the design requirements.
Through ZEMAX simulation, diffraction patterns at the central field of the system for different wavelengths can be obtained. According to Rayleigh's criterion, two non-coherent point sources with equal intensities can be just resolved if the center of one diffraction pattern falls precisely on the edge of the other. As shown in Fig. 8, we simulated and obtained the spectral resolution of DMAIS at seven wavelengths. According to Rayleigh's criterion, the spectral resolution of an optical system under ideal conditions should be less than 5 nm.
4. Experiment and data analysis
To validate the feasibility of the design, an experimental setup was constructed based on the 1:1 scale reproduction of the ZEMAX simulation in Fig. 4. The experimental system includes a displacement platform with an accuracy of 10 and some partial cage-like structures, as shown in Fig. 9.
We used a compound light source as the illumination for capturing images of the physical object. An object was placed in front of PL, and the actual spectral image was captured by TL and a gray camera, as shown in Fig. 10(b). A continuous spectrum laser with a spectral resolution of 1.4 nm, serving as the standard light source, was utilized along with a monochromator to provide monochromatic light at specific wavelengths. The wavelength range emitted by the continuous spectrum laser is $400\sim 2400\textrm{nm}$, while the monochromator covers wavelengths from $400\sim 1300\textrm{nm}$. A set of single-wavelength calibration images were captured using DMAIS under the standard light source, as shown in Fig. 10(c). This set of images covers wavelengths in range of $420\sim 700\textrm{nm}$ with a wavelength interval of 5 nm, totaling 56 images. Using the wavelength scanning method, all the single-wavelength calibration images were filtered and overlaid to obtain a calibration reference image, as shown in Fig. 10(d). The actual spectral image aligns with the positions of the same-wavelength lines in the calibration spectral image aligns with the positions of the same-wavelength lines in the calibration reference image. Consequently, the spectral image can be deconstructed using the reference image to obtain single-wavelength actual images, as shown in Fig. 10(e).
Fig. 10. Physical and spectral images. (a) Physical images; (b) Physical spectrograms. (c) Single wavelength calibration images; (d) Calibration reference image; (e) Single wavelength physical images; (f) Spectral reconstruction images.
Given the device’s varied sensitivity to different wavelengths, direct pixel value comparisons are meaningless. Normalize the pixel values of the spectral images against the non-zero pixel values of the single-wavelength calibration images to determine the relative brightness at each wavelength. Along the wavelength axis, vertically analyze the relative brightness of each pixel across different wavelengths to identify the single-wavelength real images with the highest relative brightness for each segment of the object. Perform horizontal and vertical alignment of the single-wavelength real images using projection transformation, ensuring that points in the same row are positioned at the same horizontal location and points in the same column are positioned at the same vertical location within the image. Then, iterate through the images and remove rows and columns with no content. For single-wavelength real images, spectral information is distributed in a grid-like pattern in the form of scattered points. Therefore, many rows and columns in the images are completely dark, containing no spectral information. These empty rows and columns hinder the acquisition of spectral image shape information and disrupt the spectral reconstruction results. Hence, removing empty rows and columns is crucial. After performing morphological closing on the single-wavelength real images containing only a partial object, apply Gaussian blur to extract the corresponding portion of the image. Morphological closing creates larger connected regions to fill small holes or gaps in the image, resulting in smoother and more continuous object boundaries. Gaussian blur, compared to other blur filters, prioritizes edge preservation. During the blurring process, Gaussian blur adjusts the weights based on the similarity between pixels and their surrounding pixels, thus preserving the sharpness of edges and avoiding edge blurring. Merge the obtained images to get the spectral reconstruction result image. The results as illustrated in Fig. 10(f). Since the spectral information in the single-wavelength real image is controlled by the number of micro-lens array elements, resulting in sparse bright spots and impacting the imaging quality. However, the main spectral information in the spectral reconstruction result images remains intact with clear contours. Thus, lower imaging quality has no bearing on identification results.
The full width at half maximum (FWHM) represents the distance between points on either side of the peak, where the intensity is half of the peak intensity. It is commonly used to assess the spectral resolution of a spectrometer. For any calibration image at a specific wavelength, a randomly selected bright point was chosen, and a Gaussian fit was applied to its pixel position and intensity. The experiment utilized CCD as the imaging device to capture spectral images. Due to material properties and manufacturing processes inherent to photodetector materials, CCDs exhibit variations in spectral responsivity. While the energy distribution of the laser generator remains relatively uniform across different wavelengths, discrepancies in CCD sensitivity to various wavelengths result in inconsistent brightness of light rays imaging with the same energy. Consequently, peaks of different wavelengths differ in the acquired images. To mitigate the adverse effects of differentiated spectral responsivity, spectral data normalization was performed. Here, 0 denotes no illumination, 1 represents the maximum light intensity, and 0.5 indicates the position halfway between the peak's height, known as the half-maximum position. A horizontal line is drawn at 0.5 to obtain the FWHM. The results are shown in Fig. 11. Figure 11 indicates that the peak values of four different wavelengths vary, yet their FWHM are approximately equal, each around 6. Therefore, the actual spectral resolution of the experimental setup is approximately 6 nm, slightly below the ideal condition but still better than PMVIS.
5. Conclusion
In this paper, we improved the image slicing module and designed DMAIS, aiming for high transmittance and low spectral bending. DMAIS consists of three main components: PL, SCM, and TL. The core component of DMAIS is the SCM, which is based on a dual micro-lens array. PL is positioned at the front end of SCM to capture spatial information, while TL is placed at the rear end to obtain spectral images. The dual micro-lens array of SCM replaces the mask for segmenting and collimating the light into multiple parallel micro-beams with a consistent exit direction. A planar aperture is located at the rear focal length of the first micro-lens array and the front focal length of the second micro-lens array, serving solely to filter stray light and suppress spectral line distortion. The prism is the sole dispersive element in this design.
ZEMAX simulations indicate that compared to IFIS, DMAIS achieves a 109.2% increase in energy transmittance and a 32.9% reduction in spectral bending. Under ideal conditions, distortion and spectral curvature are both suppressed to below 4.3% and 22% of per pixel, respectively, with a spectral resolution of 5 nm. Subsequently, an experimental setup was constructed based on the simulation results, and calibration images and actual spectral images were captured using this setup. The calibration images consist of a set of single-wavelength images spaced 5 nm apart in the wavelength range of $420\sim 700\textrm{nm}$, totaling 56 images. A reference image was generated by superimposing these images using the wavelength scanning method. By deconstructing actual spectral images with reference to the reference image, single-wavelength images at specific wavelengths were obtained. Guided by the theoretical FWHM, the actual spectral resolution of the experimental setup was determined to be 6 nm based on the calibration images.
IMS operates within a narrow wavelength range and has a limited field of view, often necessitating more complex hardware and software systems, resulting in high equipment costs. In contrast to IMS, DMAIS covers the entire visible light spectrum with a $48.8^\circ $ imaging field of view, enabling wide-field imaging. By utilizing SCM to segment the field of view, DMAIS achieves a simple structure. Pursuing higher spectral resolution is one of the research directions for imaging spectrometers. To achieve this goal, IFIS uses masks to reduce the size of the light entering the dispersive element and control the length of the spectral lines. However, the use of mask obstructs light propagation, severely affecting the light flux and causing spectral bending. DMAIS does not prioritize the improvement of spectral resolution. Instead, it addresses issues such as low energy transmittance and spectral bending by improving the field segmentation module of the imaging spectrometer. DMAIS employs a microlens array to focus sub-lenses instead of using small apertures in mask for sampling, thereby enhancing the optical system's energy transmittance. The increase in energy transmittance inevitably leads to an increase in the size of the incident light on the dispersive element and the length of the spectral lines. To prevent spectral overlap, reducing the dispersion rate of the dispersive element is inevitable. Therefore, the spectral resolution of DMAIS is slightly lower than that of IFIS. Figure 5 illustrates the imaging results of IFIS and DMAIS under the same mask aperture. Observation of Fig. 5 reveals that although DMAIS sacrifices some spectral resolution, the improvement in energy transmittance enhances the imaging effect of the spectral images, increases the amount of spectral information acquired, and alleviates the pressure of subsequent image processing.
Funding
Shanxi Provincial Key Research and Development Project (202102150401007).
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
Supplemental document
See Supplement 1 for supporting content.
References
1. T.U. Kampe, “The compact hyperspectral prism spectrometer: advanced imaging spectrometer for sustainable land imaging,” Remote Sensing and Modeling of Ecosystems for Sustainability XV 10767, 8–59 (2018). [CrossRef]
2. Y. K. Chung, C. M. Jo, S. K. Kim, et al., “Simulation for small lamellar grating FTIR spectrometer for passive remote sensing,” J. Opt. Soc. Korea 20(6), 669–677 (2016). [CrossRef]
3. M. Bhatt, U. Mall, C. Woehler, et al., “A comparative study of iron abundance estimation methods: Application to the western nearside of the Moon,” Icarus 248, 72–88 (2015). [CrossRef]
4. S. Kumar, P. Bhatt, B. K. Singh, et al., “An experimental set up for studying the fusion edge plasma facing materials using TOF mass spectrometry,” Int. J. Mass Spectrom. 385, 32–41 (2015). [CrossRef]
5. Z. Zhang, B. Hu, Q. Yin, et al., “Design of shortwave infrared imaging spectrometer system based on CDP,” Opt. Express 23(23), 29758–29763 (2015). [CrossRef]
6. Y. Zhang, C. Wang, H. Chen, et al., “Optical design of a crossed double-grating spectrometer for extracting pure rotational Raman lines of N2,” Optics and Lasers in Engineering 134, 106291 (2020). [CrossRef]
7. L. Feng, “Design of a nonlinearity-corrected spectrometer based on curved prism,” Opt. Commun. 434, 239–245 (2019). [CrossRef]
8. Gebirie Yizengaw Belay, “Miniaturized broadband spectrometer based on a three-segment diffraction grating for spectral tissue sensing,” Optics and Lasers in Engineering 134, 106157 (2020). [CrossRef]
9. Y. Pang, Y. Zhang, H. Yang, et al., “Compact high-resolution spectrometer using two plane gratings with triple dispersion,” Opt. Express 26(5), 6382–6391 (2018). [CrossRef]
10. Y. Shi, L. Zhang, F. Chen, et al., “Progress in spectral imaging technology of digital micromirror devices,” Optics and Precision Engineering 31(21), 3096–3110 (2023). [CrossRef]
11. L. Qian, L. Xiang, Q. Lu, et al., “Analysis and simulation of effect of spectral line bending on computational imaging spectrometry,” Acta Photonica Sinica 42(8), 897–901 (2013). [CrossRef]
12. Z. P. Zheng, B. Qiu, S. Y. Wei, et al., “Two-Dimensional Fiber Spectral Bending Correction Based on Curve Distance Method,” Opt. Express 30(2), 2230 (2022). [CrossRef]
13. X. Zhang, X. Li, and X. Tang, “Study on a plane grating spectral imaging system with smile and keystone eliminated,” J. Opt. Technol. 89(5), 262–268 (2022). [CrossRef]
14. X. Cao, H. Du, X. Tong, et al., “A prism-mask system for multispectral video acquisition,” IEEE Trans. Pattern Anal. Mach. Intell. 33(12), 2423–2435 (2011). [CrossRef]
15. X. Zhang, J. Wang, J. Zhang, et al., “A design method for direct vision coaxial linear dispersion spectrometers,” Opt. Express 30(21), 38266–38283 (2022). [CrossRef]
16. F. Yu, Q. Liu, and Y. Liu, “Design and deformation impact analysis of arrays in filter array multispectral cameras,” Acta Photonica Sinica 46(4), 0411001 (2014).
17. X. Wu, X. Yang, and H. Yu, “Optical design of offner code aperture spectral imaging system based on convex grating,” Opt. Express 29(24), 39271 (2021). [CrossRef]
18. F. Zhang, J Gao, N Wang, et al., “LWIR imaging spectrometer employing a variable gap Fabry-Perot interferometer,” Infrared and Laser Engineering 46(3), 0318001 (2017).
19. B. Liu, Y. Liu, C. Li, et al., “Design of optical system for miniature visible/near-infrared real-time imaging spectrometer,” Acta Opt. Sin. 35(6), 0630002 (2015). [CrossRef]
20. P. Li, Y. Hu, X. Zhong, et al., “Light and small mask spectral optical system,” Infrared and Laser Engineering 49(4), 414001 (2020). [CrossRef]
21. Michal E. Pawlowski, Jason G. Dwight, Thuc-Uyen Nguyen, et al., “High performance image mapping spectrometer (IMS) for snapshot hyperspectral imaging applications,” Opt. Express 27(2), 1597–1612 (2019). [CrossRef]
22. C. Zeng, Y. Han, B. Liu, et al., “Optical design of a high-resolution spectrometer with a wide field of view,” Optics and Lasers in Engineering 140, 106547 (2021). [CrossRef]
