Hyperspectral imaging system and use method thereof

Abstract

A hyperspectral imaging system is characterized by comprising an imaging unit, an optical fiber scanning device, an optical fiber spectrometer, a data processing unit and a control unit, the imaging unit comprises an imaging lens and a micro-lens array, the imaging lens images a target object onto the micro-lens array, the optical fiber scanning device scans an optical signal processed by the imaging unit, and the optical fiber spectrometer performs spectrum separation on light with different wavelengths in the scanned optical signal to form spectrum data; and the data processing unit processes the spectral data and reconstructs a hyperspectral image of the target object by using the spectral data. And only an important area is scanned point by point in an optical fiber scanning mode, and hyperspectral data acquisition of a local interested target is concerned, so that unnecessary overhead of full-frame data acquisition is greatly reduced. For an application scene needing quick or real-time monitoring, the directional scanning mode greatly shortens the required integral acquisition time, and reduces the overall time delay of the system.

Description

Technical Field

[0001] The present invention relates to the field of spectral imaging technology, and particularly to a hyperspectral imaging system based on a microlens array and fiber optic scanning. Background Art

[0002] Spectral imaging technology, also known as hyperspectral imaging (HSI) technology, is a new detection technology that combines traditional two-dimensional imaging technology and spectroscopy technology. It can obtain the two-dimensional spatial information of the target object while detecting the spectral information of each spatial pixel point, and finally form a three-dimensional data cube containing two-dimensional spatial information and one-dimensional spectral information. This unique data acquisition method enables hyperspectral imaging technology to identify and distinguish substances with similar morphological characteristics but different spectral characteristics, greatly enhancing the human's cognitive ability of the material world. Therefore, hyperspectral imaging technology shows broad application prospects in many fields such as military reconnaissance, target recognition, camouflage detection, resource exploration, environmental protection, agricultural production, food safety, biomedicine, astronomical observation, etc.

[0003] Traditional hyperspectral imaging technologies mainly include two categories: scanning type and non-scanning type. Non-scanning hyperspectral imagers usually adopt a staring imaging method and can obtain hyperspectral images of the entire field of view without scanning. However, they usually require complex optical systems and algorithms, with high costs, and limited spatial resolution or spectral resolution. Scanning hyperspectral imagers usually adopt push-broom or swing-scanning methods. They obtain one-dimensional spatial information and one-dimensional spectral information through a slit spectrometer, and then obtain the other dimension of spatial information through the movement of the platform or the swing of the scanning mirror, thereby constructing a three-dimensional data cube. The slit design allows only a part of the light to enter the spectrometer, thus limiting the light energy collection efficiency. This is particularly obvious under low-light conditions and may result in a low signal-to-noise ratio of the system.

[0004] Although the size of the slit is similar to that of a single microlens, its physical edge will produce diffraction and scattering effects on the passing light. After passing through the slit, the light beam will spread and form a diffraction pattern. Such diffraction expansion will cause part of the light energy to spread outside the expected sampling area, thereby increasing the background noise and having an adverse effect on the spectral resolution of the system.

[0005] Due to the diffraction and scattering effects, the light spot collected by the slit is prone to overlap with adjacent areas during transmission. Adjacent pixels may receive partially overlapping light energy, resulting in crosstalk. This crosstalk will cause the spectral information at different spatial positions to be mixed, thereby reducing the accuracy and resolution of the final hyperspectral image.

[0006] More importantly, whether it is a traditional scanning or non-scanning hyperspectral imaging system, hyperspectral data is usually collected for the entire imaging scene. This approach is not necessary in many application scenarios and instead leads to problems such as data redundancy, long processing times, and low efficiency. Especially in applications with high requirements for frame rate or latency, such as real-time monitoring, tracking and analysis of high-speed moving objects, etc., full-frame hyperspectral imaging often fails to meet the requirements, and the temporal consistency issues caused by long scans also affect the accuracy of the data. Summary of the Invention

[0007] In view of the problems in the existing hyperspectral imaging technology, such as data redundancy, low acquisition efficiency, and difficulty in balancing hyperspectral resolution and high frame rate during full-frame acquisition, the present invention proposes a novel hyperspectral imaging system based on a microlens array and fiber optic scanning. It only performs point-by-point scanning on important regions and focuses on the acquisition of hyperspectral data for locally interested targets, greatly reducing the unnecessary overhead of full-frame data acquisition.

[0008] To solve the above technical problems, the technical solution adopted by the present invention is as follows: A hyperspectral imaging system, characterized in that it includes: an imaging unit, a fiber optic scanning device, a fiber optic spectrometer, a data processing unit, and a control unit; the imaging unit is used to obtain an image of a target object; the fiber optic scanning device is used to scan the optical signal processed by the imaging unit; the fiber optic spectrometer is used to receive the optical signal scanned by the fiber optic scanning device and perform spectral separation on the light of different wavelengths in the optical signal to form spectral data; the data processing unit is used to receive the spectral data output by the fiber optic spectrometer, process the spectral data, and reconstruct a hyperspectral image of the target object using the spectral data; the imaging unit includes an imaging lens and a microlens array, and the imaging lens is used to image the target object onto the microlens array.

[0009] The hyperspectral imaging system proposed by the present invention can well achieve flexible, fast, and efficient regional hyperspectral imaging by introducing a fiber optic scanning device and a two-dimensional imaging sensor, combined with a microlens array, and at the same time ensure high spatial resolution and spectral resolution. Using a microlens array to replace the slit design in the traditional scheme has obvious advantages: the microlens array can evenly divide the entire imaging area into multiple small focal regions and focus the light energy more efficiently onto each region. This design not only improves the light energy utilization rate but also reduces the risk of scattering and crosstalk caused by edge effects. By optimizing the scanning strategy and adopting closed-loop control, the system can obtain high-quality spectral data in a shorter time, thereby ensuring a high signal-to-noise ratio and spatial resolution.

[0010] Further, the fiber optic scanning device drives the optical fiber to perform point-by-point scanning on the focal planes of the M×N sub-image lenses generated by the microlens array, where M≥50 and N≥50; or only performs point-by-point scanning on the focal planes of the m×n sub-image lenses within a local range, where m≤M and n≤N.

[0011] When traditional hyperspectral imagers acquire two-dimensional data of spatial and spectral information, a slit needs to be set at the front end of the spectrometer to limit the light beam. However, the slit itself will generate diffraction and scattering due to its physical edges, causing some light energy to diffract into adjacent pixels, resulting in crosstalk and reducing the signal-to-noise ratio and spectral resolution. In addition, the size of the slit will also limit the light flux of the system, making it face serious signal insufficiency problems in weak light or high-resolution applications. In contrast, the microlens array can evenly divide the imaging plane into several tiny focal points, significantly improving the light collection efficiency per unit area while reducing edge scattering and energy loss caused by the slit effect. Each microlens focal point corresponds to an independent spatial field of view, with almost no diffraction and leakage caused by the physical slit structure, and the crosstalk between spatial pixels is significantly reduced. This not only helps to improve the signal-to-noise ratio and spectral resolution of the final image, but also facilitates the efficient acquisition of some key regions.

[0012] Further, the fiber optic scanning device driving optical fiber is driven by a fiber two-dimensional planar movement mechanism.

[0013] Further, the fiber two-dimensional planar movement mechanism is a piezoelectric actuator, which uses piezoelectric ceramic materials to drive the optical fiber to move in the two-dimensional plane, and the scanning frequency is from 1 kHz to 20 kHz, and the scanning trajectory error is ≤2% through closed-loop feedback control.

[0014] After moving the optical fiber to the target sub-image focal point, the fiber optic spectrometer completes the acquisition of spectral signals within the integration time; then the optical fiber moves to the next focal point, and sequentially scans the selected M×N or m×n sub-image regions. Users can perform point-by-point scanning only on the regions of interest according to different application requirements, reducing unnecessary data acquisition and significantly improving the imaging speed and data utilization rate.

[0015] Through the fiber two-dimensional planar movement mechanism and position sensing feedback, the fiber optic scanning device can keep the movement accuracy and trajectory of the optical fiber within a small error range in the two-dimensional plane, improving the sampling consistency and the reliability of hyperspectral data. The scanning frequency of the fiber optic scanning device can reach the kHz level to meet the needs of high-speed moving targets or real-time detection; for the case of only local area scanning, the sampling rate of a single point can be further increased.

[0016] Further, the residence time of the optical fiber at each microlens focal point satisfies , where is the integration time of the fiber optic spectrometer.

[0017] Further, a beam splitting system is provided between the imaging lens and the microlens array. The beam splitting system splits the beam from the imaging unit into a first beam and a second beam; the first beam enters the microlens array, and the second beam enters the two-dimensional imaging sensor; the optical path difference between the first beam and the second beam .

[0018] The added beam splitting system in this system splits the beam from the imaging unit into two paths: one path enters the microlens array and is finally sent to the fiber optic spectrometer for spectral acquisition, and the other path directly enters the two-dimensional imaging sensor to achieve conventional two-dimensional imaging. Through this beam splitting system, hyperspectral images and two-dimensional reference images of the target scene can be obtained simultaneously, complementing each other, facilitating subsequent data fusion or positioning, and thus further improving the ability to identify and analyze the target. The system can obtain two-dimensional image information consistent with the target in real time through the two-dimensional sensor while acquiring hyperspectral data. The image frame rate of the two-dimensional sensor is usually higher than that of the fiber scanning method, which can monitor the target scene in real time, lock the ROI, and track moving targets; the hyperspectral image and the two-dimensional image have a good correspondence in time and space, which is conducive to combining spectral information with visual features for target recognition, classification, or image analysis.

[0019] Further, the two-dimensional imaging sensor is a CMOS, CCD, or InGaAs sensor; the CMOS / CCD sensor operates in the range of 200 - 1100 nm, and the InGaAs sensor operates in the range of 900 - 2500 nm.

[0020] Further, the control unit is an FPGA main control, with a clock accuracy ≤ 10 ns and a cumulative synchronization error , where is the total number of scan points.

[0021] Further, the imaging module in the fiber optic spectrometer for receiving the light from the beam splitting prism is a silicon photodiode, or an avalanche diode, or a TDI CCD.

[0022] A method for using a hyperspectral imaging system, the key points being that the hyperspectral imaging system includes an imaging lens, a microlens array, an optical fiber scanning device, an optical fiber spectrometer, a data processing unit, a spectroscopic system, a two-dimensional imaging sensor, and a control unit; the imaging lens introduces the light of the object to be measured into the imaging system; the microlens array divides the light of the object to be measured into M×N sub-images; the optical fiber scanning device guides the light of a single lens divided by the microlens array into the optical fiber spectrometer; the optical fiber spectrometer is used to receive the optical signal scanned by the optical fiber scanning device and perform spectral separation on the light of different wavelengths in the optical signal to form spectral data; the data processing unit is used to receive the spectral data output by the optical fiber spectrometer, process the spectral data, and reconstruct the hyperspectral image of the target object using the spectral data; the spectroscopic system divides the light beam from the imaging lens into a first light beam and a second light beam; the control unit is used to drive the optical fiber scanning device;

[0023] The usage steps are as follows:

[0024] a) The imaging lens images the target object onto the spectroscopic system;

[0025] b) The spectroscopic system divides the incident light into two paths: the first light beam enters the microlens array to form M×N sub-focal points; the second light beam enters the two-dimensional imaging sensor to generate a reference image;

[0026] c) The control unit selects a scanning path according to the region of interest (ROI) and traverses m×n focal points in a preset order, where m≤M and n≤N;

[0027] d) Drive the optical fiber two-dimensional planar movement mechanism for closed-loop control scanning;

[0028] e) When the optical fiber reaches each focal point of the target: trigger the optical fiber spectrometer to perform integration; synchronously record the current focal point coordinates ; the optical fiber spectrometer transmits the spectral data stream with spatial encoding to the data processing unit;

[0029] f) The data processing unit reconstructs the three-dimensional data cube and stores the original spectral data and the reconstructed data cube.

[0030] The advantages of the present invention are as follows: By means of optical fiber scanning, only important regions are scanned point by point, focusing on the acquisition of hyperspectral data of locally interested targets, greatly reducing the unnecessary overhead of full-frame data acquisition. For application scenarios that require fast or real-time monitoring, this directional scanning method significantly shortens the required integration acquisition time and reduces the overall system delay.

[0031] Using a microlens array to focus the light beam, different from the traditional way of using a slit to define the light beam, it avoids the energy loss and pixel crosstalk caused by diffraction and scattering at the slit edge; at the same time, the microlens array concentrates more light energy at the focal position, and the light flux utilization rate of the system is relatively high, significantly improving the acquisition quality in low-light environments.

[0032] By adding a beam splitting system in front of the microlens array and equipping it with a two-dimensional imaging sensor, a conventional two-dimensional image of the target scene can be obtained in real time while performing hyperspectral acquisition, meeting the requirements of observing, positioning, identifying, and tracking the target. The two-way data has a good corresponding relationship in space and time, creating better technical conditions for multi-source information fusion.

[0033] Using a fiber two-dimensional planar movement mechanism to drive the fiber to perform micron-level displacement in the two-dimensional plane can ensure high precision when the light spot aligns with each microlens focus; at the same time, through closed-loop feedback, the error of the scanning trajectory can be maintained at an extremely low level. The control unit can arrange the spectral integration and data reading processes with a clock accuracy at the nanosecond level, keeping the fiber position and spectral acquisition strictly synchronized and reducing the dynamic mismatch problem caused by time delay.

[0034] By introducing a fiber optic scanning device between the imaging unit and the fiber optic spectrometer and cooperating with key components such as a two-dimensional imaging sensor and a microlens array, fast, flexible, and efficient spectral acquisition of the region of interest is achieved, significantly improving the response speed and data quality of the system in specific application scenarios. At the same time, under the design of the microlens array, phenomena such as diffraction, scattering, and crosstalk existing in traditional slit acquisition are effectively suppressed, thereby further improving the signal-to-noise ratio and light energy utilization rate of imaging while ensuring high spectral resolution. Description of the Drawings

[0035] Figure 1 Schematic structural diagram of a hyperspectral imaging system based on a microlens array and fiber optic scanning

[0036] Figure 2 Schematic structural diagram of a hyperspectral imaging system with a beam splitting system

[0037] Figure 3 Flowchart of the usage method of the hyperspectral imaging system

[0038] Figure 4 Corresponding schematic diagram of a two-dimensional image sensor, a microlens array, and an ROI region

[0039] Wherein: 1 imaging unit, 11 imaging lens, 12 microlens array, 2 fiber optic scanning device, 21 optical fiber, 22 two-dimensional planar movement mechanism of optical fiber, 3 fiber optic spectrometer, 4 data processing unit, 5 beam splitting system, 6 two-dimensional imaging sensor, 7 control unit, 401 frame of two-dimensional image sensor, 402 frame of microlens array, 403 frame of ROI region. Detailed implementation mode

[0040] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0041] The present invention is a hyperspectral imaging system, which includes: an imaging unit 1, a fiber optic scanning device 2, a fiber optic spectrometer 3, a data processing unit 4 and a control unit 7; the imaging unit 1 is used to obtain an image of a target object, and it includes an imaging lens 11 and a microlens array 12, and the imaging lens 11 is used to image the target object onto the microlens array 12; the fiber optic scanning device 2 is used to scan the optical signal processed by the imaging unit 1, and the fiber optic scanning device 2 drives the optical fiber 21 to perform point-by-point scanning on the focal plane of M×N sub-image lenses generated by the microlens array, where M≥50 and N≥50; or only perform point-by-point scanning on the focal plane of m×n sub-image lenses within a local range, where m≤M and n≤N; the fiber optic scanning device 2 drives the optical fiber by using a two-dimensional planar movement mechanism 22 of the optical fiber; the two-dimensional planar movement mechanism 22 of the optical fiber is a piezoelectric actuator, and a piezoelectric ceramic material is used to drive the optical fiber 21 to move in a two-dimensional plane, and the scanning frequency is 1 kHz to 20 kHz, and the scanning trajectory error is ≤2% through closed-loop feedback control. The residence time of the optical fiber 21 at each microlens focus satisfies , after moving the optical fiber 21 to the focus of the target sub-image, the fiber optic spectrometer 3 completes the acquisition of the spectral signal within the integration time; then the optical fiber 21 moves to the next focus, and sequentially scans the selected sub-image region of m×n (where m≤M and n≤N). Users can perform point-by-point scanning only on the region of interest (ROI) according to different application requirements, reduce unnecessary data acquisition, and significantly improve the imaging speed and data utilization rate, wherein is the integration time of the fiber optic spectrometer 3. The fiber optic spectrometer 3 is used to receive the optical signal scanned by the fiber optic scanning device 2, and perform spectral separation on the light of different wavelengths in the optical signal to form spectral data. The imaging module in the fiber optic spectrometer 3 for receiving the light of the beam splitter prism is a silicon photodiode, or an avalanche diode (APD), or a TDI CCD. After receiving the spectral data output by the fiber optic spectrometer 3, the data processing unit 4 combines the spectral information corresponding to the focal points of each microlens with the spatial position information through algorithms such as interpolation, registration, and spectral correction to reconstruct the hyperspectral cube of the target scene. For a scene that only scans a partial area, only the three-dimensional data of the ROI part can be reconstructed, greatly reducing the data scale and processing delay. In addition, if there is a two-dimensional image as an auxiliary, more accurate correction and fusion analysis can be performed on the area of interest.

[0042] The control unit 7 is an FPGA main control, with a clock accuracy ≤ 10 ns and a cumulative synchronization error , where is the total number of scan points. Its main functions include:

[0043] (1) Precise synchronization: Ensure the movement of the fiber optic scanning device 2 and the integration time of the fiber optic spectrometer 3, and ensure that there is no timing chaos in the spectral acquisition process before or after the optical fiber 21 reaches the focus of the next sub-image.

[0044] (2) Trajectory planning: According to the shape and size of the ROI given by the user, output the corresponding control signal to make the optical fiber scan in the two-dimensional plane in the specified order, path, and speed.

[0045] (3) Closed-loop feedback: Receive the feedback information of the fiber optic two-dimensional plane moving mechanism 22 or other position sensors in real time, correct the possible deviations during the movement of the optical fiber, and ensure the scanning accuracy.

[0046] (4) Data reading and preprocessing: After the fiber optic spectrometer 3 completes the integration, read the spectral data in time and perform preliminary correction or compression for subsequent reconstruction by the data processing unit 4.

[0047] The present invention only performs point-by-point scanning on important areas through the fiber optic scanning method, focuses on the acquisition of hyperspectral data of local target areas of interest, and greatly reduces the unnecessary overhead of full-frame data acquisition. For application scenarios that require fast or real-time monitoring, this directional scanning method greatly shortens the required integration acquisition time and reduces the overall system delay.

[0048] Embodiment 1: Hyperspectral imaging system based on a microlens array and fiber optic scanning

[0049] As Figure 1As shown in the figure, the system consists of an imaging unit 1, an optical fiber scanning device 2, an optical fiber spectrometer 3, a data processing unit 4, and an FPGA control unit 7. The imaging unit 1 uses an imaging lens 11 with a focal length of 50 mm to image the target object onto the microlens array 12. The array consists of 100×100 (M = N = 100) square-arranged microlenses. The single-lens aperture is 200 μm and the focal length is 800 μm, which can divide the incident light into 10,000 independent focused spots.

[0050] The optical fiber scanning device 2 adopts the optical fiber two-dimensional planar driving scheme described in the invention patent CN108810498A. It includes a multimode optical fiber 21 with a core diameter of 9 μm and an optical fiber two-dimensional planar moving mechanism 22 - a piezoelectric ceramic actuator, which realizes two-dimensional planar scanning through closed-loop feedback control. The scanning frequency is 15 kHz, and the trajectory error ≤ 1.8%. The optical fiber 21 scans point by point in a serpentine path at the focal plane of the microlens, and the dwell time at each focus = 66 μs (meeting the dwell time , where the integration time of the optical fiber spectrometer = 60 μs). The control unit 7 uses a Xilinx Kintex-7 FPGA with a clock accuracy of 8 ns.

[0051] The optical fiber spectrometer 3 adopts a Hamamatsu S11866-128-02 type silicon photodiode module, with a spectral range of 200 - 1000 nm and a resolution of 7 nm. The data processing unit 4 realizes the mapping of spectral data and spatial coordinates through the LabVIEW platform, and reconstructs a 100×100×114 (space × spectrum) data cube.

[0052] The piezoelectric ceramic actuator drives the optical fiber to move to the (i, j) focus (as shown in the figure, moving from the lower solid line area to the upper dotted line), which takes = 50 μs; start spectral integration in the last 1 / 3 period (about 17 μs) after the moving stage; after the integration is completed, read out the spectral data through the LVDS interface within the last 6 μs; loop until the full array scan is completed, with a total time consumption of about 1.1 seconds.

[0053] The working process of the system is as follows:

[0054] a) The imaging lens 11 images the target object onto the microlens array 12, and the microlens array 12 divides the light into 100×100 sub-foci;

[0055] b) The control unit 7 selects and generates a scanning path according to the region of interest (ROI), and traverses m×n foci in a preset order, where m ≤ 100 and n ≤ 100;

[0056] c) Drive the optical fiber two-dimensional planar moving mechanism 22 to perform closed-loop control scanning;

[0057] d) When the optical fiber 21 reaches each target focus: Trigger the optical fiber spectrometer 3 to perform integration; synchronously record the current focus coordinates ; The optical fiber spectrometer 3 transmits the spectral data stream with spatial encoding to the data processing unit 4;

[0058] e) The data processing unit 4 reconstructs the three-dimensional data cube and stores the original spectral data and the reconstructed data cube.

[0059] Embodiment 2: Hyperspectral imaging system with a beam splitting system

[0060] As Figure 2 shown, on the basis of Embodiment 1, this embodiment adds a beam splitting system 5 and a two-dimensional imaging sensor 6 to form a more perfect system. The beam splitting system 5 splits the beam from the imaging unit 1 into a first beam and a second beam; the first beam enters the microlens array 12, and the second beam enters the two-dimensional imaging sensor 6; the optical path difference between the first beam and the second beam .

[0061] The added beam splitting system 5 in this system splits the beam from the imaging unit 1 into two paths: one path enters the microlens array 12 and finally is sent to the optical fiber spectrometer 3 for spectral acquisition, and the other path directly enters the two-dimensional imaging sensor 6 (such as a CCD, CMOS or InGaAs device) to achieve conventional two-dimensional imaging. Through this beam splitting system 5, a hyperspectral image and a two-dimensional reference image of the target scene can be obtained simultaneously, complementing each other, facilitating subsequent data fusion or positioning, and thus further enhancing the ability to identify and analyze the target. The correspondence between the frame 401 of the two-dimensional image sensor presented in the data processing unit 4 for the image generated by the two-dimensional imaging sensor 6 and the frame 402 of the microlens array for the spectral data of the microlens array presented in the data processing unit 4 is as Figure 4 shown.

[0062] The system can, while acquiring hyperspectral data, obtain two-dimensional image information consistent with the target in real time through the two-dimensional sensor 6. The image frame rate of the two-dimensional sensor 6 is usually higher than that of the fiber scanning method, and it can monitor the target scene in real time, lock the frame 403 of the ROI area, and track moving targets; the hyperspectral image and the two-dimensional image have a good correspondence in time and space, which is conducive to subsequent combination of spectral information and visual features for target recognition, classification or image analysis. The correspondence between the frame 403 of the ROI area and the frames 401, 402 of the two-dimensional image sensor and the frame of the microlens array is as Figure 4 shown.

[0063] The two-dimensional imaging sensor 6 is a CMOS, CCD, or InGaAs sensor; the CMOS / CCD sensor operates at 200 - 1100 nm, and the InGaAs sensor operates at 900 - 2500 nm.

[0064] Specifically: The beam splitting system 5 can be implemented using a beam splitting plane mirror or a beam splitting prism, and is arranged between the imaging lens 11 and the microlens array 12 to split the beam after the imaging lens 11 into a first beam and a second beam. The first beam enters the microlens array 12, and the second beam enters the two-dimensional imaging sensor 6. The two-dimensional imaging sensor 6 can use a CMOS, CCD, or InGaAs imaging sensor to obtain a two-dimensional grayscale image of the target object. This two-dimensional grayscale image can be used for preview and navigation. The user can observe this image and select a region of interest (ROI) on the software interface of the data processing unit 4. Then, the control unit 7 controls the fiber optic scanning device 2 to scan only this ROI according to the user's selection, thereby realizing ROI hyperspectral imaging. This method avoids hyperspectral scanning of the entire imaging screen, greatly improves the imaging speed and efficiency, reduces data redundancy, and is particularly suitable for application scenarios with high requirements for frame rate or latency.

[0065] The FPGA control unit 7 plays a coordinating and controlling role in the entire system. The control unit 7 communicates with the imaging unit 1, the fiber optic scanning device 2, the fiber optic spectrometer 3, the data processing unit 4, and the two-dimensional imaging sensor 6, coordinates the working timings and parameters of each unit, and enables them to work synchronously. For example, the control unit 7 can control the autofocus of the imaging lens 11, control the scanning trajectory, scanning speed, start and stop of the fiber optic scanning device 2, and control the exposure time, trigger mode, etc. of the fiber optic spectrometer 3. In Embodiment 3, the control unit 7 can also receive the user's ROI selection instruction from the data processing unit 4 and control the scanning range of the fiber optic scanning device 2 according to this instruction to realize ROI hyperspectral imaging.

[0066] As Figure 3 shown, the system working process is as follows:

[0067] a) The imaging lens 11 images the target object onto the beam splitting system 5;

[0068] b) The beam splitting system 5 divides the incident light into two paths: The first beam enters the microlens array 12 to form M×N sub-foci; the second beam enters the two-dimensional imaging sensor 6 to generate a reference image;

[0069] c) The control unit 7 generates a scanning path according to the selection of the region of interest (ROI), and traverses m×n foci in a preset order, where m ≤ M, n ≤ N, such as Figure 3 the frame 403 of the ROI region in

[0070] d) Drive the two-dimensional planar movement mechanism 22 of the optical fiber to perform closed-loop control scanning;

[0071] e) When the optical fiber 21 reaches each target focus: Trigger the optical fiber spectrometer 3 to integrate; Synchronously record the current focus coordinates ; The optical fiber spectrometer 3 transmits the spectral data stream with spatial encoding to the data processing unit 4;

[0072] f) The data processing unit 4 reconstructs the three-dimensional data cube and stores the original spectral data and the reconstructed data cube.

[0073] The above two embodiments have described in detail the specific implementation manners of the hyperspectral imaging system proposed by the present invention. By combining a spatial light modulator, an optical fiber scanning device, an optical fiber spectrometer, and a two-dimensional imaging sensor, the system realizes flexible and efficient ROI hyperspectral imaging and has high spatial resolution and high spectral resolution. The system has a compact structure, is easy to integrate and miniaturize, and has broad application prospects in the fields of remote sensing, medical diagnosis, food safety, environmental monitoring, etc.

[0074] It should be noted that the above embodiments are only some preferred implementation manners of the present invention. Those skilled in the art can replace or improve each component according to the above embodiments. For example, other types of scanning mechanisms, optical fiber spectrometers, spatial light modulators, etc. can be adopted. These improvements and deformations should fall within the protection scope of the present invention.

Claims

1. A hyperspectral imaging system, characterized in that: It includes: An imaging unit (1), an optical fiber scanning device (2), an optical fiber spectrometer (3), a data processing unit (4) and a control unit (7); the imaging unit (1) is used to acquire an image of a target object; the optical fiber scanning device (2) is used to scan the optical signal processed by the imaging unit (1); The optical fiber spectrometer (3) is used to receive the optical signal scanned by the optical fiber scanning device (2), and to perform spectral separation on light of different wavelengths in the optical signal to form spectral data; the data processing unit (4) is used to receive the spectral data output by the optical fiber spectrometer (3), to process the spectral data, and to reconstruct a hyperspectral image of the target object using the spectral data; the imaging unit (1) comprises an imaging lens (11) and a microlens array (12), and the imaging lens (11) is used to image the target object onto the microlens array (12).

2. A hyperspectral imaging system according to claim 1, characterized in that: The optical fiber scanning device (2) drives the optical fiber (21) to perform point-by-point scanning on the focal planes of the M×N sub-image lenses generated by the microlens array, where M≥50 and N≥50; or performs point-by-point scanning on the focal planes of the m×n sub-image lenses only within a local range, where m≤M and n≤N.

3. A hyperspectral imaging system according to claim 2, characterized in that: The optical fiber scanning device (2) drives the optical fiber by using an optical fiber two-dimensional plane moving mechanism (22).

4. A hyperspectral imaging system according to claim 3, characterized in that: The optical fiber two-dimensional plane moving mechanism (22) is a piezoelectric actuator, which uses piezoelectric ceramic material to drive the optical fiber (21) to move in the two-dimensional plane, and the scanning frequency The frequency range is 1kHz to 20kHz, and the scanning trajectory error is ≤2% through closed-loop feedback control.

5. A hyperspectral imaging system according to claim 1, 2 or 4, characterized in that: The optical fiber (21) stays at each microlens focus for a certain period of time. satisfy ,in is the integration time of the fiber optic spectrometer (3).

6. A hyperspectral imaging system according to claim 1, characterized in that: A light splitting system (5) is arranged between the imaging lens (11) and the microlens array (12), and the light splitting system (5) splits the light beam from the imaging unit (1) into a first light beam and a second light beam; the first light beam enters the microlens array (12), and the second light beam enters the two-dimensional imaging sensor (6); the optical path difference between the first light beam and the second light beam is .

7. A hyperspectral imaging system according to claim 6, characterized in that: The two-dimensional imaging sensor (6) is a CMOS, CCD or InGaAs sensor; the CMOS / CCD sensor operates at 200-1100nm, and the InGaAs sensor operates at 900-2500nm.

8. A hyperspectral imaging system according to claim 1, characterized in that: The control unit is FPGA master control, the clock accuracy is ≤10ns, and the synchronization error accumulation ,in is the total scan points.

9. A hyperspectral imaging system according to claim 1, characterized in that: The imaging module in the optical fiber spectrometer (3) for receiving the light from the beam splitter prism is a silicon photodiode, or an avalanche diode, or a TDI CCD.

10. A method for using a hyperspectral imaging system, characterized in that: The hyperspectral imaging system comprises an imaging lens (11), a microlens array (12), an optical fiber scanning device (2), an optical fiber spectrometer (3), a data processing unit (4), a light splitting system (5), a two-dimensional imaging sensor (6) and a control unit (7); the imaging lens (11) introduces light from the object to be measured into the imaging system; the microlens array (12) divides the light from the object to be measured into M×N sub-images; the optical fiber scanning device (2) introduces the light of a single lens divided by the microlens array into the optical fiber spectrometer (3); the optical fiber spectrometer (3) is used to receive the optical signal scanned by the optical fiber scanning device (2), and perform spectral separation on light of different wavelengths in the optical signal to form spectral data; the data processing unit (4) is used to receive the spectral data output by the optical fiber spectrometer (3), process the spectral data, and reconstruct a hyperspectral image of the target object using the spectral data; the light splitting system (5) divides the light beam from the imaging lens (11) into a first light beam and a second light beam; the control unit (7) is used to drive the optical fiber scanning device (2); The steps for use are as follows: a) the imaging lens (11) images the target object into the light splitting system (5); b) the light splitting system (5) splits the incident light into two paths: the first light beam enters the microlens array (12) to form M×N sub-foci; the second light beam enters the two-dimensional imaging sensor (6) to generate a reference image; c) a control unit (7) generates a scanning path according to the region of interest (ROI) selection, and traverses m×n focal points in a preset order, where m≤M, n≤N; d) driving the optical fiber two-dimensional plane moving mechanism (22) to perform closed-loop control scanning; e) When the optical fiber (21) reaches each focus of the target: trigger the optical fiber spectrometer (3) to perform integration; synchronously record the current focus coordinates ; The optical fiber spectrometer (3) transmits the spatially encoded spectral data stream to the data processing unit (4); f) The data processing unit (4) reconstructs a three-dimensional data cube and stores the original spectral data and the reconstructed data cube.

Images