An ultra-high grayscale intelligent hyperspectral imaging method and device

Abstract

The present invention relates to the field of spectral measurement technology, and specifically to an ultra-high grayscale intelligent hyperspectral imaging method and device, wherein the ultra-high grayscale intelligent hyperspectral imaging method includes the following steps: S1: initializing the intelligent hyperspectral imaging device; S2: splitting the captured light signal image into at least two beams according to a preset light intensity distribution ratio before focusing, and focusing them on different slits for shaping, and then pre-processing each split light beam; S3: performing multiple exposure acquisition on several light beams to obtain hyperspectral image data for each light beam; S4: fusing several frames of hyperspectral image data according to anchor points to complete ultra-high grayscale hyperspectral image fusion imaging. The present invention realizes multi-band hyperspectral image imaging with a common imaging window by means of light splitting, greatly expanding the hyperspectral band range, and simultaneously adopts a multiple exposure and fusion method to obtain ultra-high grayscale hyperspectral images.

Description

Technical Field

[0001] The present invention relates to the field of spectral measurement technology, and in particular to an ultra-high grayscale intelligent hyperspectral imaging method and device. Background Art

[0002] Spectral measurement technology is currently being applied in more and more scenarios, and is increasingly shifting from laboratory environments to natural environments. With the expansion of applications, the data range of a single hyperspectral camera is becoming increasingly limited, and the real-time requirements for data processing are becoming increasingly higher. The usual method is to use dual hyperspectral cameras for post-shooting processing, but the shooting time is relatively long, and data verification cannot be analyzed until the shooting is completed. There may be cases where the data does not meet the requirements and needs to be re-collected, resulting in very low test and analysis efficiency. At the same time, the installation positions of existing dual hyperspectral cameras are spaced far apart, which makes it easy for test results to deviate after fusion, and requires a lot of manpower to process the dual hyperspectral camera images. At the same time, the data accuracy of existing hyperspectral data is often limited by the sensor sampling hardware circuit. Objects that are too bright or too dark are prone to appear during image acquisition, resulting in data saturation or insufficient brightness, thereby losing data in some areas, affecting test analysis. It has the disadvantages of high data processing difficulty, complex analysis process, and low test and analysis accuracy. Summary of the Invention

[0003] The technical problem to be solved by the present invention is: the existing spectral measurement method is relatively cumbersome, and the camera position arrangement is relatively inconvenient, which leads to easy deviation of the test results after fusion, which is not conducive to high-precision image acquisition and use. At the same time, the existing hyperspectral image data processing is difficult, puts a lot of pressure on the hardware load, and is difficult to miniaturize and portability.

[0004] To solve the above technical problems, the first aspect of the present invention adopts the following technical solution: an ultra-high grayscale intelligent hyperspectral imaging method, comprising the following steps:

[0005] S1: Initialization of the intelligent hyperspectral imaging device, deploying the intelligent hyperspectral imaging device to align with the measured area;

[0006] S2: Splitting the captured light signal image into at least two beams according to a preset light intensity distribution ratio before focusing, focusing them on different slits for shaping, and preprocessing each split light beam;

[0007] S3: performing multiple exposure acquisition on a plurality of light beams to obtain hyperspectral image data of each light beam;

[0008] S4: Fuse several pieces of hyperspectral image data according to the anchor points so that the pieces of hyperspectral image data completely overlap in space, and complete the hyperspectral image fusion imaging with ultra-high grayscale.

[0009] When the present invention is working, multi-band hyperspectral imaging of a common imaging window is achieved by means of spectroscopic means, without the need for additional camera arrangement steps, and can optimize the image acquisition workflow, reduce the labor intensity of operators, and get rid of the limitations of the sensor sampling hardware circuit, making it convenient for miniaturization and portability. It improves test efficiency while reducing hardware load pressure, greatly expands the hyperspectral band range by improving the data response range and data bit depth, and simultaneously adopts a multi-exposure and fusion method to obtain ultra-high grayscale hyperspectral images, thereby improving the accuracy of object spectral measurement.

[0010] Preferably, in step S2, the captured light signal image is split into two beams according to a preset light intensity distribution ratio before focusing, and the light intensity distribution ratio of the two beams is set to 1:1, 1:1.5 or 1:4, and they are respectively focused on different slits for shaping.

[0011] Preferably, in step S2, when processing each split light beam, the following steps are adopted:

[0012] A1: Collimate the beam after it passes through the slit for shaping;

[0013] A2: Perform diffraction and spectral splitting on the collimated light beam. If complete spectral information is required, proceed to step S3; otherwise, proceed to step A3.

[0014] A3: After filtering the required light beam after the splitting is completed, go to step S3.

[0015] Preferably, in step S3, when performing multiple exposure acquisition on a plurality of light beams to acquire hyperspectral image data of each light beam, the following steps are adopted:

[0016] B1: According to the preset exposure parameter group, expose in sequence from low to high gradient, and obtain several hyperspectral image data of different exposure times for each beam;

[0017] B2: Traverse all spatial points and determine whether there are oversaturated pixels at each spatial point in descending order. If there are oversaturated pixels at the spatial point in the current hyperspectral image data, proceed to step B3. If there are no oversaturated pixels at the spatial point in the current hyperspectral image data, proceed to step B4.

[0018] B3: Determine whether there is an oversaturated pixel at the spatial point in the hyperspectral image data one level lower than the current hyperspectral image data. If the spatial point has an oversaturated pixel in the current hyperspectral image data, proceed to step B3. If the spatial point has no oversaturated pixel in the current hyperspectral image data, proceed to step B4.

[0019] B4: The hyperspectral data of all spatial points of each light beam is recorded as the ratio of the current hyperspectral image data to the corresponding exposure time to obtain the hyperspectral image data of each light beam.

[0020] When the present invention is working, by sequentially controlling the exposure time, it is possible to obtain several items of hyperspectral image data with different exposure times for each light beam, with high spatial consistency, which is convenient for subsequent image fusion work. In addition, a method for judging oversaturated pixels is adopted, so that the fused hyperspectral image data has ultra-wide spectral characteristics while ensuring the data response range, which can avoid the defects of the collected image being too bright or too dark, resulting in saturation of the hyperspectral image data or insufficient brightness, and loss of data in some areas, thereby further improving the stability of the spectral test.

[0021] Preferably, in step S4, the plurality of hyperspectral image data are fused according to the anchor points so that the plurality of hyperspectral image data completely overlap in space, and when ultra-high grayscale hyperspectral image fusion imaging is completed, the following steps are adopted:

[0022] C1: Acquire ultra-high grayscale hyperspectral image data for each beam;

[0023] C2: Mark the common points of the hyperspectral image data of each light beam, and set the hyperspectral image data of one of the light beams as the reference hyperspectral image data;

[0024] C3: Traverse all spatial points on the reference hyperspectral image data and obtain the spectral curve of each spatial point. Use linear interpolation to calculate the corresponding spatial point on the hyperspectral image data of other beams. According to the linear interpolation parameters, calculate the linear interpolation value of each band to obtain the spectral curve of the corresponding spatial point.

[0025] C4: Merge the spectral curve of each spatial point with the spectral curve of the corresponding spatial point to obtain a merged spectral curve;

[0026] C5: Complete ultra-high grayscale hyperspectral image fusion imaging by spatially overlapping several hyperspectral image data.

[0027] When the present invention is working, the spectrum merging of the hyperspectral images of several light beams is completed through linear interpolation calculation, so that the data of several hyperspectral images completely overlap in space, with good spatial consistency, and a certain degree of defect correction can be achieved, thereby enhancing the integrity of the hyperspectral image data, and being able to suppress the interference of noise on the hyperspectral image data while improving the resolution. At the same time, the computing power resources occupied by the linear interpolation calculation are low, the computing efficiency is high, and the test efficiency can be further improved.

[0028] In order to solve the above technical problems, the second aspect of the present invention adopts the following technical solution: an ultra-high grayscale intelligent hyperspectral imaging device, comprising a beam splitting mechanism for splitting light according to a preset light intensity distribution ratio, a plurality of hyperspectral modules and a processing module for data processing and overall control, the beam splitting mechanism is provided with at least two light-transmitting parts for exposing the light beams after splitting, the imaging parts of the plurality of hyperspectral modules are respectively optically connected to the corresponding light-transmitting parts on the beam splitting mechanism, the data output ends of the plurality of hyperspectral modules are data-connected to the processing module, and the processing module receives the ultra-high grayscale hyperspectral image data of each light beam after the imaging of the plurality of hyperspectral modules is completed to perform hyperspectral image fusion imaging.

[0029] Preferably, the hyperspectral module includes a slit for beam shaping, a spectroscopic module for collimating the light beam, a grating for diffraction spectrometry, and a grayscale camera for imaging. The slit, spectroscopic module, grating and imaging parts of the grayscale camera are arranged in sequence along the direction of the light path and the light path is conductive. The slit is arranged at the focal position of the light-transmitting part of the beam-splitting mechanism and is located at the object-side focal position of the spectroscopic module. The imaging part of the grayscale camera is arranged at the image-side focal position of the spectroscopic module. The grayscale camera is data-connected to the processing module.

[0030] Preferably, the hyperspectral module further comprises a filter for selectively transmitting or blocking light of a specific wavelength, and the filter is arranged in the focusing light path of the spectroscopic module.

[0031] Preferably, the grayscale camera sends control parameters of the grayscale camera synchronously when sending hyperspectral image data after completing imaging, and the control parameters of the grayscale camera include at least one of frame rate, exposure time, ROI parameter and gain.

[0032] Preferably, the processing module includes a processing board and at least one computing chip, and the computing chip is configured as at least one of FPGA, DSP and GPU.

[0033] The beneficial technical effects of the present invention include:

[0034] 1. The present invention realizes multi-band hyperspectral imaging of a common imaging window by means of spectroscopic means, without the need for additional camera arrangement steps, can optimize the workflow of image acquisition, reduce the labor intensity of operators, and get rid of the limitations of sensor sampling hardware circuits, facilitating miniaturization and portability. It improves test efficiency while reducing hardware load pressure, greatly expands the hyperspectral band range by improving the data response range and data bit depth, and simultaneously adopts a multi-exposure and fusion method to obtain ultra-high grayscale hyperspectral images, thereby improving the accuracy of object spectral measurement.

[0035] 2. The present invention can obtain several items of hyperspectral image data with different exposure times for each light beam by sequentially controlling the exposure time. The spatial consistency is high, which facilitates the subsequent image fusion work. The method of judging oversaturated pixels is adopted, so that the fused hyperspectral image data has ultra-wide spectral characteristics while ensuring the data response range. It can avoid the defects of too bright or too dark in the collected image, which leads to saturation or insufficient brightness of the hyperspectral image data and loss of data in some areas, thereby further improving the stability of the spectral test.

[0036] 3. The present invention completes the spectral merging of the hyperspectral images of several light beams through linear interpolation calculation, so that the data of several hyperspectral images completely overlap in space, with good spatial consistency, and can achieve a certain degree of defect correction, thereby enhancing the integrity of the hyperspectral image data, and can suppress the interference of noise on the hyperspectral image data while improving the resolution. At the same time, the computing power resources occupied by the linear interpolation calculation are low, the computing efficiency is high, and the test efficiency can be further improved.

[0037] Other features and advantages of the present invention will be disclosed in detail in the following specific embodiments and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The present invention will be further described below with reference to the accompanying drawings:

[0039] Attachment Figure 1 Schematic diagram of the structure of an ultra-high grayscale intelligent hyperspectral imaging device;

[0040] Attachment Figure 2 Workflow diagram for collecting multiple exposures of a beam to obtain hyperspectral image data;

[0041] Attachment Figure 3 The workflow diagram for hyperspectral image fusion imaging;

[0042] Attachment Figure 4 For reference in the second embodiment Figure 1 ;

[0043] Attachment Figure 5 For reference in the second embodiment Figure 2 . DETAILED DESCRIPTION

[0044] The following is an explanation and description of the technical solutions of the embodiments of the present invention in conjunction with the drawings of the embodiments of the present invention. However, the following embodiments are only preferred embodiments of the present invention and are not exhaustive. Based on the embodiments in the implementation manner, other embodiments obtained by those skilled in the art without creative work are all within the scope of protection of the present invention.

[0045] In the following description, terms such as "inside", "outside", "up", "down", "left", "right", etc. that indicate directions or positional relationships are only used to facilitate the description of the embodiments and simplify the description, and do not indicate or imply that the device or element referred to must have a specific direction, be constructed and operate in a specific direction. Therefore, they should not be understood as limiting the present invention.

[0046] Example 1:

[0047] Please see the attached Figure 1 and attached Figure 2 This embodiment discloses an ultra-high grayscale intelligent hyperspectral imaging method, comprising the following steps:

[0048] S1: Initialization of the intelligent hyperspectral imaging device, deploying the intelligent hyperspectral imaging device to align with the measured area;

[0049] S2: Splitting the captured light signal image into at least two beams according to a preset light intensity distribution ratio before focusing, focusing them on different slits 21 for shaping, and pre-processing each split light beam;

[0050] S3: performing multiple exposure acquisition on a plurality of light beams to obtain hyperspectral image data of each light beam;

[0051] S4: Fuse several pieces of hyperspectral image data according to the anchor points so that the pieces of hyperspectral image data completely overlap in space, and complete the hyperspectral image fusion imaging with ultra-high grayscale.

[0052] When this embodiment is working, multi-band hyperspectral imaging of a common imaging window is achieved through spectroscopic means, without the need for additional camera arrangement steps, which can optimize the image acquisition workflow, reduce the labor intensity of operators, and get rid of the limitations of the sensor sampling hardware circuit, making it convenient for miniaturization and portability. It improves test efficiency while reducing hardware load pressure, greatly expands the hyperspectral band range by improving the data response range and data bit depth, and simultaneously adopts a multi-exposure and fusion method to obtain ultra-high grayscale hyperspectral images, thereby improving the accuracy of object spectral measurement.

[0053] Preferably, in step S2, the captured light signal image is split into two beams according to a preset light intensity distribution ratio before focusing, and the light intensity distribution ratio of the two beams is set to 1:1, 1:1.5 or 1:4, and respectively focused on different slits 21 for shaping. In a specific implementation, the light intensity distribution ratio is optimized according to the actual preferred band distribution of the measured object, which can further improve the data response range and realize the ultra-wide spectral characteristics of the collected hyperspectral image data.

[0054] Preferably, in step S2, when processing each split light beam, the following steps are adopted:

[0055] A1: After the light beam is shaped by the slit 21, it is collimated;

[0056] A2: Perform diffraction and spectral splitting on the collimated light beam. If complete spectral information is required, proceed to step S3; otherwise, proceed to step A3.

[0057] A3: After filtering the required light beam after the splitting is completed, go to step S3.

[0058] In this embodiment, in step S3, when performing multiple exposure acquisition on a plurality of light beams to obtain hyperspectral image data of each light beam, the following steps are adopted:

[0059] B1: According to the preset exposure parameter group, expose in sequence from low to high gradient, and obtain several hyperspectral image data of different exposure times for each beam;

[0060] B2: Traverse all spatial points and determine whether there are oversaturated pixels at each spatial point in descending order. If there are oversaturated pixels at the spatial point in the current hyperspectral image data, proceed to step B3. If there are no oversaturated pixels at the spatial point in the current hyperspectral image data, proceed to step B4.

[0061] B3: Determine whether there is an oversaturated pixel at the spatial point in the hyperspectral image data one level lower than the current hyperspectral image data. If the spatial point has an oversaturated pixel in the current hyperspectral image data, proceed to step B3. If the spatial point has no oversaturated pixel in the current hyperspectral image data, proceed to step B4.

[0062] B4: The hyperspectral data of all spatial points of each light beam is recorded as the ratio of the current hyperspectral image data to the corresponding exposure time to obtain the hyperspectral image data of each light beam.

[0063] Please see the attached Figure 2For the spatial point x1, y1, the Nth hyperspectral data with the largest exposure time is HSIx1, y1, b, N. At this time, the spatial point x1, y1 is judged whether there are oversaturated pixels. When HSIx1, y1, b, N does not have oversaturated pixels, the hyperspectral data of the spatial point x1, y1 is determined as HSIx1, y1, b, N=HSIx1, y1, b, tN / tN. When HSIx1, y1, b, N has oversaturated pixels, the N-1th hyperspectral data HSIx1, y1, b, N-1 with a smaller exposure time is judged whether there are oversaturated pixels. This is repeated recursively until there are no oversaturated pixels at the spatial point x1, y1, and all spatial points are traversed repeatedly in order to obtain the hyperspectral image data of each light beam.

[0064] When this embodiment is working, by sequentially controlling the exposure time, it is possible to obtain several items of hyperspectral image data with different exposure times for each light beam, with high spatial consistency, which is convenient for subsequent image fusion work. In addition, the method of judging oversaturated pixels is adopted, so that the fused hyperspectral image data has ultra-wide spectral characteristics while ensuring the data response range, which can avoid the defects of the collected image being too bright or too dark, resulting in saturation of the hyperspectral image data or insufficient brightness, and loss of data in some areas, thereby further improving the stability of the spectral test.

[0065] Example 2:

[0066] Please see the attached Figure 3 This embodiment provides an ultra-high grayscale intelligent hyperspectral imaging method. The same points as those in the first embodiment will not be repeated here. The differences will be described in detail below with reference to the accompanying drawings.

[0067] Please see the attached Figure 3 To the attached Figure 5 In this embodiment, in step S4, a plurality of hyperspectral image data are fused according to the anchor points so that the plurality of hyperspectral image data completely overlap in space. When ultra-high grayscale hyperspectral image fusion imaging is completed, the following steps are adopted:

[0068] C1: Acquire ultra-high grayscale hyperspectral image data for each beam;

[0069] C2: Mark the common points of the hyperspectral image data of each light beam, and set the hyperspectral image data of one of the light beams as the reference hyperspectral image data;

[0070] C3: Traverse all spatial points on the reference hyperspectral image data and obtain the spectral curve of each spatial point. Use linear interpolation to calculate the corresponding spatial point on the hyperspectral image data of other beams. According to the linear interpolation parameters, calculate the linear interpolation value of each band to obtain the spectral curve of the corresponding spatial point.

[0071] C4: Merge the spectral curve of each spatial point with the spectral curve of the corresponding spatial point to obtain a merged spectral curve;

[0072] C5: Complete ultra-high grayscale hyperspectral image fusion imaging by spatially overlapping several hyperspectral image data.

[0073] Please see the attached Figure 4 and attached Figure 5 , the A-band hyperspectral image and the B-band hyperspectral image are fused, and the combined spectrum of the selected spatial point after fusion is shown in the attached figure. Figure 5 As shown in the figure, it can avoid collection interference by fusing the spectral information of band A and band B, and optimizing the collection of visible light band, near infrared band, short infrared band and other special bands, so as to obtain more comprehensive spectral data, covering a wider spectral range, so that the fused hyperspectral image data is as shown in the attached figure. Figure 4 As shown in the middle right image, its clarity and details are more complete and clear than the left first and left second images, which can further improve the test accuracy.

[0074] When this embodiment is working, the spectrum merging of the hyperspectral images of several light beams is completed through linear interpolation calculation, so that the data of several hyperspectral images completely overlap in space, with good spatial consistency, and a certain degree of defect correction can be achieved, thereby enhancing the integrity of the hyperspectral image data, and being able to suppress the interference of noise on the hyperspectral image data while improving the resolution. At the same time, the computing power resources occupied by the linear interpolation calculation are low, the computing efficiency is high, and the test efficiency can be further improved.

[0075] Example 3:

[0076] Please see the attached Figure 1 This embodiment provides an ultra-high grayscale intelligent hyperspectral imaging device, which applies the intelligent hyperspectral imaging method of the above embodiment, including a beam splitting mechanism 1 for splitting light according to a preset light intensity distribution ratio, a plurality of hyperspectral modules 2, and a processing module 3 for data processing and overall control. The differences are described in detail below with reference to the accompanying drawings.

[0077] In this embodiment, the beam splitting mechanism 1 is provided with at least two light-transmitting parts for exposing the light beam after splitting. In a specific implementation, the beam splitting mechanism 1 includes an imaging lens 11 and a beam splitting system 12, wherein the beam splitting system 12 can be provided with several light-transmitting parts according to actual working requirements, and the imaging parts of several hyperspectral modules 2 are respectively connected to the optical paths of the corresponding light-transmitting parts on the beam splitting mechanism 1, and the data output ends of the several hyperspectral modules 2 are all data-connected with the processing module 3. After the imaging of the several hyperspectral modules 2 is completed, the processing module 3 receives the ultra-high grayscale hyperspectral image data of each light beam to perform hyperspectral image fusion imaging.

[0078] Preferably, the hyperspectral module 2 includes a slit 21 for beam shaping, a spectroscopic module for collimating the light beam, a grating 23 for diffraction spectrometry, and a grayscale camera 24 for imaging. The slit 21, the spectroscopic module, the grating 23 and the imaging parts of the grayscale camera 24 are arranged in sequence along the direction of the light path and the light path is conductive. The slit 21 is arranged at the focal position of the light-transmitting part of the beam splitting mechanism 1 and is located at the object side focal position of the spectroscopic module. The imaging part of the grayscale camera 24 is arranged at the image side focal position of the spectroscopic module. The grayscale camera 24 is data-connected to the processing module 3.

[0079] In a specific implementation, the hyperspectral module 2 further includes a filter 25 for selectively transmitting or blocking light of a specific wavelength. The filter 25 is arranged in the focusing light path of the spectroscopic module to facilitate adjustment of the wavelength distribution.

[0080] Preferably, the grayscale camera 24 sends the control parameters of the grayscale camera 24 synchronously when sending the hyperspectral image data after completing the imaging. The control parameters of the grayscale camera 24 include at least one of the frame rate, exposure time, ROI parameter and gain, which facilitates the overall control of the processing module 3. Preferably, the processing module 3 includes a processing board 31 and at least one computing power chip 32. The computing power chip 32 is arranged on the processing board 31. The processing board 31 is connected to the grayscale camera 24 data through a corresponding data line. The computing power chip 32 is set to at least one of FPGA, DSP and GPU. Of course, any other suitable chip can also be selected according to the actual working conditions. Through the coordinated processing of multiple computing power chips 32, the computing efficiency and fusion effect can be guaranteed, thereby improving the accuracy of the test.

[0081] The beneficial technical effects of this embodiment include: the present invention realizes multi-band hyperspectral imaging of a common imaging window through spectroscopic means, does not require additional camera arrangement steps, can optimize the image acquisition workflow, reduces the labor intensity of operators, and can get rid of the limitations of the sensor sampling hardware circuit, facilitate miniaturization and portability design, improve test efficiency while reducing hardware load pressure, greatly expand the hyperspectral band range by improving the data response range and data bit depth, and simultaneously adopts a multi-exposure and fusion method to obtain ultra-high grayscale hyperspectral images, thereby improving the accuracy of object spectral measurement.

[0082] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Those skilled in the art will understand that the present invention includes, but is not limited to, the contents described in the drawings and the above specific embodiments. Any modifications that do not deviate from the functional and structural principles of the present invention are intended to be included within the scope of the claims.

Claims

1. An ultra-high grayscale intelligent hyperspectral imaging method, characterized in that: The following steps are involved: S1: Initialize the intelligent hyperspectral imaging device and deploy the intelligent hyperspectral imaging device to align with the measured area; S2: before focusing, the captured light signal image is split into at least two beams according to a preset light intensity distribution ratio, and the beams are respectively focused on different slits (21) for shaping, and each split light beam is pre-processed; S3: performing multiple exposure acquisition on a plurality of light beams to obtain hyperspectral image data of each light beam; In step S3, when performing multiple exposure acquisition on a plurality of light beams to obtain hyperspectral image data of each light beam, the following steps are adopted: B1: According to the preset exposure parameter group, expose in sequence from low to high gradient, and obtain several hyperspectral image data of different exposure times for each beam; B2: Traverse all spatial points and determine whether there are oversaturated pixels at each spatial point in descending order. If there are oversaturated pixels at the spatial point in the current hyperspectral image data, proceed to step B3. If there are no oversaturated pixels at the spatial point in the current hyperspectral image data, proceed to step B4. B3: Determine whether there is an oversaturated pixel at the spatial point in the hyperspectral image data one level lower than the current hyperspectral image data. If the spatial point has an oversaturated pixel in the current hyperspectral image data, proceed to step B3. If the spatial point has no oversaturated pixel in the current hyperspectral image data, proceed to step B4. B4: Record the hyperspectral data of all spatial points of each light beam as the ratio of the current hyperspectral image data to the corresponding exposure time to obtain the hyperspectral image data of each light beam; S4: Fuse several pieces of hyperspectral image data according to the anchor points so that the pieces of hyperspectral image data completely overlap in space, and complete the hyperspectral image fusion imaging with ultra-high grayscale.

2. The ultra-high grayscale intelligent hyperspectral imaging method according to claim 1, characterized in that: In step S2, the captured light signal image is split into two beams according to a preset light intensity distribution ratio before focusing, the light intensity distribution ratio of the two beams is set to 1:1, 1:1.5 or 1:4, and they are respectively focused on different slits (21) for shaping.

3. The ultra-high grayscale intelligent hyperspectral imaging method according to claim 1, characterized in that: In step S2, when processing each split light beam, the following steps are adopted: A1: After the light beam passes through the slit (21) and is shaped, it is collimated; A2: Perform diffraction and spectral splitting on the collimated light beam. If complete spectral information is required, proceed to step S3; otherwise, proceed to step A3. A3: After filtering the required light beam after the splitting is completed, go to step S3.

4. The ultra-high grayscale intelligent hyperspectral imaging method according to claim 1, characterized in that: In step S4, the multiple hyperspectral image data are fused according to the anchor points so that the multiple hyperspectral image data completely overlap in space. When the ultra-high grayscale hyperspectral image fusion imaging is completed, the following steps are adopted: C1: Acquire ultra-high grayscale hyperspectral image data for each beam; C2: Mark the common points of the hyperspectral image data of each light beam, and set the hyperspectral image data of one of the light beams as the reference hyperspectral image data; C3: Traverse all spatial points on the reference hyperspectral image data and obtain the spectral curve of each spatial point. Use linear interpolation to calculate the corresponding spatial point on the hyperspectral image data of other beams. According to the linear interpolation parameters, calculate the linear interpolation value of each band to obtain the spectral curve of the corresponding spatial point. C4: Merge the spectral curve of each spatial point with the spectral curve of the corresponding spatial point to obtain a merged spectral curve; C5: Complete ultra-high grayscale hyperspectral image fusion imaging by spatially overlapping several hyperspectral image data.

5. An ultra-high grayscale intelligent hyperspectral imaging device, using the ultra-high grayscale intelligent hyperspectral imaging method according to any one of claims 1 to 4, characterized in that: The invention comprises a beam splitting mechanism (1) for splitting light according to a preset light intensity distribution ratio, a plurality of hyperspectral modules (2) and a processing module (3) for data processing and overall control, wherein the beam splitting mechanism (1) is provided with at least two light-transmitting parts for exposing the light beam after splitting, the imaging parts of the plurality of hyperspectral modules (2) are respectively connected to the light paths of the light-transmitting parts corresponding to each other on the beam splitting mechanism (1), the data output ends of the plurality of hyperspectral modules (2) are all data-connected to the processing module (3), and the processing module (3) receives the ultra-high grayscale hyperspectral image data of each light beam after the imaging of the plurality of hyperspectral modules (2) is completed to perform hyperspectral image fusion imaging.

6. The ultra-high grayscale intelligent hyperspectral imaging device according to claim 5, characterized in that: The hyperspectral module (2) comprises a slit (21) for beam shaping, a spectroscopic system (22) for collimating the beam, a grating (23) for diffraction spectrometry, and a grayscale camera (24) for imaging. The imaging parts of the slit (21), the spectroscopic system (22), the grating (23), and the grayscale camera (24) are arranged in sequence along the direction of the light path conduction and the light path conduction is conducted. The slit (21) is arranged at the focal position of the light-transmitting part of the beam splitting mechanism (1) and is located at the object-side focal position of the spectroscopic system (22). The imaging part of the grayscale camera (24) is arranged at the image-side focal position of the spectroscopic system (22). The grayscale camera (24) is data-connected to the processing module (3).

7. The ultra-high grayscale intelligent hyperspectral imaging device according to claim 6, characterized in that: The hyperspectral module (2) further includes a filter (25) for selectively transmitting or blocking light of a specific wavelength, wherein the filter (25) is arranged in the focusing light path of the optical splitting system (22).

8. The ultra-high grayscale intelligent hyperspectral imaging device according to claim 6, characterized in that: The grayscale camera (24) synchronously sends control parameters of the grayscale camera (24) when sending hyperspectral image data after completing imaging, and the control parameters of the grayscale camera (24) include at least one of a frame rate, an exposure time, an ROI parameter, and a gain.

9. The ultra-high grayscale intelligent hyperspectral imaging device according to claim 5, characterized in that: The processing module (3) comprises a processing board (31) and at least one computing chip (32), wherein the computing chip (32) is configured as at least one of an FPGA, a DSP, and a GPU.

Images