A nonlinear spectral calibration method and system for hyperspectral cameras

Abstract

The present invention discloses a nonlinear spectral calibration method and system for a hyperspectral camera, comprising: S1: under any environmental parameter, collecting pixel brightness values ​​of a high-reflection area, a medium-reflection area, and a low-reflection area of ​​a multi-region standard material calibration plate as raw data; S2: constructing a correction lookup table, the correction lookup table containing fitting coefficients under different environments; S3: judging whether the environmental change value exceeds a threshold; if not, using the correction coefficient to calculate the error term; if exceeding the threshold, repeating S1, S2, and S3 to calibrate the fitting coefficient and calculate the error term under real-time environmental parameters; S4: obtaining the actual reflectivity based on the error term and preliminary reflectivity calculation; S5: calculating the wavelength offset, correcting the wavelength of the full spectrum based on the wavelength offset of the full spectrum, performing real-time compensation for the environmental parameters, and dynamically correcting the reflectivity calculation model based on environmental changes, thereby ensuring that the calibration process is not disturbed by environmental factors and remains stable.

Description

Technical Field

[0001] The present invention relates to the technical field of nonlinear spectral calibration of hyperspectral cameras, and in particular to a method and system for nonlinear spectral calibration of hyperspectral cameras. Background Art

[0002] Hyperspectral imaging technology, leveraging its ability to capture the reflectance or radiation of a target object across continuous spectral bands, has been widely applied in fields such as remote sensing and environmental monitoring. Spectral calibration, at the core of this technology, directly impacts the accuracy and application value of hyperspectral camera data. However, current spectral calibration systems suffer from multiple technical bottlenecks, severely restricting the engineering application of hyperspectral imaging technology.

[0003] Traditional spectral calibration has significant flaws in both technology and execution. In terms of technical implementation, it relies on laboratory-grade precision equipment such as monochromators and standard light sources. Not only are the equipment large in size and have high purchase and maintenance costs, making it difficult to quickly calibrate in the field or industrial sites, but changes in ambient light intensity, temperature and humidity can easily introduce coupling errors, affecting calibration stability. At the same time, the linear approximation method for processing spectral response functions ignores actual nonlinear characteristics, resulting in reduced spectral resolution and radiation measurement accuracy. In terms of solution implementation, the complex multi-device collaborative calibration process requires professional personnel to operate, with high manpower and technical barriers. The static calibration mode cannot adapt to the real-time calibration needs of dynamic environments, and the calibration results quickly become invalid when environmental parameters fluctuate. The decoupling of radiation calibration and spectral calibration makes it difficult to coordinate and optimize key parameters, which seriously affects the accuracy of the comprehensive inversion of hyperspectral data. The dual limitations of technical implementation and engineering execution make it difficult for traditional calibration systems to meet the dynamic calibration needs of complex scenes, becoming a major obstacle to the promotion and application of hyperspectral imaging technology. Summary of the Invention

[0004] In response to the problems in the related art, the present invention proposes a nonlinear spectral calibration method for a hyperspectral camera to solve the above-mentioned technical problems.

[0005] To this end, the specific technical solution adopted by the present invention is as follows: A nonlinear spectral calibration method for a hyperspectral camera, comprising:

[0006] S1: Under any environmental parameters, the hyperspectral camera to be calibrated is used to sequentially capture the pixel brightness values ​​of the high-reflection area, medium-reflection area, and low-reflection area of ​​the multi-region standard material calibration plate as raw data;

[0007] S2: Calculate the relationship parameters between the true reflectance of the same standard material and the pixel brightness value based on the nonlinear relationship between the two, and fit the relationship parameters to generate a correction lookup table. The correction lookup table contains fitting coefficients under different environments.

[0008] S3: Use the uncalibrated hyperspectral camera to collect different environmental parameters and analyze and process them to obtain the correction coefficients corresponding to the different environmental parameters. Calculate the change value between the real-time environmental parameters and the environmental parameters of S2, and determine whether the change value exceeds the threshold. If it does not exceed the threshold, use the correction coefficient to calculate the error term. If it exceeds the threshold, repeat S1, S2 and S3 under the real-time environmental parameters to calibrate the fitting coefficients and calculate the error term.

[0009] S4: Based on the nonlinear relationship between the true reflectance of the same standard material and the pixel brightness value, the calibrated fitting coefficient is brought in to calculate the preliminary reflectance, and the actual reflectance is obtained by adding the error term and the preliminary reflectance;

[0010] S5: Peak search is performed on the reflectivity curve formed by the preliminary reflectivity to obtain an actual characteristic peak wavelength, the actual characteristic peak wavelength is compared and analyzed with the characteristic wavelength of the standard substance to obtain a wavelength offset, the wavelength offset of the full spectrum segment is calculated based on the wavelength offset, and the wavelength of the full spectrum segment is corrected based on the wavelength offset of the full spectrum segment;

[0011] S6: The corrected spectral data, real-time environmental parameters, calibrated fitting coefficients, wavelength offset of the entire spectrum, and error terms are stored as generated high-precision calibration parameters in the historical calibration parameters.

[0012] This nonlinear spectral calibration method for hyperspectral cameras performs real-time compensation through environmental parameters, monitors parameters such as ambient light intensity fluctuations, temperature and humidity changes in real time, and dynamically corrects the reflectance calculation model based on environmental changes to ensure that the calibration process is not interfered with by environmental factors and remains stable. This enables hyperspectral imaging to obtain accurate and reliable calibration results in complex and changeable environments such as the wild and industrial sites, greatly improving the adaptability of the calibration system in complex environments.

[0013] Furthermore, the environmental parameters include the light intensity, temperature and humidity in the acquisition environment. The environmental parameter acquisition is achieved through the ambient light sensor and temperature and humidity sensor integrated in the hyperspectral camera to be calibrated. Specifically: obtain the temperature correction coefficient k t When the humidity and brightness conditions are fixed, starting from the starting temperature, the temperature is changed within a certain temperature range with a certain temperature interval as the step length, and the reflectance data of the calibration plate at each temperature point is collected. The reflectance at the starting temperature is used as the reference value, and the error between the reflectance and the reference value at different temperatures is calculated respectively. The linear regression method is used to fit the linear relationship between the temperature error and the reflectance error, and then the temperature correction coefficient k is obtained. t . Get the humidity correction coefficient k hWhen the humidity gradient is set, the humidity environment is changed from the starting humidity to the end humidity, and the reflectivity of the calibration plate is measured at each humidity point. The reflectivity error caused by the humidity change is calculated based on the reflectivity at a certain standard humidity. The relationship between the humidity difference and the reflectivity error is established through linear fitting to obtain the humidity correction coefficient k. h . Get the brightness correction coefficient k l When the humidity and temperature conditions are fixed, the light intensity is changed according to a certain brightness gradient, and the lighting environment is adjusted from the starting brightness to the end brightness range. The reflectance data of the calibration plate at different brightnesses are collected. The reflectance at a certain standard brightness is used as a reference to calculate the reflectance error caused by the brightness change. The linear fitting method is used to determine the correlation between the brightness difference and the reflectance error, and the brightness correction coefficient k is obtained. l .

[0014] Furthermore, the change values ​​of the real-time environmental parameters in S3 and the environmental parameters in S2 are temperature error ΔT, humidity error ΔH, and brightness error ΔL, respectively. It is judged whether the temperature error ΔT, humidity error ΔH, and brightness error ΔL exceed the threshold value. If none of the three exceed the threshold value, the corresponding correction coefficients are multiplied by the errors to obtain their respective error terms. The three error terms are accumulated and added to the preliminary reflectivity to obtain the actual reflectivity. If any of the temperature error ΔT, humidity error ΔH, and brightness error ΔL exceeds the threshold value, S1, S2, and S3 are repeated under the real-time environmental parameters to calibrate the fitting coefficients and calculate the error terms.

[0015] Furthermore, in S4, according to the nonlinear relationship between the true reflectance of the same standard material and the pixel brightness value, when the calibrated fitting coefficient is brought into the calculation of the preliminary reflectance, the pixel brightness value is the pixel brightness value of the middle reflective area of ​​the multi-region standard material calibration plate.

[0016] Furthermore, when calculating the wavelength offset of the entire spectrum based on the wavelength offset, the wavelength offset of the entire spectrum is corrected through interpolation or wavelength remapping algorithms. The interpolation algorithm reasonably infers the correction values ​​of other wavelength points in the entire spectrum based on the known wavelength offset; the wavelength remapping algorithm remaps and adjusts the wavelength distribution of the entire spectrum, so that the wavelength of the entire spectrum can be accurately corrected, making the wavelength output by the camera more consistent with the actual situation.

[0017] A hyperspectral camera nonlinear spectral calibration system, used to implement any of the above-mentioned hyperspectral camera nonlinear spectral calibration methods, comprising: a multi-region standard material calibration plate, a data processing module, an environmental sensor module, and a hyperspectral camera to be calibrated.

[0018] The multi-region standard material calibration plate includes high reflection area, medium reflection area and low reflection area;

[0019] The environmental sensor module is used to obtain environmental parameters;

[0020] The hyperspectral camera to be calibrated is used for spectrum acquisition;

[0021] The data processing module is used for data collection and processing, including:

[0022] S1: Under any environmental parameters, the hyperspectral camera to be calibrated is used to sequentially capture the pixel brightness values ​​of the high-reflection area, medium-reflection area, and low-reflection area of ​​the multi-region standard material calibration plate as raw data;

[0023] S2: Calculate the relationship parameters between the true reflectance of the same standard material and the pixel brightness value based on the nonlinear relationship between the two, and fit the relationship parameters to generate a correction lookup table. The correction lookup table contains fitting coefficients under different environments.

[0024] S3: Use the uncalibrated hyperspectral camera to collect different environmental parameters and analyze and process them to obtain the correction coefficients corresponding to the different environmental parameters. Calculate the change value between the real-time environmental parameters and the environmental parameters of S2, and determine whether the change value exceeds the threshold. If it does not exceed the threshold, use the correction coefficient to calculate the error term. If it exceeds the threshold, repeat S1, S2 and S3 under the real-time environmental parameters to calibrate the fitting coefficients and calculate the error term.

[0025] S4: Based on the nonlinear relationship between the true reflectance of the same standard material and the pixel brightness value, the calibrated fitting coefficient is brought in to calculate the preliminary reflectance, and the actual reflectance is obtained by adding the error term and the preliminary reflectance;

[0026] S5: Peak search is performed on the reflectivity curve formed by the preliminary reflectivity to obtain an actual characteristic peak wavelength, the actual characteristic peak wavelength is compared and analyzed with the characteristic wavelength of the standard substance to obtain a wavelength offset, the wavelength offset of the full spectrum segment is calculated based on the wavelength offset, and the wavelength of the full spectrum segment is corrected based on the wavelength offset of the full spectrum segment;

[0027] S6: The corrected spectral data, real-time environmental parameters, calibrated fitting coefficients, wavelength offset of the entire spectrum, and error terms are stored as generated high-precision calibration parameters in the historical calibration parameters.

[0028] The beneficial effects of the present invention are:

[0029] 1. This hyperspectral camera nonlinear spectral calibration method uses real-time compensation of environmental parameters, monitors ambient light intensity fluctuations, temperature and humidity changes, and dynamically modifies the reflectance calculation model based on environmental changes, ensuring that the calibration process is not affected by environmental factors and remains stable. This enables hyperspectral imaging to obtain accurate and reliable calibration results in complex and changing environments such as the wild and industrial sites, greatly improving the adaptability of the calibration system in complex environments.

[0030] 2. The calibration process of this hyperspectral camera nonlinear spectral calibration system does not require complex and expensive laboratory equipment such as monochromators and standard light sources. The work can be completed only through reusable calibration plates, which greatly reduces equipment acquisition and maintenance costs. The multiple use of the calibration plates further amortizes the single cost and effectively controls overall expenditures. At the same time, this technology supports rapid calibration in the field and industrial sites, eliminating the need for complex equipment setup and long-term operation by professionals, greatly reducing calibration time and significantly improving the application efficiency and flexibility of hyperspectral imaging technology, enabling its rapid application in various practical scenarios.

[0031] 3. The multi-region reference material calibration plate used in this hyperspectral camera nonlinear spectral calibration system utilizes a multi-region design with characteristic peak identification, fully covering the dynamic range of hyperspectral imaging and providing a precise spectral "fingerprint" reference, establishing a clear benchmark for spectral analysis. The nonlinear joint correction technology simultaneously processes radiometric and spectral calibration, overcoming the drawbacks of traditional separate calibration. This joint calibration model applies nonlinear correction to both radiometric and spectral responses, achieving synergistic optimization. This combination not only overcomes the accuracy loss associated with traditional linear approximations of the spectral response function, but also improves the comprehensive inversion accuracy of hyperspectral data, ensuring high-resolution and radiometric measurement accuracy in acquired spectral data. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0033] Figure 1 This is a flowchart of the steps of a nonlinear spectral calibration method for a hyperspectral camera according to the first embodiment of the present invention. DETAILED DESCRIPTION

[0034] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0035] Example 1

[0036] A nonlinear spectral calibration method for hyperspectral cameras, such as Figure 1 Shown, including:

[0037] S1: Under any environmental parameter, such as light intensity (brightness) L1, temperature T1, and humidity H1, the hyperspectral camera to be calibrated is used to sequentially capture the pixel brightness values ​​of the high-reflection area, medium-reflection area, and low-reflection area of ​​the multi-region standard material calibration plate as raw data. Among them, the high-reflection area data can provide a high signal-to-noise ratio signal and is mainly used for radiation brightness calibration; the medium-reflection area data is used for dynamic range adaptation to adapt to different lighting conditions; the low-reflection area data is mainly used to correct dark current and background noise.

[0038] S2: Calculate the relationship parameters between the true reflectance and pixel brightness value of the same standard material based on their nonlinear relationship. Fit the relationship parameters to generate a correction lookup table. The correction lookup table contains fitting coefficients under different environments. Specifically, the nonlinear relationship between the true reflectance and pixel brightness value is: R (λ) = a0 + a1D (λ) + a2D (λ) 2 +...a n D(λ) n ; Among them, R (λ) is the real reflectivity of the material, D (λ) is the pixel brightness value, the pixel brightness value is the camera output value, a0, a1...a n is the fitting coefficient. Based on the nonlinear relationship between true reflectance and pixel brightness, the spectral response of each pixel is corrected by nonlinear interpolation, effectively making up for the shortcomings of traditional linear approximation spectral response function and improving the radiation measurement accuracy and spectral resolution.

[0039] In the laboratory, preliminary experiments were conducted to analyze the correspondence between the known reflectivity of the calibration plate standard material and the pixel brightness values ​​captured by the camera, determining the parameters of the nonlinear relationship between the true reflectivity and pixel brightness. Based on the collected spectral data of the calibration plate, a series of environmental parameters were simulated, and the collected data was fitted using the least squares method to generate a calibration lookup table (LUT), which is the fitting coefficient for different environments.

[0040] S3: A hyperspectral camera under calibration is used to collect different environmental parameters and analyze them to obtain corresponding correction coefficients. Environmental parameter collection is achieved using the ambient light sensor and temperature and humidity sensor integrated into the hyperspectral camera under calibration. The temperature error ΔT, humidity error ΔH, and brightness error ΔL of the real-time environmental parameters (such as light intensity (brightness) L2, temperature T2, and humidity H2) are calculated in relation to the changes in light intensity (brightness) L1, temperature T1, and humidity H1. A determination is made as to whether any of these changes exceeds a threshold (e.g., a temperature fluctuation exceeding ±5°C, a light intensity fluctuation exceeding 10%, or a humidity fluctuation exceeding a specified range). If none of these three changes exceeds the threshold, the correction coefficient is used to calculate the error term. If any of these changes exceeds the threshold, steps S1, S2, and S3 are repeated under the real-time environmental parameters to calibrate the fitting coefficients and calculate the error term, ensuring that the camera maintains high-precision measurement performance under the new environmental conditions.

[0041] Where, obtain the temperature correction coefficient k t When the humidity and brightness conditions are fixed, starting from the starting temperature, such as 25℃, the temperature is changed in a certain temperature interval, such as 1℃, within a certain temperature range, such as 15℃-35℃. The reflectance data of the calibration plate are collected at each temperature point, and the reflectance at the starting temperature is used as the reference value. The error between the reflectance and the reference value at different temperatures is calculated respectively. The linear regression method is used to fit the linear relationship between the temperature error and the reflectance error, and then the temperature correction coefficient k is obtained. t .

[0042] Get the humidity correction coefficient k h When setting a series of humidity gradients, such as changing the humidity environment from the starting humidity to the end humidity every 10% RH, such as 30% RH-70% RH, measure the reflectivity of the calibration plate at each humidity point, and use the reflectivity at a certain standard humidity as the standard, such as 50% RH, to calculate the reflectivity error caused by humidity change. Through linear fitting, establish the relationship between humidity difference and reflectivity error, and obtain the humidity correction coefficient k. h .

[0043] Get the brightness correction coefficient k l When the humidity and temperature conditions are fixed, the light intensity is changed according to a certain brightness gradient, such as a 10% light intensity change. The lighting environment is adjusted from the starting brightness to the end brightness range, such as 300 lux to 700 lux. The reflectivity data of the calibration plate at different brightnesses are collected. The reflectivity at a certain standard brightness, such as 500 lux, is used as a reference to calculate the reflectivity error caused by the brightness change. The linear fitting method is used to determine the correlation between the brightness difference and the reflectivity error, and the brightness correction coefficient k is obtained. l .

[0044] S4: According to R(λ)=a0+a1D(λ)+a2D(λ) 2 +...a n D(λ) n , the calibrated fitting coefficient is used to calculate the preliminary reflectivity, where D(λ) uses the pixel brightness value in the middle reflective area, and the error term ΔR=ΔTk t +ΔHk h +ΔLk l , the actual reflectivity is obtained by adding ΔR to the preliminary reflectivity, thereby achieving real-time correction of the impact of environmental changes.

[0045] S5: Peak search is performed on the reflectivity curve formed by the preliminary reflectivity to obtain the actual characteristic peak wavelength. The actual characteristic peak wavelength is compared with the characteristic wavelength of the reference material to obtain the wavelength offset. The wavelength offset of the entire spectrum is calculated based on the wavelength offset. The wavelength of the entire spectrum is corrected based on the wavelength offset of the entire spectrum. When calculating the wavelength offset of the entire spectrum, the wavelength offset of the entire spectrum is corrected using interpolation or wavelength remapping algorithms. The interpolation algorithm uses the known wavelength offset to reasonably infer the correction values ​​for other wavelength points within the entire spectrum. The wavelength remapping algorithm remaps and adjusts the wavelength distribution of the entire spectrum, ensuring that the wavelength of the entire spectrum is accurately corrected, making the wavelength output by the camera more consistent with the actual situation. Reference materials with known spectral characteristic peaks have corresponding "spectral fingerprints", such as 550nm, 1050nm, and 2100nm as the characteristic peak positions of a certain reference material.

[0046] S6: The corrected spectral data, real-time environmental parameters, calibrated fitting coefficients, wavelength offset of the entire spectrum, and error terms are stored as generated high-precision calibration parameters in the historical calibration parameters.

[0047] This nonlinear spectral calibration method for hyperspectral cameras uses multi-region standard materials and environmental dynamic compensation to solve the problem that traditional calibration relies on laboratory-grade precision equipment, is highly sensitive to the environment, and has insufficient nonlinear response correction. As a result, hyperspectral cameras cannot be calibrated quickly and accurately in actual application scenarios such as field monitoring and industrial site detection, which seriously affects the accuracy of spectral data.

[0048] Example 2

[0049] A hyperspectral camera nonlinear spectral calibration system is implemented using the hyperspectral camera nonlinear spectral calibration method of Example 1, comprising: a multi-region standard material calibration plate, a data processing module, an environmental sensor module, and a hyperspectral camera to be calibrated, wherein the multi-region standard material calibration plate includes a high-reflection area, a medium-reflection area, and a low-reflection area; the environmental sensor module is used to obtain environmental parameters; the hyperspectral camera to be calibrated is used to collect spectra; and the data processing module is used to collect and process data.

[0050] The multi-region reference material calibration plate utilizes a composite structure, divided into high, medium, and low reflectivity zones, covering the 400-2500nm spectral range. Each zone is further configured with a series of gradient reflectivity sub-regions, significantly enhancing calibration accuracy and flexibility. Each zone utilizes a reference material with stable spectral characteristics (such as barium oxide, barium sulfate, or custom coating materials), whose reflectivity curves are calibrated in the laboratory and stored as benchmark data.

[0051] Standard substances with known spectral characteristic peaks (such as rare earth oxides or organic dyes) are embedded on the surface of the calibration plate. For example, reflectivity mutation points are set at specific wavelengths (such as 550nm, 1050nm, and 2100nm) to form "spectral fingerprints". These "spectral fingerprints" are the key to achieving accurate wavelength calibration. By comparing them with the spectral data collected by the camera, the wavelength offset can be accurately determined, thereby achieving high-precision wavelength calibration.

[0052] The calibration plate is installed within the field of view of the hyperspectral camera to be calibrated, facilitating efficient acquisition of image data. The calibration plate is constructed from a wear- and weather-resistant ceramic substrate or polymer coating, ensuring excellent environmental adaptability and repeated use in challenging outdoor environments. It eliminates the need for complex laboratory-level maintenance, effectively extending the system's application scenarios and lifespan.

[0053] The environmental sensor module integrates an ambient light sensor, temperature and humidity sensors (such as a digital light meter and SHT35 temperature and humidity module). Installed near the camera and calibration board, it ensures accurate sensing of the actual environmental conditions surrounding the camera and calibration board. Connected to the data processing module via data transmission lines, this module provides continuous, real-time, and accurate monitoring of key environmental parameters such as light intensity, temperature, and humidity. Through the coordinated operation of these sensors, collected data is transmitted in real time to the data processing module within the camera integration system, providing a solid and reliable data foundation for subsequent dynamic compensation of calibration parameters based on environmental parameters.

[0054] The data processing module is built into the camera integration system and connected to the calibration board's drive motor via a motor drive circuit. This precisely controls the calibration board's reciprocating motion, enabling the camera to sequentially capture each area of ​​the calibration board, ensuring comprehensive and accurate calibration data. Simultaneously, the data processing module receives data from the environmental sensor module via a data receiving circuit. Using a built-in high-performance chip, the module processes the data in real time and performs calibration operations based on a pre-set algorithm model, achieving precise calibration of the hyperspectral camera.

[0055] The hyperspectral camera to be calibrated is mounted on a stable support structure, such as a metal bracket or optical platform, to ensure stability during imaging. The camera's lens is aligned with the calibration plate and connected to the data processing module. The camera captures the calibration plate according to the module's instructions, transmitting the raw spectral image data to the module to provide the data foundation for calibration.

[0056] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A nonlinear spectral calibration method for a hyperspectral camera, characterized in that: include: S1: Under any environmental parameters, the hyperspectral camera to be calibrated is used to sequentially capture the pixel brightness values ​​of the high-reflection area, medium-reflection area, and low-reflection area of ​​the multi-region standard material calibration plate as raw data; S2: Calculate the relationship parameters between the true reflectance of the same standard material and the pixel brightness value based on the nonlinear relationship between the two, and fit the relationship parameters to generate a correction lookup table. The correction lookup table contains fitting coefficients under different environments. S3: Use the uncalibrated hyperspectral camera to collect different environmental parameters and analyze and process them to obtain the correction coefficients corresponding to the different environmental parameters. Calculate the change value between the real-time environmental parameters and the environmental parameters of S2, and determine whether the change value exceeds the threshold. If it does not exceed the threshold, use the correction coefficient to calculate the error term. If it exceeds the threshold, repeat S1, S2 and S3 under the real-time environmental parameters to calibrate the fitting coefficients and calculate the error term. S4: Based on the nonlinear relationship between the true reflectance of the same standard material and the pixel brightness value, the calibrated fitting coefficient is brought in to calculate the preliminary reflectance, and the actual reflectance is obtained by adding the error term and the preliminary reflectance; S5: Peak search is performed on the reflectivity curve formed by the preliminary reflectivity to obtain an actual characteristic peak wavelength, the actual characteristic peak wavelength is compared and analyzed with the characteristic wavelength of the standard substance to obtain a wavelength offset, the wavelength offset of the full spectrum segment is calculated based on the wavelength offset, and the wavelength of the full spectrum segment is corrected based on the wavelength offset of the full spectrum segment; S6: The corrected spectral data, real-time environmental parameters, calibrated fitting coefficients, wavelength offset of the entire spectrum, and error terms are stored as generated high-precision calibration parameters in the historical calibration parameters.

2. The nonlinear spectral calibration method for a hyperspectral camera according to claim 1, characterized in that: The true reflectance of the same standard material is positively correlated with the pixel brightness value.

3. A hyperspectral camera nonlinear spectral calibration method according to claim 1 or 2, characterized in that: Environmental parameters include light intensity, temperature and humidity in the collection environment. Environmental parameter collection is achieved through the ambient light sensor and temperature and humidity sensor integrated in the hyperspectral camera to be calibrated.

4. The nonlinear spectral calibration method for a hyperspectral camera according to claim 3, wherein: Get the temperature correction coefficient k t When the humidity and brightness conditions are fixed, starting from the starting temperature, the temperature is changed within a certain temperature range with a certain temperature interval as the step length, and the reflectance data of the calibration plate at each temperature point is collected. The reflectance at the starting temperature is used as the reference value, and the error between the reflectance and the reference value at different temperatures is calculated respectively. The linear regression method is used to fit the linear relationship between the temperature error and the reflectance error, and then the temperature correction coefficient k is obtained. t .

5. The nonlinear spectral calibration method for a hyperspectral camera according to claim 4, characterized in that: Get the humidity correction coefficient k h When the humidity gradient is set, the humidity environment is changed from the starting humidity to the end humidity, and the reflectivity of the calibration plate is measured at each humidity point. The reflectivity error caused by the humidity change is calculated based on the reflectivity at a certain standard humidity. The relationship between the humidity difference and the reflectivity error is established through linear fitting to obtain the humidity correction coefficient k. h .

6. The nonlinear spectral calibration method for a hyperspectral camera according to claim 5, characterized in that: Get the brightness correction coefficient k l When the humidity and temperature conditions are fixed, the light intensity is changed according to a certain brightness gradient, and the lighting environment is adjusted from the starting brightness to the end brightness range. The reflectance data of the calibration plate at different brightnesses are collected. The reflectance at a certain standard brightness is used as a reference to calculate the reflectance error caused by the brightness change. The linear fitting method is used to determine the correlation between the brightness difference and the reflectance error, and the brightness correction coefficient k is obtained. l .

7. The nonlinear spectral calibration method for a hyperspectral camera according to claim 6, characterized in that: The change values ​​of the real-time environmental parameters in S3 and the environmental parameters in S2 are temperature error ΔT, humidity error ΔH, and brightness error ΔL, respectively. It is judged whether the temperature error ΔT, humidity error ΔH, and brightness error ΔL exceed the threshold. If none of the three exceed the threshold, the corresponding correction coefficient is used to multiply the error to obtain the respective error terms. The three error terms are accumulated and added to the preliminary reflectivity to obtain the actual reflectivity.

8. The nonlinear spectral calibration method for a hyperspectral camera according to claim 7, characterized in that: If any of the temperature error ΔT, humidity error ΔH, and brightness error ΔL exceeds the threshold, then under the real-time environmental parameters, S1, S2, and S3 are repeated to calibrate the fitting coefficients and calculate the error terms.

9. The nonlinear spectral calibration method for a hyperspectral camera according to claim 8, characterized in that: When calculating the wavelength offset of the entire spectrum based on the wavelength offset, the wavelength offset of the entire spectrum is corrected by an interpolation or wavelength remapping algorithm.

10. A nonlinear spectral calibration system for a hyperspectral camera, characterized in that: A method for nonlinear spectrum calibration of a hyperspectral camera according to any one of claims 1 to 9, comprising: a multi-region standard material calibration plate, a data processing module, an environmental sensor module, and a hyperspectral camera to be calibrated. The multi-region standard material calibration plate includes high reflection area, medium reflection area and low reflection area; The environmental sensor module is used to obtain environmental parameters; The hyperspectral camera to be calibrated is used for spectrum acquisition; The data processing module is used for data collection and processing, including: S1: Under any environmental parameters, the hyperspectral camera to be calibrated is used to sequentially capture the pixel brightness values ​​of the high-reflection area, medium-reflection area, and low-reflection area of ​​the multi-region standard material calibration plate as raw data; S2: Calculate the relationship parameters between the true reflectance of the same standard material and the pixel brightness value based on the nonlinear relationship between the two, and fit the relationship parameters to generate a correction lookup table. The correction lookup table contains fitting coefficients under different environments. S3: Use the uncalibrated hyperspectral camera to collect different environmental parameters and analyze and process them to obtain the correction coefficients corresponding to the different environmental parameters. Calculate the change value between the real-time environmental parameters and the environmental parameters of S2, and determine whether the change value exceeds the threshold. If it does not exceed the threshold, use the correction coefficient to calculate the error term. If it exceeds the threshold, repeat S1, S2 and S3 under the real-time environmental parameters to calibrate the fitting coefficients and calculate the error term. S4: Based on the nonlinear relationship between the true reflectance of the same standard material and the pixel brightness value, the calibrated fitting coefficient is brought in to calculate the preliminary reflectance, and the actual reflectance is obtained by adding the error term and the preliminary reflectance; S5: Peak search is performed on the reflectivity curve formed by the preliminary reflectivity to obtain an actual characteristic peak wavelength, the actual characteristic peak wavelength is compared and analyzed with the characteristic wavelength of the standard substance to obtain a wavelength offset, the wavelength offset of the full spectrum segment is calculated based on the wavelength offset, and the wavelength of the full spectrum segment is corrected based on the wavelength offset of the full spectrum segment; S6: The corrected spectral data, real-time environmental parameters, calibrated fitting coefficients, wavelength offset of the entire spectrum, and error terms are stored as generated high-precision calibration parameters in the historical calibration parameters.

Images