Infrared calibration method and device based on dynamic background model and emissivity self-correction

Abstract

The application discloses an infrared calibration method and device based on a dynamic background model and emissivity self-correction, belongs to the technical field of infrared remote sensing calibration, and is based on the characteristics of gain stability of a refrigeration infrared camera and background drift with thermal balance, combines spectral response functions and physical anchor radiance, and constructs a calibration model containing a dynamic background term; then, gray scale, temperature and time data of a multi-temperature calibration board and natural water at a take-off and landing point are collected at discrete flight nodes of a UAV, are converted into effective spectral radiance through Planck's law and spectral weighted integration; an emissivity grid sequence is constructed, initial parameters are obtained through a two-point method, a global optimal response slope is fitted through a least square method, and an instantaneous background is recalculated, optimal emissivity and calibration parameters are locked through water body inversion error closed loop, the discrete background is fitted into a negative exponential dynamic evolution model, background continuous prediction is realized, and finally, high-precision radiation inversion is completed through model substitution. The application can accurately compensate background drift and self-correct emissivity error.

Description

Technical Field

[0001] This invention relates to the field of infrared radiation calibration technology, and in particular to an infrared calibration method and apparatus based on a dynamic background model and emissivity self-correction. Background Technology

[0002] Infrared remote sensing detectors mounted on unmanned aerial vehicles (UAVs) have wide applications in field scientific experiments such as thermal environment monitoring and target detection. In order to obtain the absolute radiation intensity of the target, the infrared detector must be rigorously radiometrically calibrated to establish a quantitative relationship between the image grayscale value and the target spectral radiance.

[0003] Currently, existing infrared detector calibration techniques mainly rely on blackbody calibration in a laboratory environment. In a laboratory setting, the camera's gain coefficient and background bias are obtained using a high-precision, temperature-controlled blackbody.

[0004] However, existing technologies have the following shortcomings in the field operation of drones:

[0005] Calibration coefficient failure due to environmental inconsistencies: Factors such as temperature fluctuations, atmospheric path radiation, UAV fuselage heating, and temperature-induced contamination of the infrared detector head in the field flight environment cause differences between actual observation conditions and the controlled laboratory environment. Laboratory calibration coefficients cannot account for the variable background radiation in the field, leading to significant detection errors when directly applied.

[0006] Dynamic background drift is difficult to estimate in real time: During startup and flight, the thermal equilibrium state of the optomechanical system of an infrared-cooled camera changes dynamically over time. Existing technologies typically use a fixed background coefficient, ignoring the background drift characteristics over time, and lack effective real-time reference methods to accurately estimate the current background value during flight missions.

[0007] Uncertainty in the emissivity of calibration reference objects: During field calibration, ground calibration plates are often subject to surface oxidation, moisture condensation, or environmental dust contamination, causing their actual emissivity to deviate from the theoretical value. Existing calibration methods often assume that the emissivity is constant. This assumption can introduce significant source-end radiation errors in complex field environments, thereby affecting the overall calibration accuracy.

[0008] Therefore, how to combine the characteristics of field experiments to establish a high-precision calibration method that can correct background drift in real time and achieve self-correction of the emissivity of the calibration source is a technical problem that urgently needs to be solved in the field of UAV infrared remote sensing. Summary of the Invention

[0009] The purpose of this invention is to solve the problems of background dynamic drift, laboratory calibration coefficient failure and uncertain emissivity of field calibration sources caused by environmental interference in infrared cameras during UAV field flight experiments. It provides an infrared calibration method and device based on dynamic background model and emissivity self-correction, aiming to achieve high-precision quantitative inversion of infrared radiation energy under complex field environments.

[0010] The present invention adopts the following technical solution, and the specific steps are as follows:

[0011] An infrared calibration method based on a dynamic background model and emissivity self-calibration includes the following steps:

[0012] S1: Based on the characteristics of stable gain and background drift with thermal equilibrium of cooled infrared cameras, an infrared radiation calibration model with dynamic background term is constructed by combining the camera spectral response function and the spectral radiance of the physical anchor point preset at the stable detection lower limit temperature.

[0013] S2: At discrete time points during the UAV's field flight, collect image grayscale, measured temperature, and flight time of multiple temperature calibration plates at the take-off and landing points and natural water bodies in the flight area. Convert the measured temperature into effective spectral radiance by weighted integration of Planck's law and spectral response.

[0014] S3: Construct a calibration plate emissivity grid sequence, calculate the initial response slope and initial instantaneous background using the calibration sample point with the largest temperature difference, then fit the global optimal response slope using the least squares method, recalculate the instantaneous background at each time, and lock the optimal emissivity and corresponding calibration parameters using water body inversion error closure loop;

[0015] S4: Based on the physical characteristics of camera thermal balance, the discrete-time background is fitted into a negative exponential dynamic background evolution model to achieve continuous prediction of background drift.

[0016] S5: Substitute the grayscale of the target image at any time and the corresponding dynamic background into the infrared radiometric calibration model to complete high-precision infrared radiometric inversion.

[0017] An infrared calibration device based on a dynamic background model and emissivity self-calibration includes the following modules:

[0018] Data acquisition module: Used to acquire the grayscale values ​​of images of the calibration reference and verification source captured by the infrared camera at discrete time points during UAV flight experiments. Simultaneously record the measured temperatures of the calibration reference and the verification source, as well as the data acquisition time;

[0019] Parameter initialization module: Used to select the two sets of calibration sample points with the largest temperature difference from the data acquisition module, and calculate the initial response slope through linear regression. ; and based on the preset physical anchor point spectral radiance Increase the spectral radiance of the target The initial instantaneous background response at that moment is obtained by reverse calculation. This module is also used to call the built-in spectral response weighted integral algorithm to convert the acquired measured temperature into effective radiance;

[0020] Emittance self-calibration and global optimization module: used to construct the emittance of the calibration board. The grid sequence, and for each The values, combined with the initial parameters, are used to fit all observed samples using the least squares method to solve for the globally optimal response slope. Background response at each discrete time point Simultaneously, by utilizing the inversion error feedback from the verification source, the optimal emissivity is determined. and its corresponding calibration parameters;

[0021] Dynamic background modeling module: used to process the optimal background response sequence acquired at each discrete time point. As input, a negative exponential compensation model is established based on the physical properties of thermal equilibrium, and the result is fitted to obtain the time-dependent... A continuously changing background evolution function;

[0022] Radiometric Inversion and Correction Module: This module performs pixel-level processing on the raw images of the target object, retrieving the corresponding data from the Dynamic Background Modeling module based on the image acquisition timestamp. Value, and combined with the globally optimal response slope This converts the image grayscale values ​​into a corrected image of the true target spectral radiance.

[0023] An electronic device includes: one or more processors; and a memory for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors cause the one or more processors to implement the method.

[0024] A computer-readable storage medium having executable instructions stored thereon, which, when executed by a processor, cause the processor to implement the method described thereon.

[0025] The present invention has the following beneficial effects:

[0026] This invention addresses the calibration failure problem caused by dynamic background drift of infrared cameras in UAV field experiments, proposing an infrared calibration method and device based on a dynamic background model and emissivity self-calibration. This invention introduces physical anchor point spectral radiance... Redefining the background response In conjunction with the physical characteristic that cooled infrared cameras tend to reach thermal equilibrium over time, a method is adopted... The model is continuously fitted. Compared to traditional calibration methods that treat the background as a constant, this invention can accurately capture and compensate for background drift throughout the entire operation, significantly improving the accuracy of radiometric inversion in complex field environments.

[0027] This invention achieves online self-calibration of the emissivity of an external calibration reference through a closed-loop mechanism of grid search and feedback verification from natural water bodies. Based on Planck's law, this invention effectively solves the source-end error caused by damage to the surface characteristics of the calibration plate or environmental interference in the field by constructing an emissivity grid and combining it with natural isothermal sources such as lake water bodies for verification.

[0028] This invention combines the working cycle of UAV battery swapping observation with the least squares method to integrate redundant data for global parameter optimization, thereby improving calibration accuracy and stability without increasing the additional load. Attached Figure Description

[0029] Figure 1 The flowchart shows the infrared calibration method based on dynamic background model and emissivity self-calibration.

[0030] Figure 2 The background at discrete times is fitted to a scatter plot based on a thermal equilibrium physical property model. Detailed Implementation

[0031] To make the objectives, technical solutions, and advantages of this invention clearer, the 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 merely illustrative of the invention and are not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of the invention described below can be combined with each other as long as they do not conflict with each other. To achieve the above objectives, the invention adopts the following technical solutions.

[0032] This invention provides an infrared calibration method based on a dynamic background model and emissivity self-calibration, comprising the following steps:

[0033] S1: Establish an infrared radiation calibration model based on physical anchor points

[0034] Taking into account the stable gain and easily shifted background of cooled infrared cameras with thermal equilibrium, an infrared radiation calibration model is constructed:

[0035] ;

[0036] in, The target theoretical spectral radiance; Output grayscale values ​​to the camera; The slope of the response to be solved is the reciprocal of the gain; To determine the radiant energy and spectral response function corresponding to the lower limit temperature for stable camera operation. Preset physical anchor point spectral radiance; For runtime The location corresponds to Camera background response value at energy level.

[0037] Function: By introducing physical anchor points This transforms the complex background drift into a response increment relative to a fixed energy reference, giving the calibration parameters a clear physical meaning.

[0038] S2: Calibration data acquisition at discrete time points in the field

[0039] During the flight of the UAV, images of the natural lake water body in the flight area are acquired as a verification source. During the intervals between multiple battery swaps of the UAV, the specially made high emissivity calibration plates at different temperatures located at the take-off and landing points are observed from a vertical perspective.

[0040] Synchronously record each running time Contact temperature of the lower calibration plate Corresponding image grayscale and the measured temperature of the water body With grayscale .

[0041] The conversion relationship between the measurement temperature and the effective spectral radiance needs to be established in advance, and Planck's law is used to calculate the blackbody at absolute temperature. Spectral radiance below Combined with the pre-acquired relative spectral response function of the infrared camera , The wavelength is within the camera's effective wavelength range. The effective spectral radiance is calculated by performing a weighted integral of the spectral response within the internal spectral response. The calculation formula is as follows:

[0042] ;

[0043] S3: Emissivity self-calibration and parameter calculation based on grid search and water body feedback

[0044] Mesh construction: Setting the calibrator emissivity Candidate grid sequence;

[0045] Initial instantaneous background Calculate: at the current running time Select the two calibration sample points with the largest difference in temperature readings. and ,in and These refer to the grayscale values ​​of the images output by the infrared camera observing the high-temperature and low-temperature calibration plates, respectively. and This refers to using Planck's law and the camera's spectral response function based on the measured contact temperature of high-temperature and low-temperature calibration plates. The calculated theoretical spectral radiance.

[0046] A preliminary linear fit is performed using the two-point method, and the initial response slope at that moment is calculated. :

[0047] ;

[0048] Let the target theoretical spectral radiance Equal to the spectral radiance of the physical anchor point Substituting into the formula, we can deduce the initial instantaneous background at that moment. :

[0049] ;

[0050] By minimizing random noise interference at the point of maximum temperature difference, initial parameters are provided for subsequent global fitting, ensuring the physical anchor point. Align with camera response benchmark.

[0051] Global response slope Least squares solution: Using Planck's law and the camera's spectral response function to measure temperature Converted to spectral radiance This summarizes all observations of different temperature calibration plates during multiple UAV battery swapping intervals at a specific emissivity. The following observation point data, the first The data set's runtime, image grayscale values, and true spectral radiance. Minimize the sum of squared residuals using the least squares method :

[0052] ;

[0053] By solving this system of equations, the reciprocal of the most robust global response gain under the environmental conditions of this voyage can be obtained. .

[0054] Instantaneous background Recalculation: using the established Regarding runtime Combining n sets of different temperature calibration plate data observed at various times { }, recalculate the time relative to The baseline response residuals are weighted and averaged to obtain the corresponding baseline scalar at that moment. :

[0055] ;

[0056] Emissivity self-calibration closed-loop water body verification: using the calibration model and the running time of the UAV passing over the water body. , Inverting the spectral radiance of lake water , through grid traversal Since the emissivity of water is close to 1, when the inversion results are consistent with... Theoretical values ​​of water spectral radiance converted using Planck's law and the camera's spectral response function Error At its minimum, the optimal emission rate is locked. and its corresponding slope and each observation time .

[0057] Function: It solves the error caused by changes in the surface characteristics of the calibration plate in the field environment, and ensures the accuracy of the model input source.

[0058] S4: Dynamic background evolution modeling based on thermal equilibrium physical properties

[0059] The i-th discrete time point in each of the m (m≥3) takeoffs and landings of the UAVs Acquired As a sample { The background evolution function is fitted using the nonlinear least squares method:

[0060] ;

[0061] in, For the initial background response, For the increasing trend of thermal equilibrium, This is the thermal equilibrium time constant.

[0062] Function: This model conforms to the physical law that infrared cooled cameras tend to thermal equilibrium as the operation time increases. By describing the saturation growth process of the camera's internal energy over time, it accurately compensates for the environmental heat accumulation error caused by the operation time of the UAV and realizes continuous prediction of the background drift within the calibration gap.

[0063] S5: Application of Full Radiation Inversion

[0064] For any runtime Acquired target image Substitute The function calculates the corrected true spectral radiance image through a calibration model.

[0065] This invention also provides an infrared calibration device based on a dynamic background model and emissivity self-calibration, comprising the following modules:

[0066] 1. Data Acquisition Module:

[0067] Used to acquire the image grayscale values ​​of the calibration reference and verification source captured by the infrared camera at discrete time points in UAV flight experiments. It also records the measured temperatures of the calibration reference and the verification source, as well as the time of data acquisition, simultaneously.

[0068] 2. Parameter initialization module:

[0069] This is used to select the two sets of calibration sample points with the largest temperature difference from the data acquisition module, and to calculate the initial response slope through linear regression. ; and based on the preset physical anchor point spectral radiance Increase the spectral radiance of the target The initial instantaneous background response at that moment is obtained by reverse calculation. This module is also used to call the built-in spectral response weighted integral algorithm to convert the acquired measured temperature into effective radiance.

[0070] 3. Emissivity self-calibration and global optimization module:

[0071] Used to construct the emissivity of the calibration plate The grid sequence, and for each The values, combined with the initial parameters, are used to fit all observed samples using the least squares method to solve for the globally optimal response slope. Background response at each discrete time point Simultaneously, by utilizing the inversion error feedback from the verification source (lake water), the optimal emissivity is determined. And its corresponding calibration parameters.

[0072] 4. Dynamic Background Modeling Module:

[0073] Used to obtain the optimal background response sequence at each discrete time point As input, a negative exponential compensation model is established based on the physical properties of thermal equilibrium. The fitting yielded the result over time. A continuously changing background evolution function.

[0074] 5. Radiation Inversion and Correction Module:

[0075] This is used for pixel-level processing of the raw image of the target under test, and calls the corresponding data from the dynamic background modeling module based on the image acquisition timestamp. Value, and combined with the globally optimal response slope This converts the image grayscale values ​​into a corrected image of the true target spectral radiance. Specific Implementation

[0076] like Figure 1 As shown: This invention proposes an infrared calibration method based on a dynamic background model and emissivity self-calibration, comprising the following steps:

[0077] S1: Based on the spectral response function of the infrared camera and the camera's nominal stable detection lower limit temperature, preset physical anchor point spectral radiance. In this embodiment, the lower limit temperature for stable detection by the infrared camera is taken. The corresponding spectral radiance is Construct the scaling equation:

[0078] ;

[0079] in, The target theoretical spectral radiance; Output grayscale values ​​to the camera; The slope of the response to be solved is the reciprocal of the gain; For the camera during runtime Detected at the location Real-time grayscale response at energy levels.

[0080] S2: The drone performs flight missions along a preset route. Every 40 minutes or so, the drone returns to the take-off and landing point to change batteries, completing calibration source and verification source observations and time recording.

[0081] S21 Calibration Source Observation: The UAV hovers above the take-off and landing point, observing two calibration plates with different water temperatures from a vertical, top-down perspective (e.g., one set of temperatures corresponds to different image grayscale values). , , , ).

[0082] S22 Verification Source Observation: Acquiring images of natural lakes within the experimental area during flight. And record the measured water temperature. .

[0083] S23 Time Recording: Synchronously records the running time of each group of data acquisition moments. .

[0084] S24 In order to realize the measurement of temperature using Planck's law and the camera's spectral response function. Converted to spectral radiance To achieve a precise mapping from measured temperature to the effective spectral radiance detected by an infrared camera, this invention employs a weighted integral method based on the spectral response for calculation. The specific processing steps are as follows:

[0085] S241 calculates spectral radiance based on Planck's law: First, Planck's law is used to calculate the blackbody's spectral radiance at absolute temperature. Theoretical spectral radiance :

[0086] ;

[0087] in, is Planck's constant. At the speed of light, Boltzmann's constant, λ is the wavelength.

[0088] Interpolation and normalization of the S242 spectral response function:

[0089] The spectral response function is constructed by pre-measuring the relative spectral response function of the infrared camera. To ensure integration accuracy, within the camera's effective wavelength range... A uniform spectral grid is established within the grid, and the original spectral response data is resampled using a linear interpolation algorithm to obtain a continuous spectral response distribution.

[0090] Weighted integral calculation of effective spectral radiance of S243:

[0091] spectral radiance With relative spectral response function Couple the parameters and calculate the effective spectral radiance over the entire camera response bandwidth. Its mathematical expression is the ratio of the weighted integral of the spectral radiance to the integral of the spectral response function:

[0092] ;

[0093] This process is achieved through numerical integration, which transforms continuous integration into a summation operation on a discrete grid, thereby obtaining the true value of effective spectral radiance that has a physical correspondence with the grayscale value output by the infrared camera.

[0094] S3: Parameter calculation and emissivity self-correction based on grid search and water body feedback;

[0095] S31 Mesh Construction: Setting Emissivity exist A mesh is constructed within the range with a step size of 0.005.

[0096] S32 Initial Instantaneous Background Calculation: At the current calibration time, select the two calibration sample points with the largest difference in temperature measurement values. and A preliminary linear fit is performed using the two-point method to calculate the initial response slope at that moment. :

[0097] ;

[0098] Increase the spectral radiance of the target Equal to the spectral radiance of the physical anchor point Substituting into the formula, we can deduce the initial instantaneous background at that moment. :

[0099] ;

[0100] S33 Global Response Slope Least squares solution: Using Planck's law and the camera's spectral response function to measure temperature Converted to spectral radiance This summarizes all observations of different temperature calibration plates during multiple UAV battery swapping intervals at a specific emissivity. The following observation point data Minimize the sum of squared residuals using the least squares method :

[0101] ;

[0102] By solving this system of equations, the reciprocal of the most robust global response gain under the environmental conditions of this voyage can be obtained. .

[0103] S34 Instantaneous Background Recalculation: Utilizing the already determined globally optimal slope For drone take-off and landing calibration nodes Combining n sets of different temperature calibration plate data observed at various times { }, recalculate the time relative to The baseline response residuals are weighted and averaged to obtain the corresponding baseline scalar at that moment. :

[0104] ;

[0105] S35 Emissivity Self-Calibration Closed-Loop Water Body Validation: Utilizing Calibration Model and Drone Passage Time over Water Bodies , Inverting the spectral radiance of lake water , through grid traversal When the inversion results are consistent with Theoretical values ​​of water spectral radiance converted using Planck's law and the camera's spectral response function Error At its minimum, the optimal emission rate is locked. and its corresponding slope and each observation time .

[0106] S4: Dynamic background evolution modeling based on thermal equilibrium physical properties

[0107] The calculated discrete points { Substituting into the thermal equilibrium model:

[0108] ;

[0109] Least squares fitting yields an analytical expression describing the continuous change of the background over time.

[0110] S5: Application of Full Radiation Inversion

[0111] For any time The target of the shooting The fitted at that moment To compensate, substitute the values ​​into the model to calculate the spectral radiance of the target:

[0112] ;

[0113] It is applied to output a physically consistent true spectral radiance image for all pixels of the entire image.

[0114] Figure 2 This is a schematic diagram of fitting the thermal equilibrium physical property model of the discrete-time background response values ​​in an embodiment of the present invention. The horizontal axis represents the operating time of the UAV infrared camera, in minutes; the vertical axis represents the instantaneous background grayscale value. The scattered points in the figure represent the background response of each discrete calibration node calculated through step S3. The solid line is the continuous evolution curve obtained by fitting the thermal equilibrium model in step S4. This figure intuitively shows the physical process by which the background response gradually tends to the thermal equilibrium saturation state as the running time increases.

Claims

1. An infrared calibration method based on a dynamic background model and emissivity self-calibration, characterized in that, Includes the following steps: S1: Based on the characteristics of stable gain and background drift with thermal equilibrium of cooled infrared cameras, an infrared radiation calibration model with dynamic background term is constructed by combining the camera spectral response function and the spectral radiance of the physical anchor point preset at the stable detection lower limit temperature. S2: At discrete time points during the UAV's field flight, collect image grayscale, measured temperature, and flight time of multiple temperature calibration plates at the take-off and landing points and natural water bodies in the flight area. Convert the measured temperature into effective spectral radiance by weighted integration of Planck's law and spectral response. S3: Construct a calibration plate emissivity grid sequence, calculate the initial response slope and initial instantaneous background using the calibration sample point with the largest temperature difference, then fit the global optimal response slope using the least squares method, recalculate the instantaneous background at each time, and lock the optimal emissivity and corresponding calibration parameters using water body inversion error closure loop; S4: Based on the physical characteristics of camera thermal balance, the discrete-time background is fitted into a negative exponential dynamic background evolution model to achieve continuous prediction of background drift. S5: Substitute the grayscale of the target image at any time and the corresponding dynamic background into the infrared radiometric calibration model to complete high-precision infrared radiometric inversion.

2. The infrared calibration method based on a dynamic background model and emissivity self-calibration according to claim 1, characterized in that, Step S1 is as follows: The infrared radiation calibration model is as follows: ； in, The target theoretical spectral radiance; Output grayscale values ​​to the camera; The slope of the response to be solved is the reciprocal of the gain; To determine the radiant energy and spectral response function corresponding to the lower limit temperature for stable camera operation. Preset physical anchor point spectral radiance; For runtime The location corresponds to Camera background response value at energy level.

3. The infrared calibration method based on a dynamic background model and emissivity self-calibration according to claim 2, characterized in that, Step S2 is as follows: During the flight of the UAV, images of the natural lake water body in the flight area were acquired as a verification source, and during the intervals between multiple battery swaps of the UAV, vertical top-down observations were conducted on n calibration plates at different temperatures located at the take-off and landing points. Synchronously record each running time Contact temperature of the lower calibration plate Corresponding image grayscale and the measured temperature of the water body With grayscale .

4. The infrared calibration method based on a dynamic background model and emissivity self-calibration according to claim 3, characterized in that, Step S2 further includes: A conversion relationship between temperature and effective spectral radiance is established beforehand, and Planck's law is used to calculate the blackbody at absolute temperature. Spectral radiance below Combined with the pre-acquired relative spectral response function of the infrared camera , The wavelength is within the camera's effective wavelength range. The effective spectral radiance is calculated by performing a weighted integral of the spectral response within the internal spectral response. The calculation formula is as follows: 。 5. The infrared calibration method based on a dynamic background model and emissivity self-calibration according to claim 4, characterized in that, Step S3 is as follows: Setting the emissivity of the calibration plate Candidate grid sequence; At the current runtime Select the two calibration sample points with the largest difference in temperature readings. and ,in and These refer to the grayscale values ​​of the images output by the infrared camera observing the high-temperature and low-temperature calibration plates, respectively. and This refers to using Planck's law and the camera's spectral response function based on the measured contact temperature of high-temperature and low-temperature calibration plates. The calculated theoretical spectral radiance; A preliminary linear fit is performed using the two-point method, and the initial response slope at that moment is calculated. : ； Let the target theoretical spectral radiance Equal to the spectral radiance of the physical anchor point Substituting into the formula, we can deduce the initial instantaneous background at that moment. : ； Temperature measurement is performed using Planck's law and the camera's spectral response function. Converted to spectral radiance This summarizes the emissivity of different temperature calibration plates observed during multiple UAV battery swapping intervals. The following observation point data, the first The data set's runtime, image grayscale values, and true spectral radiance. Minimize the sum of squared residuals using the least squares method : ； By solving this system of equations, the reciprocal of the most robust global response gain under the environmental conditions of this voyage can be obtained. ; Utilize the established Regarding runtime Combining n sets of different temperature calibration plate data observed at various times { }, recalculate the time relative to The baseline response residuals are weighted and averaged to obtain the corresponding baseline scalar at that moment. : ； Using calibration models and the running time of UAVs passing over water bodies , Inverting the spectral radiance of lake water Emittance by traversing the grid When the inversion results are consistent with Theoretical values ​​of water spectral radiance converted using Planck's law and the camera's spectral response function Error At its minimum, the optimal emission rate is locked. and its corresponding slope and each observation time .

6. The infrared calibration method based on a dynamic background model and emissivity self-calibration according to claim 5, characterized in that, Step S4 is as follows: The i-th discrete time point in each of the m takeoffs and landings of the UAV Acquired As a sample { The background evolution function is fitted using the nonlinear least squares method: ； in, For the initial background response, For the increasing trend of thermal equilibrium, This is the thermal equilibrium time constant.

7. The infrared calibration method based on a dynamic background model and emissivity self-calibration according to claim 6, characterized in that, Step S6 is as follows: For any runtime Acquired target image Substitute The function calculates the corrected true spectral radiance image through a calibration model.

8. An infrared calibration device based on a dynamic background model and emissivity self-calibration, characterized in that, Includes the following modules: Data acquisition module: Used to acquire the grayscale values ​​of images of the calibration reference and verification source captured by the infrared camera at discrete time points during UAV flight experiments. Simultaneously record the measured temperatures of the calibration reference and the verification source, as well as the data acquisition time; Parameter initialization module: Used to select the two sets of calibration sample points with the largest temperature difference from the data acquisition module, and calculate the initial response slope through linear regression. ; and based on the preset physical anchor point spectral radiance Increase the spectral radiance of the target The initial instantaneous background response at that moment is obtained by reverse calculation. This module is also used to call the built-in spectral response weighted integral algorithm to convert the acquired measured temperature into effective radiance; Emittance self-calibration and global optimization module: used to construct the emittance of the calibration board. The grid sequence, and for each The values, combined with the initial parameters, are used to fit all observed samples using the least squares method to solve for the globally optimal response slope. Background response at each discrete time point Simultaneously, by utilizing the inversion error feedback from the verification source, the optimal emissivity is determined. and its corresponding calibration parameters; Dynamic background modeling module: used to process the optimal background response sequence acquired at each discrete time point. As input, a negative exponential compensation model is established based on the physical properties of thermal equilibrium, and the result is fitted to obtain the time-dependent... A continuously changing background evolution function; Radiometric Inversion and Correction Module: This module performs pixel-level processing on the raw images of the target object, retrieving the corresponding data from the Dynamic Background Modeling module based on the image acquisition timestamp. Value, and combined with the globally optimal response slope This converts the image grayscale values ​​into a corrected image of the true target spectral radiance.

9. An electronic device, characterized in that, include: One or more processors; A memory for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors cause the one or more processors to implement the method of any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, It stores executable instructions that, when executed by a processor, cause the processor to perform the method described in any one of claims 1 to 7.

Images