A satellite infrared channel radiometric calibration method and system

Abstract

This invention discloses a satellite infrared channel radiometric calibration method and system, belonging to the field of equipment calibration technology. The method includes: scanning a deep-space scene and a blackbody at different temperatures using a satellite scanner to obtain deep-space DN values ​​and blackbody DN values; determining the radiometric bias of the satellite infrared channel using a matrix factorization algorithm; establishing a radiometric transfer equation reflecting the relationship between the DN values ​​obtained from the satellite infrared channel scan and the radiance based on the radiometric bias; determining the maximum order of the radiometric transfer equation based on the obtained blackbody DN values ​​using a ridge regression algorithm; updating the radiometric transfer equation based on the maximum order and determining the correction coefficients of each order of the updated radiometric transfer equation using the least squares method; updating the radiometric transfer equation using the scanning angle polarization factor caused by the satellite scanner to correct the radiant intensity; and outputting the updated radiometric transfer equation to complete the satellite infrared channel radiometric calibration. This improves the accuracy of satellite observation.

Description

Technical Field

[0001] This invention belongs to the field of equipment calibration technology, specifically relating to a satellite infrared channel radiometric calibration method and system. Background Technology

[0002] DN (Digital Number) is short for digital count value. It is the digital output value obtained by the sensor after receiving light or radiation signals and processing them through electronic equipment. In satellite remote sensing, it is the numerical value converted from the actual physical quantity (such as radiation intensity) by the sensor through a quantization and digitization process. Radiance intensity refers to the radiant energy flux density per unit area within a unit solid angle, usually used to describe the intensity of light or electromagnetic waves. It is the actual physical quantity received by the satellite infrared channel, reflecting the radiation characteristics of the target object in a certain wavelength band. Satellite infrared channel radiometric calibration refers to accurately determining the relationship between the digital count value (DN value) received by the satellite infrared detector and the actual radiance through mathematical modeling and experimental calibration. This process is a key step in ensuring the reliability and accuracy of satellite observation data, especially for fields requiring high-precision observation, such as weather forecasting, environmental monitoring, and remote sensing.

[0003] Radiometric calibration is a crucial step in satellite remote sensing to ensure the accuracy and consistency of measurement data. Over long-term operation, sensors are affected by factors such as electronic offset, temperature variations, and aging, leading to deviations in measurement data. Radiometric calibration accurately establishes the relationship between DN values ​​and actual radiance, thereby eliminating systematic errors and ensuring the reliability of satellite observation data. This provides high-precision support for applications such as environmental monitoring and weather forecasting.

[0004] In existing technologies, during radiometric calibration, the fitting order between the DN value and the actual radiance is often determined subjectively, or a low-order fitting model is directly selected. This results in the fitting model being unable to accurately describe the relationship between the DN value and the actual radiance, leading to low fitting accuracy. Furthermore, the high-order terms are ignored in their detailed description of the radiometric relationship, resulting in a large deviation in the final obtained radiance of the scanned area. Consequently, the accuracy of satellite observation data is low, affecting the normal operation of the satellite. Summary of the Invention

[0005] To address the technical problem in existing technologies where the order of the fitting between the DN value and the actual radiance is often subjectively determined during radiometric calibration, or a low-order fitting model is directly selected, resulting in the fitting model failing to accurately describe the relationship between the DN value and the actual radiance, leading to low fitting accuracy, neglecting the fine description of the radiation relationship by higher-order terms, and causing a large deviation in the final obtained radiance of the scanned area, thus resulting in low accuracy of satellite observation data and affecting the normal operation of the satellite, this invention provides a satellite infrared channel radiometric calibration method and system.

[0006] First aspect

[0007] This invention provides a satellite infrared channel radiometric calibration method, comprising:

[0008] S1: Scan deep space scenes and blackbodies at different temperatures using a satellite scanner to obtain multiple deep space DN values ​​and multiple blackbody DN values;

[0009] S2: Combining matrix factorization algorithm, determine the radiation bias of satellite infrared channel based on the obtained deep space DN value;

[0010] S3: Combine radiation bias to establish a radiation transfer equation that reflects the relationship between the DN value obtained from satellite infrared channel scanning and radiance;

[0011] S4: Based on the obtained blackbody DN value, the maximum order of the radiative transfer equation is determined by combining the ridge regression algorithm;

[0012] S5: Update the radiative transfer equation based on the maximum order, and determine the correction coefficients of each order of the updated radiative transfer equation using the least squares method.

[0013] S6: Update the radiative transfer equation using the scanning angle polarization factor caused by the satellite scanner to correct the radiative intensity;

[0014] S7: Output the updated radiative transfer equation and end the radiometric calibration of the satellite's infrared channel.

[0015] Second aspect

[0016] The present invention provides a satellite infrared channel radiometric calibration system for performing the satellite infrared channel radiometric calibration method in the first aspect.

[0017] Compared with the prior art, the present invention has at least the following beneficial technical effects:

[0018] In this invention, a deep-space scene with zero radiation intensity and blackbodies at different temperatures are first scanned using a satellite scanner to obtain deep-space DN values ​​and blackbodies DN values. Then, based on this determined data, a matrix factorization algorithm is used to determine the radiation bias of the satellite infrared channel based on the obtained deep-space DN values. This accurately removes the influence of electronic offsets and anomalous noise, thereby improving the accuracy and stability of the satellite infrared channel radiometric calibration and ensuring data reliability. The maximum order of the radiative transfer equation is determined using a ridge regression algorithm, effectively avoiding excessive fitting errors caused by subjective selection of the order, improving the accuracy and robustness of the radiometric calibration, and simultaneously resolving overfitting or underfitting phenomena, ensuring more accurate calculation of the satellite infrared channel radiance. Finally, the correction coefficients for each order of the updated radiative transfer equation are accurately determined using the least squares method. Then, the radiation intensity was corrected by the scanning angle polarization factor caused by the satellite scanner. Finally, the radiation transfer equation reflecting the relationship between DN value and radiation intensity was accurately fitted. It can accurately and efficiently obtain the radiation intensity of the satellite scanning area. By combining deep space scene, blackbody data, matrix factorization algorithm and ridge regression algorithm, the influence of electronic offset and noise is accurately removed. At the same time, the optimal order of the radiation transfer equation is adaptively determined to avoid fitting error and improve the accuracy and stability of radiometric calibration. The combination of least squares method and scanning angle correction ensures that the calculation of satellite infrared channel radiation intensity is accurate and reliable. Attached Figure Description

[0019] The preferred embodiments will now be described in a clear and easy-to-understand manner, in conjunction with the accompanying drawings, to further explain the above-mentioned characteristics, technical features, advantages, and implementation methods of the present invention.

[0020] Figure 1 This is a flowchart illustrating a satellite infrared channel radiometric calibration method provided by the present invention;

[0021] Figure 2 This is a schematic diagram of the structure of a satellite infrared channel radiometric calibration system provided by the present invention. Detailed Implementation

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the specific implementation methods of the present invention will be described below with reference to the accompanying drawings. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings and other implementation methods can be obtained based on these drawings without any creative effort.

[0023] To keep the drawings concise, each figure only schematically shows the parts relevant to the invention, and these do not represent the actual structure of the product. Furthermore, to facilitate understanding, in some figures, only one of components with the same structure or function is schematically depicted, or only one is labeled. In this document, "one" not only means "only one," but can also mean "more than one."

[0024] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0025] In this document, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to fixed connections, detachable connections, or integral connections; they can refer to mechanical connections or electrical connections; they can refer to direct connections or indirect connections through an intermediate medium; and they can refer to the internal communication between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0026] Furthermore, in the description of this invention, the terms "first," "second," etc., are used only for distinguishing descriptions and should not be construed as indicating or implying relative importance.

[0027] Example 1

[0028] In one embodiment, refer to the appendix to the specification. Figure 1 The diagram shows a flowchart of the satellite infrared channel radiometric calibration method provided by the present invention.

[0029] This invention provides a satellite infrared channel radiometric calibration method, comprising:

[0030] S1: Use a satellite scanner to scan deep space scenes and blackbodies at different temperatures to obtain multiple deep space DN values ​​and multiple blackbody DN values.

[0031] Satellite scanners are remote sensing sensor devices installed on satellites to scan Earth, deep space, or other target scenes. They receive electromagnetic radiation and convert it into digital signals (DN values) for scientific measurement and analysis. Deep space scenes refer to the empty cosmic background without stars or other radiation sources. Their radiation intensity is close to zero, and they contain almost no thermal radiation, making them ideal reference "zero-radiation" scenes used to calibrate the electronic offsets of sensors. Blackbodies are ideal radiators that completely absorb and emit thermal radiation; their radiation intensity depends only on their own temperature. The radiation of a blackbody at different temperatures follows known mathematical relationships (such as Planck's law), making it a standard reference source in radiometric calibration.

[0032] It should be noted that satellite scanners acquire the DN values ​​of both deep space and blackbody radiation by scanning deep-space scenes with near-zero radiation intensity and blackbody radiation at different temperatures. The deep-space data is used to determine the sensor's electronic offset (radiative bias), while the blackbody data provides a standard reference for temperature and radiation intensity. Deep-space data provides a "zero baseline" for calibration, while blackbody data provides a known physical standard for subsequent radiance calculations. This process is fundamental to radiometric calibration, ensuring data reliability and accuracy.

[0033] S2: Combining matrix factorization algorithm, determine the radiation bias of satellite infrared channel based on the obtained deep space DN value.

[0034] Matrix factorization algorithms are mathematical methods that decompose a matrix into multiple sub-matrices with specific properties. Satellite infrared channels refer to specific channels in the infrared band through which satellite remote sensing sensors receive radiation signals from the Earth's surface or atmosphere. Infrared channels can monitor the temperature and radiation intensity of objects and are widely used in meteorology, environmental monitoring, and other fields. Radiation bias refers to the non-zero fundamental signal introduced by the sensor during observation due to noise in the electronic system itself or hardware characteristics. Even in the absence of radiation input (such as in deep space scenarios), a certain digital count value (DN value) will still be generated. Radiation bias needs to be removed through calibration to ensure the accuracy of the observation data.

[0035] Specifically, based on the deep-space DN values ​​acquired by the satellite scanner, a matrix factorization algorithm (such as RPCA) is used to decompose the deep-space DN value matrix into a low-rank matrix and a sparse matrix. The low-rank matrix represents the radiation bias of the satellite's infrared channel, i.e., the reference offset introduced by the electronic system, while the sparse matrix reflects the influence of anomalous noise. This decomposition method allows for the accurate determination of the radiation bias while removing noise interference, ensuring the accuracy and stability of subsequent data processing. This step provides an accurate electronic offset reference for radiometric calibration, laying the foundation for high-precision calibration.

[0036] In one possible implementation, the matrix factorization algorithm is specifically the RPCA algorithm.

[0037] Among them, the RPCA algorithm (Robust Principal Component Analysis) is a mathematical method for processing data decomposition. It decomposes the original data matrix into two parts: a low-rank matrix and a sparse matrix. It is suitable for removing noise and outliers and is widely used in signal processing.

[0038] S2 specifically includes:

[0039] S201: Combine deep-space DN values ​​into a deep-space DN value matrix:

[0040]

[0041] Where D represents the deep space DN value matrix, N SV,k,m This represents the DN value corresponding to the m-th field of view in the k-th scan, where k = 1, 2, ..., N, m = 1, 2, ..., M, N represents the total number of scans, and M represents the total number of fields of view in the scan.

[0042] S202: Decomposition of the deep space DN value matrix using the RPCA algorithm:

[0043] min L,S ‖L‖ * +λ‖S‖1

[0044] D = L + S

[0045] Where L represents the low-rank matrix reflecting the radiation bias, and S represents the coefficient matrix reflecting the anomalous noise, ‖L‖ * Let L be the nuclear norm of L, which is the sum of the singular values ​​of L; let ||S||1 be the L1 norm of S; let λ be the regularization parameter that balances the weights of L and S; and let min be the null terminator. L,S ‖L‖ * +λ‖S‖1 represents taking the value such that ‖L‖ * +λ‖S‖1 takes the minimum values ​​of L and S.

[0046] It should be noted that the deep space DN value matrix is ​​decomposed into a low-rank matrix L and a sparse matrix S by the RPCA algorithm, where L represents the normal part of the radiation bias and S represents the noise anomalous component, effectively separating the normal radiation bias from the noise and improving the accuracy of the data.

[0047] In one possible implementation, the regularization parameter is calculated as follows:

[0048]

[0049] |λ (t+1) -λ (t) |<

[0050]

[0051] Where, λ (t) and λ (t+1) Let represent the regularization parameters obtained in the t-th and t+1-th iterations, respectively, represent the termination threshold, and η represent the learning rate. This indicates the partial derivative. This represents the RPCA decomposition and reconstruction error with respect to λ. express The gradient, ← indicates update.

[0052] It should be noted that the iterative optimization process dynamically adjusts parameter values ​​to ensure that the algorithm gradually approaches the optimal solution with each update. Introducing a learning rate and a termination threshold controls the convergence speed, avoiding adjustments that are too fast or too slow, thereby improving computational efficiency and stability. Partial derivative and gradient calculations make the parameter update process more accurate, adaptively balancing reconstruction error and regularization terms, reducing the complexity of manual parameter tuning, and ensuring the algorithm exhibits good robustness and stability across different data scales.

[0053] Optionally, ∈ can specifically be 10 -5 Or 10 -6 The learning rate η can be initially set to 0.01, and then decays exponentially to 0.001. The decomposition and reconstruction error is specifically as follows:

[0054]

[0055] Among them, || || F Let μ denote the Frobenius norm, and μ denote the control over the reconstruction error and the regularization term ‖L‖. * The balancing parameter between the weights +λ‖S‖1 is calculated as follows: By adaptively adjusting the parameters using the Frobenius norm and the maximum dimension of the data matrix, the weights of reconstruction error and regularization terms are automatically balanced according to the data size. This adaptive computation avoids the complexity of manual parameter tuning, improves the robustness and generalization ability of the algorithm, and ensures stable results under different data conditions, effectively improving the accuracy and efficiency of radiation bias extraction.

[0056] S203: Calculate the radiation bias by averaging the low-rank matrix for each field of view:

[0057]

[0058] Where, N SV L represents radiation bias. k,m This represents the element obtained from the m-th field of view during the k-th scan of the low-rank matrix L.

[0059] It should be noted that by minimizing the weighted sum of the nuclear norm and L1 norm, RPCA can effectively separate normal radiation bias from anomalous noise, improving data accuracy and stability. In S201, deep-space DN values ​​are organized into a matrix. In S202, the matrix is ​​decomposed using the RPCA algorithm. In S203, the mean of the low-rank matrix is ​​taken to obtain accurate radiation bias values, thus providing a stable and reliable data foundation for subsequent radiometric calibration. Here, the radiation bias is a digital count of the space field of view used to remove electronic offsets.

[0060] S3: Combine radiation bias to establish a radiation transfer equation that reflects the relationship between the DN value obtained from satellite infrared channel scanning and radiance.

[0061] The radiative transfer equation is a mathematical model describing the relationship between digital count (DN) values ​​and actual physical radiance. Based on the radiative bias determined in the previous step, the radiative transfer equation links the DN values ​​obtained from satellite infrared channel scanning with the actual radiance. This equation can compensate for the bias information of the sensor system and, through the introduction of multi-order correction coefficients, more accurately fits the nonlinear relationship between DN values ​​and radiance, ensuring the accuracy and reliability of calibration data and providing a foundation for subsequent radiance calculations.

[0062] In one possible implementation, the radiation transfer equation is specifically:

[0063] R = a0 + a1·D N +a2·D N 2 +…+a n ·D N n

[0064] a0 = N SV

[0065] Where, N sc a represents the radiance obtained from the scan. v D represents the correction coefficient of order v, where v = 1, 2, ..., n, and n represents the maximum order. N This represents the DN value obtained from the scan, where R indicates the relationship between DN and D. N The corresponding radiance.

[0066] It should be noted that by introducing a polynomial form, a precise relationship is established between the scanned digital count values ​​and the actual radiance. By using order correction coefficients and searching for the maximum order, the nonlinear relationship can be better fitted, compensating for sensor system biases and improving the accuracy and precision of radiometric calibration. Furthermore, the flexible polynomial order design gives the model stronger adaptability and generalization ability, making it suitable for different observation scenarios and data, ensuring more stable and reliable observation results.

[0067] S4: Based on the obtained blackbody DN value, the maximum order of the radiative transfer equation is determined by combining the ridge regression algorithm.

[0068] Ridge regression is a regression analysis method used to solve multicollinearity problems. By introducing an L2 regularization term (penalty term) into the loss function, it effectively limits the excessively large model coefficients, avoids overfitting, and improves the model's generalization ability. The maximum order refers to the highest order of the polynomial in the radiative transfer equation, used to describe the complex nonlinear relationship between DN values ​​and radiance. Determining the maximum order aims to balance fitting accuracy and model complexity, preventing overfitting or underfitting.

[0069] It should be noted that, based on the blackbody DN values ​​and corresponding temperatures obtained from scanning, ridge regression was used to fit radiative transfer equations of different orders, and the fitting error was calculated. By introducing a regularization penalty term, ridge regression can effectively control model complexity, avoid overfitting caused by high-order fitting, and determine the optimal maximum order. Finally, the performance of models of different orders was evaluated using the Akaike Information Criterion (AIC) and the Bayesian Information Criterion (BIC), and the maximum order with the smallest fitting error and highest stability was selected, thereby improving the accuracy and robustness of radiative calibration and ensuring more accurate radiance calculation.

[0070] In one possible implementation, S4 specifically includes:

[0071] S401: Calculate the true radiance corresponding to each blackbody DN value by combining the blackbody DN value and the temperature corresponding to that value.

[0072]

[0073] Among them, T i and R i Let λ represent the i-th temperature of the blackbody and the true radiance corresponding to the i-th temperature of the blackbody, respectively; exp represent the natural exponential function; C1 and C2 represent the first and second radiation constants, respectively; and λ represent the first and second radiation constants, respectively. 波长 This indicates the wavelength emitted by the satellite's infrared channel for observing the blackbody.

[0074] It should be noted that previous calibration methods often relied on subjective experience to select the maximum order or directly fitted the radiative transfer equation with a second-order function, ignoring significant fitting errors. This resulted in large deviations in the satellite's infrared channel radiometric calibration, making it impossible to accurately calculate radiance and consequently causing inaccurate detection of various parameters by the satellite. This proposed solution utilizes the ridge regression algorithm to fit the obtained DN value of a blackbody with known temperature. Before establishing the radiative transfer equation, it assesses the maximum order necessary for its establishment, avoiding excessive fitting errors caused by manually selecting the order.

[0075] S402: Substitute the blackbody DN value into the calculation of the radiation transfer equations at different orders to obtain the fitted radiance at different orders.

[0076] S403: Using the ridge regression algorithm, establish a loss function for the true radiance and the fitted radiance:

[0077]

[0078] Among them, L oss Let O represent the loss function, and O represent the total number of blackbody temperature changes. γ represents the fitted radiance obtained by fitting the blackbody DN value collected for the temperature of the i-th blackbody, and γ = 0.5 represents the penalty term coefficient.

[0079] S404: Determine the sum of squared residuals between the true radiance and the fitted radiance, with the goal of minimizing the loss function.

[0080] min RSS L oss

[0081]

[0082] Where RSS represents the sum of squared residuals, min RSS L oss This indicates taking the minimum L. oss The RSS feed below.

[0083] S405: Calculate the Akaike information criterion value and the Bayesian information criterion value for the sum of squared residuals:

[0084]

[0085] Where AIC and BIC represent the Akaike Information Criterion value and the Bayesian Information Criterion value, respectively, and log represents the logarithmic function.

[0086] S406: Calculate the ratio of the Akaike information criterion value to the Bayesian information criterion value, and output the order of the radiative transfer equation with the minimum ratio as the maximum order:

[0087]

[0088] in, Indicates taking the minimum n below.

[0089] Specifically, the true radiance of a blackbody is calculated based on Planck's formula using its temperature and density (DN) value, providing a physical standard reference value. Radiation transfer equations of different orders are used to fit the relationship between the blackbody's DN value and the true radiance, yielding fitting results for different orders. A ridge regression algorithm is used to introduce a loss function, combined with a regularization term, to balance the model's fitting accuracy and complexity, preventing overfitting. The residual sum of squares (RSS) is determined by minimizing the loss function, further evaluating the model's error. The Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC) are calculated to evaluate the performance of models of different orders. By comparing the ratio of AIC to BIC, the order corresponding to the smallest ratio is selected as the optimal maximum order output, ultimately determining the optimal model for the radiation transfer equation. This method, through automatic order selection, avoids fitting errors caused by subjective human factors, improving the accuracy, stability, and robustness of radiation calibration, and ensuring accurate calculation of radiance.

[0090] S5: Update the radiative transfer equation based on the maximum order, and determine the correction coefficients of each order of the updated radiative transfer equation using the least squares method.

[0091] The least squares method is a mathematical optimization technique that solves for model parameters by minimizing the sum of squared errors between observed and predicted values. Correction coefficients are the coefficients of polynomials of various orders in the radiative transfer equation, used to describe the relationship between digital count (DN) values ​​and radiance. Based on a determined maximum order, the radiative transfer equation is transformed into a matrix equation, and the least squares method is used to solve the equation to obtain the correction coefficients of each order. The least squares method ensures that the model fitting error is minimized, making the radiative transfer equation more accurately reflect the correspondence between DN values ​​and actual radiance, thereby improving calibration accuracy.

[0092] In one possible implementation, S5 specifically includes:

[0093] S501: Transform the radiation transfer equation into a matrix equation:

[0094] R = Xa

[0095]

[0096] Where a represents the correction coefficient vector, R represents the true radiance vector, and X represents the vector including D. N Design matrix for each term, This represents the DN value associated with the i-th temperature and order v of the blackbody.

[0097] S502: Solve the matrix equation using the least squares method to obtain the correction coefficient vector, i.e., the correction coefficients of each order:

[0098]

[0099] Wherein, the subscript T indicates transpose. Indicates taking The smallest a, This indicates that it has been launched.

[0100] It should be noted that the radiative transfer equation is transformed into a matrix equation, simplifying the polynomial fitting problem into a linear algebraic problem. Here, the actual radiance is treated as a vector, and the matrix is ​​designed to contain DN value terms of different orders. The correction coefficients to be determined are represented as vectors. This matrix equation is solved using the least squares method to minimize the error between the model's predicted values ​​and the actual radiance, thus obtaining the correction coefficients of each order. The least squares method can accurately fit the data with minimal error, improving the accuracy and stability of the radiative transfer equation. Through this method, the correction coefficients of each order of the radiative transfer equation are finally determined, making the relationship between DN values ​​and radiance more accurate and contributing to improved reliability of the calibration results.

[0101] S6: Update the radiative transfer equation using the scanning angle polarization factor caused by the satellite scanner to correct for radiative intensity.

[0102] The scanning angle polarization factor is a polarization effect caused by the structure and optical characteristics of the satellite scanner, which affects the measurement results of radiation intensity as the scanning angle changes. This polarization effect introduces systematic errors in the observation data at different scanning angles, requiring correction. Considering the polarization effect caused by the change in the satellite scanner's scanning angle, the radiation transfer equation is updated and corrected by introducing the scanning angle polarization factor. This step effectively eliminates radiation intensity errors at different scanning angles, ensuring that the final calculated radiation intensity remains consistent and accurate across all scanning angles, further improving data quality.

[0103] In one possible implementation, the corrected radiation intensity is specifically:

[0104] R′=R·[1+P(θ)cos 2 (θ)]

[0105] Where R′ represents the corrected radiation intensity, θ represents the scanning angle of the satellite scanner, P(θ) represents the scanning angular polarization factor related to the satellite structure, cos represents the cosine function, and R represents the polarization factor related to D. N The corresponding radiance.

[0106] The scanning angle polarization factor was obtained by conducting radiation response tests on the satellite scanner under ground-based laboratory conditions using a standard radiation source and a test device with a controllable scanning angle. By changing the scanning angle, the radiation signal received by the scanner was recorded and compared with the actual radiation intensity of the standard radiation source to calculate the polarization factor. It is understandable that changes in the scanning angle may lead to radiation measurement errors. By introducing a cosine function and the polarization factor to correct the observed data, the influence of angle deviation is effectively eliminated, ensuring the consistency and accuracy of radiation intensity at different scanning angles and improving the reliability of the measurement results.

[0107] S7: Output the updated radiative transfer equation and end the radiometric calibration of the satellite's infrared channel.

[0108] In one possible implementation, after S7, the following is also included:

[0109] After a preset time interval, repeat steps S1 to S7.

[0110] Understandably, the calibration process is repeated periodically to continuously update the radiation bias and correction coefficients, ensuring the accuracy of the satellite's infrared channel radiometric calibration, adapting to potential performance drift of the sensor over time, and ensuring that the observation data maintains stability and accuracy over the long term.

[0111] It should be noted that those skilled in the art can set the preset duration according to actual needs, and this invention does not limit this.

[0112] In one possible implementation, it also includes:

[0113] The target scene is scanned by a satellite scanner to obtain the target DN value.

[0114] Substitute the target DN value into the radiative transfer equation to determine the radiance of the target scene.

[0115] It should be noted that by scanning the target scene with a satellite scanner to obtain the target DN value, and then substituting it into a calibrated radiative transfer equation, the radiance of the target scene is calculated. This process establishes a precise correspondence between satellite observation data and actual physical quantities, ensuring the accuracy of radiance and providing reliable data support for the analysis and application of the radiation characteristics of the target scene. It is widely used in environmental monitoring and remote sensing analysis.

[0116] In practical applications, high-precision radiometric calibration of satellite infrared channels was achieved through systematic satellite infrared channel radiometric calibration. First, deep-space DN values ​​and blackbody DN values ​​were obtained by scanning deep-space scenes and blackbody at different temperatures, used for electronic offset correction and radiometric reference standards, respectively. A matrix factorization algorithm was used to accurately determine the radiometric bias and remove noise interference. A radiometric transfer equation relating DN values ​​to radiance was established to compensate for system bias information. The ridge regression algorithm was used to determine the optimal order, avoiding excessive fitting errors and improving model accuracy. The correction coefficients of the equation were solved using the least squares method to ensure minimal fitting errors. A scanning angle polarization factor was introduced to correct radiant intensity and eliminate scanning angle errors. Finally, an accurate radiometric transfer equation was output. This method, through scientific algorithms and parameter optimization, effectively improves the accuracy, stability, and robustness of radiometric calibration, ensuring the reliability of observational data.

[0117] Compared with the prior art, the present invention has at least the following beneficial technical effects:

[0118] In this invention, a deep-space scene with zero radiation intensity and blackbodies at different temperatures are first scanned using a satellite scanner to obtain deep-space DN values ​​and blackbody DN values. Then, based on this determined data, a matrix factorization algorithm is used to determine the radiation bias of the satellite infrared channel based on the obtained deep-space DN values. This accurately removes the influence of electronic offsets and anomalous noise, thereby improving the accuracy and stability of satellite infrared channel radiometric calibration and ensuring data reliability. The maximum order of the radiative transfer equation is determined using a ridge regression algorithm, effectively avoiding excessive fitting errors caused by subjective selection of the order, improving the accuracy and robustness of radiometric calibration, and resolving overfitting or underfitting phenomena to ensure more accurate calculation of satellite infrared channel radiance. Finally, the correction coefficients for each order of the updated radiative transfer equation are accurately determined using the least squares method. Then, the radiation intensity was corrected by the scanning angle polarization factor caused by the satellite scanner. Finally, the radiation transfer equation reflecting the relationship between DN value and radiation intensity was accurately fitted. It can accurately and efficiently obtain the radiation intensity of the satellite scanning area. By combining deep space scene, blackbody data, matrix factorization algorithm and ridge regression algorithm, the influence of electronic offset and noise is accurately removed. At the same time, the optimal order of the radiation transfer equation is adaptively determined to avoid fitting error and improve the accuracy and stability of radiometric calibration. The combination of least squares method and scanning angle correction ensures that the calculation of satellite infrared channel radiation intensity is accurate and reliable.

[0119] Example 2

[0120] Reference manual attached Figure 2 The diagram shows a structural schematic of a satellite infrared channel radiometric calibration system provided by the present invention.

[0121] The present invention also provides a satellite infrared channel radiometric calibration system 20, applied to the above-mentioned satellite infrared channel radiometric calibration method, comprising:

[0122] Processor 201.

[0123] The memory 202 stores computer-readable instructions, which, when executed by the processor 201, implement the satellite infrared channel radiometric calibration method as described in the method embodiment.

[0124] The satellite infrared channel radiometric calibration system 20 provided by the present invention can perform the above-mentioned satellite infrared channel radiometric calibration method and achieve the same or similar technical effects. To avoid repetition, the present invention will not elaborate further.

[0125] Compared with the prior art, the present invention has at least the following beneficial technical effects:

[0126] In this invention, a deep-space scene with zero radiation intensity and blackbodies at different temperatures are first scanned using a satellite scanner to obtain deep-space DN values ​​and blackbody DN values. Then, based on this determined data, a matrix factorization algorithm is used to determine the radiation bias of the satellite infrared channel based on the obtained deep-space DN values. This accurately removes the influence of electronic offsets and anomalous noise, thereby improving the accuracy and stability of satellite infrared channel radiometric calibration and ensuring data reliability. The maximum order of the radiative transfer equation is determined using a ridge regression algorithm, effectively avoiding excessive fitting errors caused by subjective selection of the order, improving the accuracy and robustness of radiometric calibration, and resolving overfitting or underfitting phenomena to ensure more accurate calculation of satellite infrared channel radiance. Finally, the correction coefficients for each order of the updated radiative transfer equation are accurately determined using the least squares method. Then, the radiation intensity was corrected by the scanning angle polarization factor caused by the satellite scanner. Finally, the radiation transfer equation reflecting the relationship between DN value and radiation intensity was accurately fitted. It can accurately and efficiently obtain the radiation intensity of the satellite scanning area. By combining deep space scene, blackbody data, matrix factorization algorithm and ridge regression algorithm, the influence of electronic offset and noise is accurately removed. At the same time, the optimal order of the radiation transfer equation is adaptively determined to avoid fitting error and improve the accuracy and stability of radiometric calibration. The combination of least squares method and scanning angle correction ensures that the calculation of satellite infrared channel radiation intensity is accurate and reliable.

[0127] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0128] The above embodiments merely illustrate several implementation methods of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. A method for radiometric calibration of a satellite infrared channel, characterized in that, include: S1: Scan deep space scenes and blackbodies at different temperatures using a satellite scanner to obtain multiple deep space DN values ​​and multiple blackbody DN values; S2: Combining matrix factorization algorithm, determine the radiation bias of the satellite infrared channel based on the obtained deep space DN value; S3: Based on the aforementioned radiation bias, establish a radiation transfer equation to reflect the relationship between the DN value obtained from satellite infrared channel scanning and the radiance; S4: Based on the obtained blackbody DN value, the maximum order of the radiative transfer equation is determined by combining the ridge regression algorithm; S5: Update the radiation transfer equation based on the maximum order, and determine the correction coefficients of each order of the updated radiation transfer equation using the least squares method. S6: Update the radiative transfer equation using the scanning angle polarization factor caused by the satellite scanner to correct the radiative intensity; S7: Output the updated radiative transfer equation and end the radiometric calibration of the satellite infrared channel.

2. The satellite infrared channel radiometric calibration method according to claim 1, characterized in that, The matrix factorization algorithm is specifically the RPCA algorithm; S2 specifically includes: S201: Combine the deep space DN values ​​into a deep space DN value matrix: Where D represents the deep space DN value matrix, N SV,k,m This represents the DN value corresponding to the m-th field of view in the k-th scan, where k = 1, 2, ..., N, m = 1, 2, ..., M, N represents the total number of scans, and M represents the total number of fields of view in the scan. S202: Decompose the deep space DN value matrix using the RPCA algorithm: minutes L,S "L" * +λ‖S‖1 D = L + S Where L represents the low-rank matrix reflecting the radiation bias, and S represents the coefficient matrix reflecting the anomalous noise, ‖L‖ * Let L be the nuclear norm of L, which is the sum of the singular values ​​of L; let ||S||1 be the L1 norm of S; let λ be the regularization parameter that balances the weights of L and S; and let min be the null terminator. L,S ‖L‖ * +λ‖S‖1 represents taking the value such that ‖L‖ * +λ‖S‖1 takes the minimum values ​​of L and S; S203: Calculate the radiation bias by averaging the low-rank matrix for each field of view: Where, N SV L represents radiation bias. k,m This represents the element obtained from the m-th field of view during the k-th scan of the low-rank matrix L.

3. The satellite infrared channel radiometric calibration method according to claim 2, characterized in that, The regularization parameter is calculated as follows: |l (t+1) -l (t) |<∈ Where, λ (t) and λ (t+1) Let represent the regularization parameters obtained in the t-th and t+1-th iterations, respectively, represent the termination threshold, and η represent the learning rate. This indicates the partial derivative. This represents the RPCA decomposition and reconstruction error with respect to λ. express The gradient, ← indicates update.

4. The satellite infrared channel radiometric calibration method according to claim 2, characterized in that, The radiation transfer equation is specifically as follows: R=a0+a1·D N +a2·D N 2 +…+a n ·D N n a0=N SV Where, N SV a represents the radiance obtained from the scan. v D represents the correction coefficient of order v, where v = 1, 2, ..., n, and n represents the maximum order. N This represents the DN value obtained from the scan, where R indicates the relationship between DN and D. N The corresponding radiance.

5. The satellite infrared channel radiometric calibration method according to claim 1, characterized in that, S4 specifically includes: S401: Calculate the true radiance corresponding to each blackbody DN value by combining the blackbody DN value and the temperature corresponding to the blackbody DN value: Among them, T i and R i Let λ represent the i-th temperature of the blackbody and the true radiance corresponding to the i-th temperature of the blackbody, respectively; exp represent the natural exponential function; C1 and C2 represent the first and second radiation constants, respectively; and λ represent the first and second radiation constants, respectively. 波长 This indicates the wavelength emitted by the satellite's infrared channel for observing a blackbody; S402: Substitute the blackbody DN value into the calculation of the radiation transfer equations at different orders to obtain the fitted radiance at different orders; S403: Using the ridge regression algorithm, establish a loss function for the true radiance and the fitted radiance: Among them, L oss Let O represent the loss function, where O represents the total number of blackbody temperature changes. This represents the fitted radiance obtained by fitting the blackbody DN value collected for the temperature of the i-th blackbody, where γ = 0.5 represents the penalty term coefficient. S404: Determine the sum of squared residuals between the true radiance and the fitted radiance, with the goal of minimizing the loss function: min RSS L oss Where RSS represents the sum of squared residuals, min RSS L oss This indicates taking the minimum L. oss The RSS feed below; S405: Calculate the Akaike information criterion value and the Bayesian information criterion value with respect to the sum of squared residuals: Where AIC and BIC represent the Akaike Information Criterion value and the Bayesian Information Criterion value, respectively, and log represents the logarithmic function; S406: Calculate the ratio of the Akaike information criterion value to the Bayesian information criterion value, and output the order of the radiative transfer equation with the minimum ratio as the maximum order: in, Indicates taking the minimum n below.

6. The satellite infrared channel radiometric calibration method according to claim 1, characterized in that, S5 specifically includes: S501: Transform the radiation transfer equation into a matrix equation: R = Xa Where a represents the correction coefficient vector, R represents the true radiance vector, and X represents the vector including D. N Design matrix for each term, This represents the DN value associated with the i-th temperature and order v of the blackbody; S502: Solve the matrix equation using the least squares method to obtain the correction coefficient vector, i.e., the correction coefficients of each order: Wherein, the subscript T indicates transpose. Indicates taking The smallest a, This indicates that it has been launched.

7. The satellite infrared channel radiometric calibration method according to claim 1, characterized in that, The corrected radiation intensity is as follows: R′=R·[1+P(θ)cos 2 (i)] Where R′ represents the corrected radiation intensity, θ represents the scanning angle of the satellite scanner, P(θ) represents the scanning angular polarization factor related to the satellite structure, cos represents the cosine function, and R represents the polarization factor related to D. N The corresponding radiance.

8. The satellite infrared channel radiometric calibration method according to claim 1, characterized in that, Following S7, it also includes: After a preset time interval, repeat steps S1 to S7.

9. The satellite infrared channel radiometric calibration method according to claim 1, characterized in that, Also includes: The target scene is scanned by the satellite scanner to obtain the target DN value; Substitute the target DN value into the radiative transfer equation to determine the radiance of the target scene.

10. A satellite infrared channel radiometric calibration system, characterized in that, Used to perform the satellite infrared channel radiometric calibration method according to any one of claims 1 to 9.

Images