Short-wave infrared hyperspectral data relative radiation processing coefficient generation method

By screening and correcting the shortwave infrared hyperspectral data in multi-dimensional statistical samples, the final radiation treatment coefficient is generated, which solves the problem of unstable data quality and difficult to guarantee linear changes, and achieves efficient and stable data correction effect.

CN119023072APending Publication Date: 2024-11-26CHINA CENT FOR RESOURCES SATELLITE DATA & APPL
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410998654.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-24
Publication Date
2024-11-26

AI Technical Summary

Technical Problem

The prior art is difficult to effectively deal with radiation distortion and interference fringes in short-wave infrared hyperspectral imaging data, resulting in unstable data quality, and the statistical histogram matching method is difficult to ensure linear changes, affecting subsequent calibration work.

Method used

By analyzing and pre-processing the original 0-segment scene data, multi-dimensional statistical sample data of imaging time, gain and integral conditions are screened, Etalon effect coefficient and non-uniformity coefficient are calculated, sensor chip array data is corrected, and the final The radiation treatment coefficient includes the product of the Etalon effect coefficient, the nonuniformity coefficient and the residual nonuniformity coefficient between the slices.

Benefits of technology

Effective correction of short-wave infrared hyperspectral data is achieved, the problems of low signal-to-noise ratio and unstable data quality are overcome, linear changes in the correction results and efficient and stable algorithms are ensured, and residual inhomogeneity errors are completely eliminated.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119023072A_ABST
    Figure CN119023072A_ABST
Patent Text Reader

Abstract

A short-wave infrared hyperspectral data relative radiation processing coefficient generation method comprises the following steps: based on statistical sample data, calculating an Etalon effect coefficient and a non-uniformity coefficient of each sensor sheet array, carrying out self-correction, and carrying out inter-sheet correction on self-corrected data by using a laboratory inter-sheet non-uniformity coefficient; self-adaptively generating an inter-chip non-uniformity coefficient; and comprehensively processing the Etalon effect coefficient and the heterogeneity coefficient of each sensor sheet array, the inter-sheet heterogeneity coefficient of the laboratory and the inter-sheet heterogeneity coefficient generated in a self-adaptive manner to generate a final radiation processing coefficient. The invention provides the effective and stable radiation processing coefficient generation method for the in-orbit short-wave infrared hyperspectral load 0-level data, and the production of the radiation processing coefficient for the in-orbit short-wave infrared hyperspectral load 0-level data can be carried out with high precision.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the field of hyperspectral remote sensing image processing and relates to a method for generating relative radiation processing coefficients of short-wave infrared hyperspectral data. Background Art

[0002] As an important technology in the field of earth observation, imaging spectroscopy can obtain high-resolution spectral information of surface objects while clearly imaging the surface. With the further development of aerospace remote sensing technology, satellite-borne hyperspectral imagers and their data processing technology have become one of the important research hotspots in the field of remote sensing.

[0003] my country's hyperspectral comprehensive observation satellite is equipped with a hyperspectral imager, which can obtain imaging spectral data in the visible and short-wave infrared bands. It is widely used in environmental monitoring, land and resources surveys, crop growth monitoring, water quality monitoring and other fields. After the satellite is in orbit, it is limited by temperature changes, instrument development technology, materials and other levels, resulting in radiation problems such as radiation distortion and interference fringes in short-wave infrared data, which requires the development of data processing methods for correction. The general method is to use a large amount of data statistical histograms, and then perform histogram matching methods to achieve it. However, for short-wave infrared hyperspectral data, on the one hand, the signal-to-noise ratio of short-wave infrared data is lower than that of the visible spectrum, and the data quality stability is weak, making it difficult to use a unified standard histogram to calibrate each imaging CCD unit; on the other hand, from the perspective of quantitative application, the pixel grayscale value output by the statistical histogram matching method is difficult to completely guarantee linear changes compared to the original histogram, which brings difficulties to the subsequent calibration work. Summary of the invention

[0004] The technical problem solved by the present invention is to overcome the deficiencies of the prior art, provide a method for generating relative radiation processing coefficients of short-wave infrared hyperspectral data, and solve the problem of radiation defects in short-wave infrared hyperspectral imaging data of on-orbit satellites.

[0005] The technical solution of the present invention is: a method for generating relative radiation processing coefficients of short-wave infrared hyperspectral data, comprising:

[0006] Step 1, parsing and preprocessing the acquired original level 0 scene segmentation data to obtain preprocessed level 0 scene segmentation data and auxiliary data;

[0007] Step 2: According to the imaging time, gain, and integral combination conditions recorded in the auxiliary data, the grayscale values ​​of the pre-processed 0-level image scene data are divided into several combinations; in each combination, data covering different objects and different geographical locations without exposure and cloud cover are selected as statistical sample data of the combination; finally, a multidimensional statistical sample set including three dimensions of imaging time, gain, and integral is formed;

[0008] Step 3, for each statistical sample data set obtained, the Etalon effect coefficient and the non-uniformity coefficient of each sensor array are calculated, and each sensor array data is corrected based on the calculated Etalon effect coefficient and the non-uniformity coefficient;

[0009] Step 4, using the inter-chip non-uniformity coefficient of the sensor chips measured in the laboratory, perform preliminary correction of the inter-chip non-uniformity error on each sensor chip array data after correction;

[0010] Step 5, based on the data of each sensor array after the initial correction, generate the residual non-uniformity coefficient between the slices, and correct the residual non-uniformity error;

[0011] Step 6: Multiply the Etalon effect coefficient, the non-uniformity coefficient, the sensor chip non-uniformity coefficient measured in the laboratory, and the chip residual non-uniformity coefficient of each sensor chip array to generate the final radiation processing coefficient.

[0012] The step of parsing and preprocessing the acquired original level 0 scene segmentation data to obtain the preprocessed level 0 scene segmentation data and auxiliary data includes:

[0013] According to the satellite-ground interface file, the level 0 scene data is parsed to obtain auxiliary data; the travel time information in the auxiliary data is extracted;

[0014] According to the dark current time-varying parameter file, the dark current time-varying parameter of each detector of the short-wave infrared hyperspectral array is obtained;

[0015] Based on the extracted row-time information and dark current time-varying parameters, the grayscale value of each detector in each frame of 0-level scene data after deducting the dark current is calculated;

[0016] The blind pixel file is read, and the grayscale value at the blind pixel position after deducting the dark current is repaired, and finally the pre-processed 0-level scene data is obtained.

[0017] The gray value DN after deducting the dark current of each detector in each frame of 0-level scene data is obtained by calculation=DN 原始 -Time*ab, where Time is the imaging time corresponding to each frame of image, DN 原始 is the detector grayscale value before deducting the dark current, DN is the detector grayscale value after deducting the dark current, a and b represent the two time-varying parameters of slope a and cutoff b contained in the dark current time-varying parameter file.

[0018] When DN<0, set DN to 0.

[0019] The step of calculating the Etalon effect coefficient and the non-uniformity coefficient of each sensor array and correcting the data of each sensor array based on the calculated Etalon effect coefficient and the non-uniformity coefficient includes:

[0020] Assume that the short-wave infrared hyperspectral camera consists of K sensor arrays;

[0021] For each film array, the Etalon effect coefficient of each scene data in the statistical sample data set is calculated to obtain the Etalon effect coefficient of N scenes data, and M Etalon effect coefficients that meet the conditions are screened out; the M Etalon coefficients are averaged to obtain the comprehensive Etalon effect coefficient as the final Etalon effect coefficient of the film array;

[0022] The comprehensive Etalon effect coefficients of the obtained K slice arrays are multiplied by the grayscale values ​​of the K slice array data corresponding to each scene, so as to realize the adaptive correction of each scene data;

[0023] Based on the corrected data, the non-uniformity coefficient of each array detector is calculated; the obtained non-uniformity coefficient is multiplied by the grayscale value of each scene data to achieve adaptive correction of each scene data.

[0024] The M Etalon effect coefficients meeting the conditions are screened out, and the screening conditions are:

[0025] Etalon i <3*STDE,0 <i≤M

[0026] 0 <Etalon i <1.5,0 <i≤M

[0027] In the formula, Etalon i is the Etalon effect coefficient of the i-th scene data that meets the conditions, and STDE is the standard deviation of N Etalon coefficients.

[0028] The calculation to obtain the non-uniformity coefficient of each array detector includes:

[0029] The mean value MEAN of the array detector along the track is calculated, and the fitting value MEAN' of the mean value of the array detector along the track in the spatial dimension is calculated by the polynomial fitting method, and then the non-uniformity coefficient Cof = MEAN' / MEAN of each array detector is calculated.

[0030] The correcting of the residual non-uniformity error comprises:

[0031] The following process is implemented for each scene data:

[0032] (1) Downsample the spatial dimensions of the detector array twice to obtain a sample data set with three spatial resolutions;

[0033] (2) Calculate the average grayscale value of each pixel along the track corresponding to the three spatial resolutions;

[0034] (3) For the obtained average grayscale values ​​along the track, in the row direction, the average values ​​MODEL1_MEAN, MODEL2_MEAN, ... MODELK_MEAN of each sensor array and the overall average value MODEL_MEAN of the K arrays are calculated respectively;

[0035] (4) For each scene data, calculate the non-uniformity coefficient of the corresponding resolution of each array;

[0036] Res i _Cof_MODEL1=MODEL_MEAN / MODEL1_MEAN

[0037] Res i _Cof_MODEL2=MODEL_MEAN / MODEL2_MEAN

[0038] ·

[0039] ·

[0040] ·

[0041] Res i _Cof_MODELK=MODEL_MEAN / MODELK_MEAN

[0042] Where i=1,2,3 represents the i-th spatial dimension resolution;

[0043] (5) Multiplying the coefficients of the three resolutions of each array, we can get the non-uniformity coefficient of the array:

[0044] Cof_MODEL1=Res1_Cof_MODEL1*Res2_Cof_MODEL1*Res3_Cof_MODEL1

[0045] Cof_MODEL2=Res1_Cof_MODEL2*Res2_Cof_MODEL2*Res3_Cof_MODEL2

[0046] ·

[0047] ·

[0048] ·

[0049] Cof_MODELK=Res1_Cof_MODELK*Res2_Cof_MODELK*Res3_Cof_MODELK

[0050] Repeat steps (1) to (5) for each scene data to calculate the comprehensive non-uniformity coefficients of the K slice arrays of each scene data, and then average the comprehensive non-uniformity coefficients of the corresponding slice arrays of each scene to obtain the error-corrected non-uniformity coefficients of the K slice arrays.

[0051] The advantages of the present invention compared with the prior art are:

[0052] 1. According to the physical process of sensor imaging, the radiation distortion generated in each link is corrected, especially the modulation error caused by the inherent Etalon effect of short-wave infrared is separated and removed, which has clear physical significance.

[0053] 2. The process of correcting the non-uniformity error between sensor chips first uses the laboratory non-uniformity coefficient to process each sensor array separately and then uniformly calculates the non-uniformity between each sensor chip, ensuring the efficiency, stability and reliability of the algorithm.

[0054] 3. The residual non-uniformity error correction process between sensor slices takes into account the low signal-to-noise ratio of short-wave infrared hyperspectral camera data. First, the image is downsampled twice in the spatial dimension, and then the non-uniformity error correction coefficient is calculated for each image of spatial resolution. Finally, the non-uniformity coefficients of each detector in the three dimensions are averaged to obtain the final residual non-uniformity error correction coefficient. This method takes into account the errors of different distribution frequencies in the spatial dimension, so that the final residual non-uniformity error can be eliminated more thoroughly.

[0055] 4. The on-orbit statistical samples are screened according to the sensor imaging time, gain, and integral combination conditions, which overcomes the disadvantage of the unstable relative radiation processing coefficient caused by the single condition screening of statistical samples in the existing technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 Flowchart of technique for emissivity generation.

[0057] Figure 2 Screening flowchart for the sample database.

[0058] Figure 3 This is a schematic diagram of a satellite's shortwave infrared sensor array. DETAILED DESCRIPTION

[0059] The specific implementation manner is described by taking a certain type of short-wave infrared hyperspectral sensor as an example.

[0060] The detector array of a certain type of short-wave infrared hyperspectral sensor consists of 4 arrays, each of which consists of 512 detectors in the spatial dimension and 180 detectors in the spectral dimension. The image size of one frame (push-scan line) of the sensor imaging the ground is 2048 columns * 180 rows. Figure 3 shown.

[0061] The specific implementation of the present invention is described in detail below with reference to the accompanying drawings and embodiments. Figure 1 , Figure 2 and Figure 3 The process of the embodiment can be divided into six steps, and the specific method, formula and process for implementing each step are as follows:

[0062] 1. Analyze and preprocess the original level 0 scene segmentation data to obtain preprocessed level 0 scene segmentation data and auxiliary data.

[0063] (1) First, the level 0 scene data is parsed to obtain image data and auxiliary data; then the line time information in the auxiliary data is further extracted.

[0064] (2) Read the dark current time-varying parameter file, which contains the slope and truncation parameters. The dark current of a frame of data in each orbit of the satellite is obtained through the above two parameters, that is, the absolute line time of the frame data multiplied by the slope, plus the truncation, and the result is the dark current of the frame data. The absolute line time of the frame data refers to the time from the camera startup time to the imaging experience of the frame data; taking the linear time-varying parameter file as an example, the file contains two time-varying parameters, slope a and truncation b. Read a and b, and based on the line time information and time-varying parameters of each frame data, calculate the gray value of each detector in each frame data after deducting the dark current according to the following formula:

[0065] DN=DN 原始 -Time*ab

[0066] When DN<0, set DN to 0

[0067] Where Time is the imaging time corresponding to each frame of image, DN 原始 is the detector grayscale value before deducting the dark current, and DN is the detector grayscale value after deducting the dark current.

[0068] (3) Read the blind pixel file, repair the grayscale value at the blind pixel position after deducting the dark current, and finally obtain the pre-processed 0-level scene data.

[0069] 2. If Figure 2As shown, according to the imaging time, gain, and integral combination conditions recorded in the auxiliary data, the grayscale values ​​of the preprocessed level 0 image scene data are divided into several combinations; in each combination, data covering different ground objects and different geographical locations without exposure and cloud cover are screened as statistical sample data of the combination; finally, a statistical sample set including three spatial dimensions of imaging time, gain, and integral is formed.

[0070] 3. For each sample data set obtained, adaptive correction is performed on each sensor array data;

[0071] (1) The short-wave infrared hyperspectral sensor consists of four CCD imaging arrays. For each array, the Etalon coefficient of each scene data is first calculated to obtain the Etalon coefficient of N scenes data. According to the following formula, M Etalon coefficients that meet the conditions are screened out, and then the M Etalon coefficients are averaged to obtain the comprehensive Etalon coefficient as the final Etalon coefficient of the array.

[0072] Etalon ii <3*STDE,0 <ii≤M

[0073] 0 <Etalon ii <1.5,0 <ii≤M

[0074] In the formula, Etalon ii is the Etalon coefficient that meets the conditions, and STDE is the standard deviation of N Etalon coefficients.

[0075] The comprehensive Etalon coefficients of the four slice arrays are multiplied by the grayscale values ​​of the four slice array data corresponding to each scene to achieve adaptive correction of the data of each scene.

[0076] (2) Based on the corrected data, we first calculate the mean value of the array detector along the track: 2048*180 Then, the polynomial fitting method is used to calculate the spatial dimension fitting value MEAN' of the average value of the array detector along the track 2048*180 , and then use the following formula to get the non-uniformity coefficient Cof of each array detector 2048*180 Finally, the non-uniformity coefficient is used to perform adaptive correction on each scene data.

[0077] Cof 2048*180 = MEAN' 2048*180 / MEAN 2048*180

[0078] 4. Use the inter-array non-uniformity coefficient measured in the laboratory to perform inter-module non-uniformity correction on the corrected data.

[0079] 5. Correct the residual non-uniformity error of the corrected data.

[0080] First, for each scene data, implement the following process:

[0081] (1) The spatial dimension of the detector array is downsampled twice to obtain a total of three spatial dimension resolution sample data sets, and the corresponding number of pixels per frame data is 2048*180, 1024*180, and 512*180 respectively;

[0082] (2) Calculate the average grayscale value of each pixel along the track corresponding to each spatial dimension resolution.

[0083] (3) For the average grayscale value along the track obtained in (2), calculate the average value of each slice array in the row direction MODEL1_MEAN 1*180 , MODEL2_MEAN 1*180 , MODEL3_MEAN 1*180 , MODEL4_MEAN 1*180 and the overall average value of the four arrays MODEL_MEAN 1*180 .

[0084] (4) For each scene data, calculate the non-uniformity coefficient of the corresponding resolution of each array.

[0085] Res i _Cof_MODEL1 1*180 =MODEL_MEAN 1*180 / MODEL1_MEAN 1*180

[0086] Res i _Cof_MODEL2 1*180 =MODEL_MEAN 1*180 / MODEL2_MEAN 1*180

[0087] Res i _Cof_MODEL3 1*180 =MODEL_MEAN 1*180 / MODEL3_MEAN 1*180

[0088] Res i _Cof_MODEL4 1*180 =MODEL_MEAN 1*180 / MODEL4_MEAN 1*180

[0089] Where i = 1, 2, 3

[0090] (5) Multiplying the three resolution coefficients of each module can obtain the comprehensive non-uniformity coefficient of the module.

[0091] Cof_MODEL1 1*180 =Res1_Cof_MODEL1 1*180 *Res2_Cof_MODEL1 1*180 *

[0092] Res3_Cof_MODEL1 1*180

[0093] Cof_MODEL2 1*180 =Res1_Cof_MODEL2 1*180 *Res2_Cof_MODEL2 1*180 *

[0094] Res3_Cof_MODEL2 1*180

[0095] Cof_MODEL3 1*180 =Res1_Cof_MODEL3 1*180 *Res2_Cof_MODEL3 1*180 *

[0096] Res3_Cof_MODEL3 1*180

[0097] Cof_MODEL4 1*180 =Res1_Cof_MODEL4 1*180 *Res2_Cof_MODEL4 1*180 *

[0098] Res3_Cof_MODEL4 1*180

[0099] Then, the processes (1) to (5) are repeated for each scene data to calculate the comprehensive non-uniformity coefficient of the four slice arrays of each scene data, and the non-uniformity coefficients of each scene are averaged to obtain the final non-uniformity coefficient.

[0100] 6. Multiply the Etalon effect coefficient, non-uniformity coefficient, laboratory-measured sensor chip non-uniformity coefficient, and chip residual non-uniformity coefficient of each sensor chip array to generate the final radiation processing coefficient.

[0101] Although the present invention has been disclosed as above in the form of a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art can make possible changes and modifications to the technical solution of the present invention by using the technical contents disclosed above without departing from the spirit and scope of the present invention. Therefore, any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall fall within the protection scope of the technical solution of the present invention.

Claims

1. A method for generating relative radiation processing coefficients of short-wave infrared hyperspectral data, characterized in that: include: Parsing and preprocessing the acquired original level 0 scene segmentation data to obtain preprocessed level 0 scene segmentation data and auxiliary data; According to the imaging time, gain, and integral combination conditions recorded in the auxiliary data, the grayscale values ​​of the pre-processed 0-level image scene data are divided into several combinations; in each combination, data covering different objects and different geographical locations without exposure and cloud cover are selected as the statistical sample data of the combination; finally, a statistical sample set including three spatial dimensions of imaging time, gain, and integral is formed; For each statistical sample data in the obtained statistical sample set, the Etalon effect coefficient and the non-uniformity coefficient of each sensor array are calculated, and each sensor array data is corrected based on the calculated Etalon effect coefficient and the non-uniformity coefficient; Using the inter-chip non-uniformity coefficient of the sensor measured in the laboratory, the inter-chip non-uniformity error of each sensor array data after correction is initially corrected; Based on the data of each sensor array after the initial correction, the residual non-uniformity coefficient between the slices is generated, and the residual non-uniformity error is corrected; The final radiation processing coefficient is generated by multiplying the Etalon effect coefficient, non-uniformity coefficient, laboratory-measured sensor chip non-uniformity coefficient, and chip residual non-uniformity coefficient of each sensor array.

2. The method for generating relative radiation processing coefficients of short-wave infrared hyperspectral data according to claim 1, characterized in that: The step of parsing and preprocessing the acquired original level 0 scene segmentation data to obtain the preprocessed level 0 scene segmentation data and auxiliary data includes: According to the satellite-ground interface file, the level 0 scene data is parsed to obtain auxiliary data; the travel time information in the auxiliary data is extracted; According to the dark current time-varying parameter file, the dark current time-varying parameter of each detector of the short-wave infrared hyperspectral array is obtained; Based on the extracted row-time information and dark current time-varying parameters, the grayscale value of each detector in each frame of 0-level scene data after deducting the dark current is calculated; The blind pixel file is read, and the grayscale value at the blind pixel position after deducting the dark current is repaired, and finally the pre-processed 0-level scene data is obtained.

3. The method for generating relative radiation processing coefficients of short-wave infrared hyperspectral data according to claim 2, characterized in that: The gray value DN after deducting the dark current of each detector in each frame of 0-level scene data is obtained by calculation=DN 原始 -Time*ab, where Time is the imaging time corresponding to each frame of image, DN 原始 is the detector grayscale value before deducting the dark current, DN is the detector grayscale value after deducting the dark current, a and b represent the two time-varying parameters of slope a and cutoff b contained in the dark current time-varying parameter file.

4. The method for generating relative radiation processing coefficients of short-wave infrared hyperspectral data according to claim 3, characterized in that: When DN<0, set DN to 0.

5. The method for generating relative radiation processing coefficients of short-wave infrared hyperspectral data according to claim 1, characterized in that: The step of calculating the Etalon effect coefficient and the non-uniformity coefficient of each sensor array and correcting the data of each sensor array based on the calculated Etalon effect coefficient and the non-uniformity coefficient includes: Assume that the short-wave infrared hyperspectral camera consists of K sensor arrays; For each film array, the Etalon effect coefficient of each scene data in the statistical sample data set is calculated to obtain the Etalon effect coefficient of N scenes data, and M Etalon effect coefficients that meet the conditions are screened out; the M Etalon coefficients are averaged to obtain the comprehensive Etalon effect coefficient as the final Etalon effect coefficient of the film array; The comprehensive Etalon effect coefficients of the obtained K slice arrays are multiplied by the grayscale values ​​of the K slice array data corresponding to each scene, so as to realize the adaptive correction of each scene data; Based on the corrected data, the non-uniformity coefficient of each array detector is calculated; the obtained non-uniformity coefficient is multiplied by the grayscale value of each scene data to achieve adaptive correction of each scene data.

6. The method for generating relative radiation processing coefficients of short-wave infrared hyperspectral data according to claim 5, characterized in that: The M Etalon effect coefficients meeting the conditions are screened out, and the screening conditions are: Etalon ii <3*STDE,0 <ii≤M 0<Etalon ii <1.5,0<ii≤M In the formula, Etalon ii is the Etalon effect coefficient of the i-th scene data that meets the conditions, and STDE is the standard deviation of N Etalon coefficients.

7. The method for generating relative radiation processing coefficients of short-wave infrared hyperspectral data according to claim 5, characterized in that: The calculation to obtain the non-uniformity coefficient of each array detector includes: Calculate the mean value MEAN of the array detector along the track, and use the polynomial fitting method to calculate the fitting value MEAN of the mean value of the array detector along the track in the spatial dimension , , and then calculate the non-uniformity coefficient Cof of each array detector = MEAN , / MEAN.

8. The method for generating relative radiation processing coefficients of short-wave infrared hyperspectral data according to claim 7, characterized in that: The correcting of the residual non-uniformity error comprises: The following process is implemented for each scene data: (1) Downsample the spatial dimensions of the detector array twice to obtain a sample data set with three spatial resolutions; (2) Calculate the average grayscale value of each pixel along the track corresponding to the three spatial resolutions; (3) For the obtained average grayscale values ​​along the track, in the row direction, the average values ​​MODEL1_MEAN, MODEL2_MEAN, ... MODELK_MEAN of each sensor array and the overall average value MODEL_MEAN of the K arrays are calculated respectively; (4) For each scene data, calculate the non-uniformity coefficient of the corresponding resolution of each array; Where i=1,2,3 represents the i-th spatial dimension resolution; (5) Multiplying the coefficients of the three resolutions of each array, we can get the non-uniformity coefficient of the array: Repeat steps (1) to (5) for each scene data to calculate the comprehensive non-uniformity coefficients of the K slice arrays of each scene data, and then average the comprehensive non-uniformity coefficients of the corresponding slice arrays of each scene to obtain the error-corrected non-uniformity coefficients of the K slice arrays.