A de-rotation type satellite-borne hyperspectral camera yaw calibration data processing method

Abstract

The present application relates to yaw calibration data simulation analysis, in order to solve the existing flyback type hyperspectral camera yaw calibration cannot effectively and reliably process the collected yaw imaging data, and a flyback type spaceborne hyperspectral camera yaw calibration data processing method is provided, comprising: S1, extracting the yaw calibration effective area data of the original hyperspectral camera yaw calibration data; S2, calculating the local range threshold; S3, calculating the local gray mean value of the yaw calibration effective area data; S4, fitting the local gray mean value space dimension of the yaw calibration effective area data; S5, calculating the local relative correction coefficient of the yaw calibration effective area data; S6, calculating the different gray level relative correction coefficient of the yaw calibration effective area data; S7, correcting the original hyperspectral camera yaw calibration data through the different gray level relative correction coefficient, thereby completing the flyback type spaceborne hyperspectral camera yaw calibration data processing.

Description

TECHNICAL FIELD

[0001] The application relates to simulation analysis of yaw calibration data of a hyperspectral camera, in particular to a yaw calibration data processing method for a backscanning satellite-borne hyperspectral camera. BACKGROUND

[0002] The principle of yaw calibration is to control satellite yaw rotation, so that different pixels of a detector make multiple observations on the same uniform target (such as a cloud layer, a desert or an ocean, etc.) in a short time, the response data of these pixels to the same radiation input are processed, the relative gain and offset differences between the pixels are calculated, and then the strip noise or response non-uniformity caused by inconsistent sensitivity in the detector array is eliminated, so as to realize the consistency correction of the radiation response between the pixels; it is a kind of on-orbit relative radiation correction method, which does not need a ground calibration field and only relies on satellite attitude maneuvering and the statistical characteristics of a uniform scene to complete the correction.

[0003] According to the satellite orbit attitude principle, the yaw calibration in the passive push-broom mode can ensure that the projection imaging area of the camera detector on the ground runs along a straight line after the yaw direction of the remote sensing camera is rotated by 90 degrees. The passive push-broom mode of the existing panchromatic and multispectral satellite-borne remote sensing camera is a conventional imaging mode, which meets the demand of the conventional yaw calibration passive push-broom mode; the yaw calibration data obtained by the passive push-broom mode is standard yaw calibration data, that is, the current pixel of the yaw calibration data and the 45° adjacent pixel above it are imaging data of the same ground object; the existing yaw data processing method only needs to extract a row of data in the 45° diagonal direction, that is, the row of data is considered as the imaging result of all the pixels of the detector on the same ground object in the spatial dimension.

[0004] In order to obtain enough exposure time of ground object spectral data, the frame frequency of the hyperspectral remote sensing camera is relatively low compared with the panchromatic and multispectral cameras, and the active backscanning compensation mode of the satellite platform is usually used to meet the low frame frequency of the hyperspectral camera for ground imaging. Figure 2 As shown in the figure, according to the satellite orbit imaging principle, the conventional active backscanning mode yaw calibration cannot guarantee that the projection imaging area of the camera detector on the ground runs strictly along a straight line, that is, there is an image shift deviation in the yaw imaging process; and it also cannot meet the requirements of the existing yaw calibration method for processing and precision evaluation of the collected yaw imaging data.

[0005] Therefore, it is necessary to seek an effective processing and precision evaluation of the yaw data for the new yaw calibration mode of the backscanning satellite-borne hyperspectral camera, which is of great significance for the yaw calibration mode demonstration in the development stage of the hyperspectral camera and the yaw calibration data processing in the on-orbit stage. SUMMARY

[0006] The application aims to solve the technical problems that the existing yaw calibration of a back-scan type spaceborne hyperspectral remote sensing camera cannot effectively and reliably process the collected yaw imaging data, and the yaw imaging process has image shift deviation.

[0007] To achieve the above-mentioned purpose, the technical solution provided by the application is as follows.

[0008] A yaw calibration data processing method for a back-scan type spaceborne hyperspectral camera, which is characterized by comprising the following steps.

[0009] S1, extracting yaw calibration effective area data of original hyperspectral camera yaw calibration data;

[0010] S2, calculating a local range threshold according to a hyperspectral camera visual axis center trajectory in a hyperspectral camera yaw imaging process;

[0011] S3, calculating a local gray mean value of the yaw calibration effective area data according to the local range threshold;

[0012] S4, performing local gray mean value spatial dimension fitting on the yaw calibration effective area data according to the local gray mean value;

[0013] S5, calculating a local relative correction coefficient of the yaw calibration effective area data according to the local gray mean value spatial dimension fitting;

[0014] S6, calculating different gray level relative correction coefficients of the yaw calibration effective area data according to the local relative correction coefficient;

[0015] S7, performing relative radiation correction on the original hyperspectral remote sensing data frame by frame through the different gray level relative correction coefficients, so as to complete the yaw calibration data processing of the back-scan type spaceborne hyperspectral camera.

[0016] Further, step S1 is specifically as follows.

[0017] The yaw calibration effective area data Data of the original hyperspectral camera yaw calibration data Data_Org is extracted according to the following formula:

[0018] Data iFrame,iCol,iBand =Data_Org iFrame+iCol-1,iCol,iBand

[0019] iFrame∈1~Frames

[0020] iCol∈1~Cols

[0021] iBand∈1~Bands

[0022] Wherein: Frames is the total number of frames of the yaw calibration effective area data, Cols is the total number of columns of the yaw calibration data space dimension, Bands is the total number of spectral bands of the yaw calibration hyperspectrum, iFrame is the current frame number to be traversed, iCol is the current column number to be traversed, and iBand is the current spectral band number to be traversed.

[0023] Further, step S2 is specifically:

[0024] S2.1, according to the longitude coordinate sequence Lon and the latitude coordinate sequence Lat of the center trajectory of the line of sight of the hyperspectral camera in the process of the yaw imaging of the hyperspectral camera, a straight line direction is fitted by using the least square method, a slope of the fitted straight line is obtained, and the calculation formula is as follows:

[0025] Lat=k×Lon+b

[0026]

[0027] Wherein: k is the slope of the fitted straight line, b is the intercept of the fitted straight line, np is the number of the longitude and latitude coordinate sequence points, and i is the current longitude and latitude coordinate sequence point number to be traversed and calculated;

[0028] S2.2, according to the slope k of the fitted straight line, the inter-frame longitude and latitude deviations of the trajectory coordinate sequence in the direction of the fitted straight line are solved;

[0029] It is assumed that Pi is a point in the trajectory longitude coordinate sequence, the longitude and latitude deviations d_Loni and d_Lati of the adjacent point Pi+1 and Pi in the direction of the fitted straight line are calculated by the following formulae;

[0030] d_Lon i =((P_Lon i+1 -P_Lon i )u_Lon+(P_Lat i -P_Lat i+1 )u_Lat)u_Lon

[0031] d_Lat i =((P_Lon i+1 -P_Lon i )u_Lon+(P_Lat i -P_Lat i+1 )u_Lat)u_Lat

[0032]

[0033] Wherein: u_Lon and u_Lat are the longitude component and the latitude component of the normal direction of the fitted straight line, P_Lon i is the longitude coordinate of the P point of the i-th frame, and P_Lat id_Lon is the longitude coordinate of the i-th frame d point, and d_Lat is the latitude coordinate of the i-th frame d point;

[0034] S2.3, according to the average longitude and latitude deviation between frames, the average pixel error e is calculated by the following formula:

[0035]

[0036]

[0037] Wherein: Di is the inter-frame physical distance deviation, R is the earth radius, and GSD is the ground resolution of the hyperspectral camera;

[0038] S2.4, according to the average pixel error e, the local range pixel number threshold T is calculated by the following formula:

[0039]

[0040] Wherein: E0 is the set maximum local range yaw displacement pixel error.

[0041] Further, step S3 is specifically:

[0042] The local gray mean value M of the yaw calibration effective area data is calculated by the following formula: iT The total column number Cols of the yaw calibration data space is divided into nt segments with T as the local range.

[0043]

[0044] Wherein: nt is the maximum local range number contained in a row of data, iT is the current segment in nt segments, j1 is the pixel position index value of the current segment, (iT-1)·T+1 to iT·T is the pixel position index value range of the current segment.

[0045] Further, step S4 is specifically:

[0046] S4.1, according to the local gray mean value M iT , the piecewise cubic spline fitting method is adopted, and the output fitting coefficient pp is obtained by the following formula:

[0047] pp=spline(X M ,M)

[0048]

[0049] M=[M1,M2,…,M iT ]

[0050] iT∈1~nt

[0051] Wherein: spline is a method for solving cubic spline fitting coefficients, and the input is the local gray mean value M iT The number series and the corresponding local range midpoint coordinate number series X M , pp is the output fitting coefficient;

[0052] S4.2, using piecewise cubic spline fitting, the relative correction trend item MTrend of each pixel position of the current line data is obtained by the following formula:

[0053] MTrend = ppval (pp, X Cols )

[0054] X Cols = [1, 2, …, Cols]

[0055] MTrend = [MTrend1, MTrend2, …, MTrend Cols ]

[0056] Wherein: ppval is a method for cubic spline fitting using fitting coefficients, and the input is the yaw calibration effective area data Data data column direction coordinate X Cols and the fitting coefficient pp, and the output is the trend item MTrend, so as to complete the local gray mean value space dimension fitting.

[0057] Further, step S5 is specifically:

[0058] S5.1, remove the relative correction trend item MTrend from the yaw calibration effective area data Data according to the following formula, to obtain the data DataT after removing the trend item:

[0059] DataT iFrame,iCols,iBand = Data iFrame,iCols,iBand -MTrend iCols

[0060] S5.2, calculate the relative correction coefficient Coeff according to the following formula:

[0061]

[0062] Further, step S6 is specifically:

[0063] Each frame of data of the yaw calibration effective area data Data can calculate a relative correction coefficient Coeff, and the relative correction coefficients Coeffs of different gray scales are obtained by counting all frames of data:

[0064] The gray value range of hyperspectral camera imaging data is divided into L gray scales, the gray scale where the mean value of the trend item MTrend of each frame of data is located is counted, and the relative correction coefficient Coeffs is calculated by the following formula:

[0065]

[0066] Wherein: iL is the iLth gray scale, FrameiL is the frame set corresponding to the iLth gray scale, n iL is the total number of frames of FrameiL, and j is the current frame number in FrameiL that needs to be traversed.

[0067] Advantages of the present application:

[0068] [1] The yaw calibration data processing method of the present application can effectively process the data with image shift deviation in the yaw imaging process of the hyperspectral camera. Since the image shift deviation caused by the satellite yaw imaging is a slowly changing deviation, the local area of the camera detector in the yaw imaging process is approximated to image the same ground element, and the local range relative correction coefficient is generated, so that the data with image shift deviation in the yaw calibration mode of the back-scan type can be effectively and reliably processed.

[0069] [2] The present application can effectively correct the inconsistent problem of the nonlinear relative response of the hyperspectral camera detector at different gray scales by using the statistical method to obtain the relative correction coefficient of different gray scales.

[0070] [3] By comparing the present yaw calibration method calculated by the present yaw calibration method, the correction improvement effect of the present application can be clearly obtained, and the setting of the local range threshold parameter can be adjusted according to the comparison result, so as to improve the correction accuracy and processing efficiency of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0071] Figure 1 is the flowchart of the embodiment of the yaw calibration data processing method of the present application for a back-scan type spaceborne hyperspectral camera;

[0072] Figure 2 is a comparison diagram of the conventional yaw calibration imaging and the yaw calibration imaging with image shift deviation in the background technology;

[0073] (a) is a first frame comparison diagram;

[0074] (b) is a second frame comparison diagram;

[0075] (c) is a third frame comparison diagram;

[0076] Figure 3 is a schematic diagram of the original hyperspectral camera yaw calibration data in the embodiment of the present application;

[0077] Figure 4 is a schematic diagram of the yaw calibration effective area data in the embodiment of the present application;

[0078] Figure 5 This is a schematic diagram of the direction of the straight line for fitting the imaging trajectory in an embodiment of the present invention;

[0079] Figure 6 This is a schematic diagram of the spatial dimension fitting of the local gray-level mean in an embodiment of the present invention;

[0080] (a) Single-spectral data for the effective area of ​​yaw calibration;

[0081] (b) is a schematic diagram of the spatial dimension fitting of the local gray-level mean;

[0082] Figure 7 This is a schematic diagram of the relative correction coefficients for different gray levels in an embodiment of the present invention;

[0083] Figure 8 This is a schematic diagram showing the original hyperspectral remote sensing data HSI_Org and the ROI (Region Area of ​​Land) in an embodiment of the present invention.

[0084] Figure 9 This is a schematic diagram of single-band data of a uniform region ROI in an embodiment of the present invention;

[0085] Figure 10 This is a schematic diagram comparing data after correction using relative correction coefficients for different gray levels in an embodiment of the present invention;

[0086] (a) is a schematic diagram of the original hyperspectral remote sensing data;

[0087] (b) A schematic diagram of the data after correction by relative correction coefficients for different gray levels;

[0088] Figure 11 This is a schematic diagram comparing the curves of relative correction errors Err and Err_Ref after correction using relative correction coefficients at different gray levels and correction using existing yaw calibration relative correction coefficients in an embodiment of the present invention. Detailed Implementation

[0089] like Figure 1 As shown, a method for processing yaw calibration data from a backscan type spaceborne hyperspectral camera includes the following steps:

[0090] S1. Yaw calibration data preprocessing;

[0091] Based on the yaw calibration principle, extract the effective yaw calibration region data Data from the original hyperspectral camera yaw calibration data Data_Org;

[0092] Data iFrame,iCol,iBand =Data_Org iFrame+iCol-1,iCol,iBand

[0093] iFrame∈1~Frames

[0094] iCol∈1~Cols

[0095] iBand∈1~Bands

[0096] Wherein: Data_Org and Data are three-dimensional data, and the three dimensions are frame number, spatial dimension column number, and hyperspectral band number.

[0097] Frames is the total frame number of the yaw calibration effective area data, Cols is the total column number of the spatial dimension of the yaw calibration data, Bands is the total band number of the hyperspectrum of the yaw calibration, iFrame is the current frame number to be traversed, iCol is the current column number to be traversed, and iBand is the current band number to be traversed.

[0098] S2, local range threshold calculation;

[0099] Due to the active back-scan imaging principle of the satellite platform, it is impossible to ensure that all pixels in the yaw back-scan imaging process are strictly over the same ground element, that is, there is a deviation of image motion in the yaw imaging process. In order to reduce the influence of the deviation of image motion on the yaw calibration, it is necessary to determine the threshold T of the number of pixels in the local range, which is required to approximately image the same ground element in the yaw process.

[0100] S2.1, according to the longitude coordinate sequence Lon and the latitude coordinate sequence Lat of the center trajectory of the hyperspectral camera in the yaw imaging process of the hyperspectral camera, a straight line direction is fitted by using the least square method, the slope of the fitted straight line is obtained, and the calculation formula is as follows:

[0101] Lat=k×Lon+b

[0102]

[0103]

[0104] Wherein: k is the slope of the fitted straight line, b is the intercept of the fitted straight line, np is the number of longitude and latitude coordinate sequences (that is, the number of frames of yaw calibration data), and i is the current longitude and latitude coordinate sequence point number to be traversed and calculated;

[0105] S2.2, according to the slope k of the fitted straight line, the inter-frame longitude and latitude deviation of the trajectory coordinate sequence in the direction of the fitted straight line is solved.

[0106] It is assumed that Pi is a point in the trajectory longitude coordinate sequence, the longitude and latitude deviations d_Loni and d_Lati of the adjacent point Pi+1 and Pi in the direction of the fitted straight line are calculated by the following formula:

[0107]

[0108] d_Loni = ((P_Lon i+1 - P_Lon i ) u_Lon + (P_Lat i - P_Lat i+1 ) u_Lat) u_Lon

[0109] d_Lat i = ((P_Lon i+1 - P_Lon i ) u_Lon + (P_Lat i - P_Lat i+1 ) u_Lat) u_Lat

[0110]

[0111] wherein: u_Lon and u_Lat are longitude component and latitude component of the normal direction of the fitting straight line, i.e. the projection direction of the longitude and latitude deviation of adjacent points, P_Lon i is the longitude coordinate of the P point of the i-th frame, P_Lat i is the latitude coordinate of the P point of the i-th frame, d_Lon is the longitude coordinate of the d point of the i-th frame, and d_Lat is the latitude coordinate of the d point of the i-th frame;

[0112] S2.3, according to the average longitude and latitude deviation between frames, the average pixel error e is calculated by the following formula:

[0113]

[0114]

[0115] wherein: Di is the physical distance deviation between frames, R is the radius of the earth, and GSD is the ground resolution of the hyperspectral camera;

[0116] S2.4, set the maximum local range yaw image motion error as E0 pixels, i.e. the local range yaw image motion error is considered to be approximately imaged to the same ground element within E0, according to the average pixel error e, the local range pixel number threshold T is calculated by the following formula:

[0117]

[0118] wherein: E0 is the set maximum local range yaw image motion pixel error;

[0119] By calculating the local range pixel number threshold, the influence of image motion deviation on yaw calibration can be effectively reduced;

[0120] S3, local gray mean value calculation;

[0121] The local gray mean value M of the effective area data of yaw calibration is calculated by the following formulaiT The total number of columns Cols in the yaw calibration data space is divided into nt segments with T as the local range;

[0122]

[0123] Wherein: nt is the maximum number of local ranges contained in a row of data, iT is the current segment in nt segments, j1 is the pixel position index value of the current segment, (iT-1)·T+1 to iT·T is the pixel position index value range of the current segment.

[0124] S4, local gray mean value space fitting;

[0125] S4.1, according to the local gray mean value M iT , the output fitting coefficient pp is obtained by the following formula using the piecewise cubic spline fitting method;

[0126] pp=spline(X M ,M)

[0127]

[0128] M=[M1,M2,…,M iT ]

[0129] iT∈1~nt

[0130] Wherein: spline is a method for solving cubic spline fitting coefficients, the input is the local gray mean value M iT , the corresponding local range midpoint coordinate sequence X M , and the output fitting coefficient pp;

[0131] S4.2, using piecewise cubic spline fitting, the relative correction trend item MTrend of each pixel position of the current row of data is obtained by the following formula;

[0132] MTrend=ppval(pp,X Cols )

[0133] X Cols =[1,2,…,Cols]

[0134] MTrend=[MTrend1,MTrend2,…,MTrend Cols ]

[0135] Wherein: ppval is a method for cubic spline fitting using fitting coefficients, the input is the yaw calibration effective area data Data column direction coordinate X Cols and the fitting coefficient pp, and the output is the trend item MTrend, thereby completing the local gray mean value space fitting.

[0136] S5, local relative correction coefficient generation;

[0137] S5.1, remove the relative correction trend item MTrend from the yaw calibration effective area data Data according to the following formula to obtain the data DataT after removing the trend item;

[0138] DataT iFrame,iCols,iBand = Data iFrame,iCols,iBand -MTrend iCols

[0139] S5.2, calculate the relative correction coefficient Coeff according to the following formula;

[0140]

[0141] S6, different gray level relative correction coefficient generation;

[0142] Each frame of data of the yaw calibration effective area data Data can calculate a relative correction coefficient Coeff, and the relative correction coefficients Coeffs of different gray levels are obtained by counting all frame data;

[0143] The gray value range of the hyperspectral camera imaging data is divided into L gray levels, the trend item MTrend average value of each frame of data is counted, and the relative correction coefficient Coeffs is calculated by the following formula;

[0144]

[0145] Wherein: iL is the iLth gray level, FrameiL is the frame set corresponding to the iLth gray level, n iL is the total number of frames of FrameiL, and j is the current frame number in FrameiL that needs to be traversed.

[0146] The yaw calibration relative correction coefficient CoeffRef generated by the existing yaw calibration method is compared with the different gray level relative correction coefficients Coeffs calculated in the embodiment to verify the correction effect of the embodiment.

[0147] S7, calculate the yaw calibration relative correction coefficient CoeffRef according to the existing yaw calibration method by the following formula;

[0148]

[0149] S8, relative correction of hyperspectral remote sensing data;

[0150] The original hyperspectral remote sensing data HSI Org is frame by frame relatively radiometrically corrected by different gray level relative correction coefficients Coeffs and yaw calibration relative correction coefficients CoeffRef, and the data HSI corrected by different gray level relative correction coefficients and the data HSI Ref corrected by yaw calibration relative correction coefficients are obtained according to the following formula:

[0151] HSI iFrame,iCol,iBand = HSI Org iFrame,iCol,iBand ⊙ Coeffs iL,iCols,iBand

[0152] HSI_Ref iFrame,iCol,iBand = HSI Org iFrame,iCol,iBand ⊙ CoeffRef iCols,iBand

[0153] iFrame ∈ 1 ~ HSI Frames

[0154] iCol ∈ 1 ~ Cols

[0155] iBand ∈ 1 ~ Bands

[0156] Wherein: ⊙ is point multiplication of the original hyperspectral remote sensing data frame by frame relative correction coefficient, HSI Frames is the total frame number of the original hyperspectral remote sensing data;

[0157] When the original hyperspectral remote sensing data HSI Org is corrected by different gray level relative correction coefficients Coeffs, the correction coefficient Coeffs corresponding to the gray level range iL in which the current pixel gray value of the original hyperspectral remote sensing data is located needs to be used according to the gray level range iL. iL .

[0158] S9, extraction of uniform area of hyperspectral data;

[0159] The region data ROI and ROI Ref with uniform distribution of ground objects are extracted from the data HSI corrected by different gray level relative correction coefficients and the data HSI Ref corrected by yaw calibration relative correction coefficients as evaluation data.

[0160] S10, calculation and comparison of relative correction accuracy.

[0161] The relative correction error of the uniform area is calculated by the following formula to evaluate the yaw calibration relative correction accuracy, and the smaller the relative correction error is, the higher the yaw calibration accuracy is;

[0162] The relative correction error Err of the single spectrum of the uniform area is calculated by the following formula, which is defined as the single column error ErrCol of the uniform area ROI after relative correction divided by the average value MROI of the uniform area, wherein the column direction is along the track direction;

[0163]

[0164]

[0165] ErrCol jj = |MCol jj - MROI

[0166]

[0167] wherein, ROI is the uniform region data to be evaluated in single spectral band, with size of m rows and n columns, MCol jj is the mean value of the jjth column of ROI, MROI is the mean value of ROI, ErrCol jj is the error of the jjth column, Err is the total relative correction error of ROI, and ii and jj are the traversal index values of rows and columns.

[0168] The relative correction errors Err and Err_Ref of the uniform regions ROI and ROI_Ref respectively in each spectral band after correction of the relative correction coefficients of different gray scales and the relative correction coefficients of yaw calibration are calculated, and the comparison results of the two can quantitatively evaluate that the yaw calibration data processing method of the flyback type spaceborne hyperspectral camera in this embodiment has higher, more reliable and effective calibration accuracy.

[0169] In this embodiment, the relative correction coefficients of different gray scales are generated by processing the yaw calibration data of the flyback type spaceborne hyperspectral camera. As shown in Figure 3 , the frame number of the original hyperspectral camera yaw calibration data Data_Org is 400 frames, the spatial dimension column number Cols of the yaw calibration data is 200, and the hyperspectral spectral band number Bands is 15.

[0170] The hyperspectral camera yaw calibration data is processed according to the above method as follows:

[0171] 1. Preprocessing of yaw calibration data;

[0172] The yaw calibration effective region data Data is extracted from the original hyperspectral camera yaw calibration data Data_Org, and according to the above method, the total frame number Frames of the original hyperspectral camera yaw calibration data Data is 200 frames, the total column number Cols of the spatial dimension of the yaw calibration data is 200, and the total spectral band number Bands of the yaw calibration is 15;

[0173] As shown in Figure 4 , the single spectral band image of the yaw calibration effective region data Data can be clearly seen, and due to the image shift deviation, all pixels in the horizontal spatial dimension direction are not imaged on the same ground element.

[0174] 2. Local threshold calculation;

[0175] To ensure that pixels within a local area of ​​the detector approximate the same ground element during yaw, it is necessary to determine a threshold T for the number of pixels within that local area.

[0176] 2.1, such as Figure 5 As shown, based on the longitude coordinate sequence Lon and latitude coordinate sequence Lat of the hyperspectral camera's line-of-sight center trajectory during the yaw imaging process, the direction of the straight line is fitted using the least squares method, and the slope k of the fitted straight line is 8.791.

[0177] 2.2. Based on the slope k of the fitted line, calculate the inter-frame latitude and longitude deviation of the trajectory coordinate sequence in the direction of the fitted line;

[0178] 2.3. Based on the average latitude and longitude deviation between frames, calculate the average pixel error e, where R (Earth radius) is taken as 6371 km and GSD (Ground resolution) is taken as 2.5 m. The calculated average pixel error e is 0.006 pixels.

[0179] 2.4. Set the maximum yaw image shift error in the local area to E0 as 0.2 pixels. That is, if the yaw image shift error in the local area is within E0 of 0.2 pixels, it is considered to be approximately imaging the same ground element. Then the threshold T for the number of pixels in the local area is 33 pixels.

[0180] 3. Calculation of local grayscale mean;

[0181] For example Figure 6 Taking the 46th row of the first spectral band as an example, the local gray-scale mean sequence of a row of data is [561.06, 498.12, 484.10, 491.57, 529.15, 688.60], where the maximum number of local ranges contained in a row of data is 6.

[0182] 4. Local gray-level mean spatial dimension fitting;

[0183] Based on the local grayscale mean M iT The method of piecewise cubic spline fitting is used to fit the relative correction trend term MTrend of the current row data in Data.

[0184] 5. Generation of local relative correction coefficients;

[0185] Remove the relative correction trend term MTrend from the effective yaw calibration data Data to obtain the trend term-free data DataT. Then, calculate the relative correction coefficient Coeff for each frame of yaw calibration data.

[0186] 6. Generation of relative correction coefficients for different gray levels;

[0187] likeFigure 7 As shown, by statistically analyzing all frame data, the relative correction coefficients Coeffs for different gray levels can be obtained, where the gray value range is set to L, which represents 5 gray levels.

[0188] 7. Based on the original hyperspectral camera yaw calibration data Data_Org, calculate the yaw calibration relative correction coefficient CoeffRef according to existing yaw calibration methods;

[0189] 8. Relative correction of hyperspectral remote sensing data;

[0190] like Figure 10 As shown, relative radiometric correction was performed on the original hyperspectral remote sensing data HSI_Org frame by frame using relative correction coefficients Coeffs and yaw calibration relative correction coefficients CoeffRef at different gray levels. The data HSI after correction with relative correction coefficients at different gray levels and the data HSI_Ref after correction with relative correction coefficients at yaw calibration were calculated.

[0191] 9. Extraction of uniform regions from hyperspectral data;

[0192] like Figure 8 As shown, the corresponding uniform area data ROI and ROI_Ref of the HSI data after correction by relative correction coefficients at different gray levels and the HSI_Ref data after correction by relative correction coefficients for yaw calibration are extracted respectively.

[0193] 10. Calculation and comparison of relative correction accuracy;

[0194] like Figure 9 As shown, the relative correction error Err of a single spectral band in a uniform region calculated by the above method is defined as the ratio of the single column error ErrCol of the uniform region ROI after relative correction to the mean MROI of the uniform region.

[0195] The relative correction errors Err and Err_Ref of the uniform regions ROI and ROI_Ref after correction by relative correction coefficients of different gray levels and yaw calibration relative correction coefficients are calculated for each spectral segment. By comparing the two, the calibration accuracy improvement effect of the patented method can be quantitatively evaluated.

[0196] like Figure 11 As shown, the average relative error of the original data is 9.57%, the average relative correction error of the existing yaw calibration method is 4.39%, and the average relative correction error of the yaw calibration method in this embodiment is 2.33%. The yaw calibration method in this embodiment improves the relative correction performance of this hyperspectral remote sensing data by 47% compared to the existing yaw calibration methods.

Claims

1. A method for processing yaw calibration data from a retrace-type spaceborne hyperspectral camera, characterized in that, Includes the following steps: S1. Extract the effective yaw calibration region data from the raw hyperspectral camera yaw calibration data; S2. Calculate the local range threshold based on the trajectory of the hyperspectral camera's line of sight center during the hyperspectral camera's yaw imaging process; S3. Calculate the local grayscale mean of the effective area data for yaw calibration based on the local range threshold. S4. Fit the spatial dimension of the local gray-level mean to the effective area data of yaw calibration based on the local gray-level mean. S5. Calculate the local relative correction coefficient of the effective yaw calibration area data based on the spatial dimension fitting of the local gray-level mean. S6. Calculate the relative correction coefficients for different gray levels of the effective area data of yaw calibration based on the local relative correction coefficients; S7. The original hyperspectral remote sensing data is subjected to relative radiometric correction frame by frame by using relative correction coefficients of different gray levels, thereby completing the yaw calibration data processing of the backscan type spaceborne hyperspectral camera.

2. The method for processing yaw calibration data of a retrace-type spaceborne hyperspectral camera according to claim 1, characterized in that, Step S1 is as follows: Extract the effective yaw calibration region data Data from the original hyperspectral camera yaw calibration data Data_Org according to the following formula; Data iFrame,iCol,iBand =Data_Org iFrame+iCol-1,iCol,iBand iFrame∈1~Frames iCol∈1~Cols iBand∈1~Bands Wherein: Frames is the total number of frames of data in the effective yaw calibration region, Cols is the total number of columns in the spatial dimension of the yaw calibration data, Bands is the total number of spectral bands in the yaw calibration hyperspectral data, iFrame is the current frame number to be traversed, iCol is the current column number to be traversed, and iBand is the current spectral band number to be traversed.

3. The method for processing yaw calibration data of a retrace-type spaceborne hyperspectral camera according to claim 2, characterized in that, Step S2 is as follows: S2.

1. Based on the longitude coordinate sequence Lon and latitude coordinate sequence Lat of the hyperspectral camera's line-of-sight center trajectory during yaw imaging, the least squares method is used to fit the direction of the straight line and obtain the slope of the fitted straight line. The calculation formula is as follows: Lat=k×Lon+b Where: k is the slope of the fitted line, b is the intercept of the fitted line, np is the number of points in the latitude and longitude coordinate sequence, and i is the number of the current latitude and longitude coordinate sequence points that need to be traversed and calculated. S2.

2. Based on the slope k of the fitted straight line, solve for the inter-frame latitude and longitude deviation of the trajectory coordinate sequence in the direction of the fitted straight line; Let Pi be a point in the trajectory longitude coordinate sequence, and let d_Lon be the latitude and longitude deviation between its neighboring point Pi+1 and Pi in the direction of the fitted straight line. i and d_Lat i Calculate using the following formula; d_Lon i =((P_Lon i+1 -P_Lon i )u_Lon+(P_Lat i -P_Lat i+1 )u_Lat)u_Lon d_Lat i =((P_Lon i+1 -Yield i )u_Lon+(P_Lat i -Lobe i+1 )u_Lat)u_Lat Where: u_Lon and u_Lat are the longitude and latitude components of the normal direction of the fitted straight line, respectively, and P_Lon i Let P_Lat be the longitude coordinate of point P in the i-th frame. i Let d_Lon be the longitude coordinate of point P in the i-th frame, and d_Lat be the dimensional coordinate of point d in the i-th frame. S2.

3. Based on the average latitude and longitude deviation between frames, calculate the average pixel error e using the following formula; Where: Di is the inter-frame physical distance deviation, R is the Earth radius, and GSD is the ground resolution of the hyperspectral camera; S2.

4. Based on the average pixel error e, calculate the local range pixel number threshold T using the following formula; Where: E0 is the set maximum yaw image shift pixel error in the local range.

4. The method for processing yaw calibration data of a retrace-type spaceborne hyperspectral camera according to claim 3, characterized in that, Step S3 is as follows: The local grayscale mean M of the effective yaw calibration data is calculated using the following formula. iT The calculation shows that the total number of columns Cols in the yaw calibration data spatial dimension is divided into nt segments with T as the local range; Where: nt is the maximum number of local ranges contained in a row of data, iT is the current segment in the nt segment, j1 is the pixel position index value of the current segment, and (iT-1)·T+1 to iT·T is the range of pixel position index values ​​of the current segment.

5. The method for processing yaw calibration data of a retrace-type spaceborne hyperspectral camera according to claim 1, characterized in that, Step S4 is as follows: S4.1, Based on the local grayscale mean M iT The piecewise cubic spline fitting method is used to obtain the output fitting coefficient pp using the following formula; pp=spline(X M ,M) M=[M1,M2,…,M iT ] iT∈1~nt Where: spline is a method for solving the cubic spline fitting coefficients, and its input is the local gray-level mean M. iT The sequence X and the coordinates of the intermediate points of the corresponding local range M pp represents the output fitting coefficients; S4.

2. Piecewise cubic spline fitting is used to obtain the relative correction trend term MTrend for each pixel position in the current row of data using the following formula; MTrend=ppval(pp,X Cols ) X Cols =[1,2,…,Cols] MTrend=[MTrend1,MTrend2,…,MTrend Cols ] Where: ppval is a cubic spline fitting method using fitting coefficients, and its input is the yaw calibration effective area data Data column direction coordinate X. Cols The fitting coefficients pp are used to output the trend term MTrend, thus completing the local gray-level mean space dimension fitting.

6. The method for processing yaw calibration data of a retrace-type spaceborne hyperspectral camera according to claim 5, characterized in that, Step S5 is as follows: S5.

1. Remove the relative correction trend term MTrend from the yaw calibration effective area data Data according to the following formula to obtain the data DataT after removing the trend term; DataT iFrame,iCols,iBand =Data iFrame,iCols,iBand -MTrend iCols S5.2 Calculate the relative correction coefficient Coeff according to the following formula; 7. The method for processing yaw calibration data of a retrace-type spaceborne hyperspectral camera according to claim 1, characterized in that, Step S6 is as follows: Each frame of the yaw calibration effective area data Data can be used to calculate a relative correction coefficient Coeff. By statistically analyzing all frames of data, we can obtain the relative correction coefficients Coeffs for different gray levels. The grayscale range of the hyperspectral camera imaging data is divided into L grayscale levels. The grayscale level at which the mean trend term MTrend of each frame of data is located is statistically analyzed. The relative correction coefficient Coeffs is calculated using the following formula. Where: iL represents the iL-th gray level, FrameiL represents the set of frames corresponding to the iL-th gray level, and n iL is the total number of frames in FrameiL, and j is the current frame number in FrameiL that needs to be traversed.

Images