Method and system for strip noise suppression of cross-track wide swath scanning camera

Abstract

The application provides a strip noise suppression method and system of a cross-track wide swath scanning camera, comprising the following steps: S1, calculating the position of each pixel in the along-track direction of the original image data matrix to form a position matrix; S2, calculating a sorting index matrix according to the position matrix, and constructing a rearrangement function and a restoration function based on the sorting index matrix; S3, constructing an optimization problem based on the cross-track direction variation and the rearranged variation data in the along-track direction, and iteratively solving the optimization problem. The application can remove the strip noise while retaining the image discontinuity caused by the earth curvature, can improve the image quality, does not affect the quantitative application, is easy to implement, and can be universally applied to the strip noise suppression of the cross-track wide swath scanning camera.

Description

TECHNICAL FIELD

[0001] The present application relates to a remote sensing image denoising method, in particular to a strip noise suppression method and system of a cross-track wide swath scanning camera. BACKGROUND

[0002] The cross-track wide swath scanning camera is widely used in the field of global remote sensing imaging because it can quickly acquire global images. Typical instruments include the Moderate-resolution Imaging Spectroradiometer (MODIS) carried on the Terra satellite and the Aqua satellite, the Visible Infrared Imaging Radiometer (VIIRS) carried on the NPP satellite, the Medium Resolution Spectral Imager carried on the FY-3 satellite, and the like.

[0003] When the cross-track wide swath scanning camera is used for quantitative applications, the acquired image data needs to be calibrated with high precision. However, due to the non-uniformity of the linear array detector and the change in the on-orbit detector response, the acquired image has strip noise. After the absolute radiometric calibration coefficient is converted into a physical quantity, the image still has a certain degree of strip noise.

[0004] For strip noise, the existing noise suppression methods include: (1) statistical methods, such as histogram matching and moment matching; (2) filtering methods, such as band-pass filtering, wavelet domain filtering, and constrained optimization; and (3) machine learning methods.

[0005] For cross-track wide swath scanning camera, the existing literature [1] (Preesan Rakwat in, Wataru Takeuchi, Yoshifumi Yasuoka, Stripe Noise Reduction in MODIS Data by Combining Histogram Matching With Facet Filter, IEEE Transactions on Geoscience Remote Sensing, 45(6), 2007.) gives a method of combining histogram matching and facet filtering for strip noise suppression. Literature [2] (Marouan Bouali, Said Ladjal, Toward Optimal Destriping of MODIS Data Using a Unidirectional Variational Model, 49(8), 2011.) gives a method of strip removal based on unidirectional variation. Literature [3] (Gang Zhou, Houzhang Fang, Cen Lu, Robust destriping of MODIS and hyperspectral data using a hybrid unidirectional total variation model, Optik, 126(7-8), 2015.) gives a method of strip removal based on unidirectional variation and weight matrix. Literature [4] (Ranil Basnayake, Erik Bollt, Nicholas Tufillaro, Regularization Destriping of Remote Sensing imagery. Nonlinear Processes in Geophysics, 24(3), 2017.) gives a method of strip removal based on unidirectional variation and constraint threshold template. Literature [5] (Pengfei Xiao, Yecai Guo, Peixian Zhuang, Removing Stripe Nose From Infrared Cloud Images via Deep Convolutional Networks, IEEE Photonics Journal, 10(4), 2018.) gives a method of strip noise removal based on convolutional neural network.

[0006] The cross-track wide swath scanning camera has a width of more than 2000km, and due to the influence of the curvature of the earth, the image has a certain particularity. The image of one scanning period corresponds to a "Bowtie" shaped region on the earth. The image at a position deviating from the subsatellite point is discontinuous in a period, which should be distinguished from the discontinuity caused by the strip noise. The existing strip noise suppression method is based on the continuity between adjacent image elements, and does not consider distinguishing and maintaining the discontinuity caused by the "Bowtie" effect at the same time of denoising.

[0007] In view of the defects of the prior art, the optimization solving method combining spatial position rearrangement and one-way variation is adopted to suppress the strip noise of the image along the track direction while trying to ensure the preservation of the details of the original image. SUMMARY

[0008] In view of the defects of the prior art, the purpose of the present application is to provide a strip noise suppression method and system for a cross-track wide swath scanning camera.

[0009] The strip noise suppression method for a cross-track wide swath scanning camera provided by the present application comprises the following steps:

[0010] Step S1: based on the original image data matrix, the position of each pixel in the along-track direction is calculated to form a position matrix;

[0011] Step S2: the sorting index matrix is calculated according to the position matrix, and the rearrangement function and the restoration function are constructed based on the sorting index matrix;

[0012] Step S3: based on the cross-track direction variation and the rearranged variation data in the along-track direction, an optimization problem is constructed and iteratively solved.

[0013] Preferably, in the step S1, the position Y(i,j) of each pixel V(i,j) in the along-track direction is calculated based on the original image data matrix V with a size of M rows and N columns, wherein i represents the row number, i=1, 2,..., M; j represents the column number, j=1, 2,..., N; and a position matrix Y with a size of M rows and N columns is formed.

[0014] Preferably, in the step S1, the position Y(i,j) of each pixel V(i,j) in the along-track direction is calculated based on the original image data matrix V with a size of M rows and N columns, wherein i represents the row number, i=1, 2,..., M; j represents the column number, j=1, 2,..., N; and a position matrix Y with a size of M rows and N columns is formed.

[0015] When the linear array of the cross-track wide swath scanning camera contains P image elements, the original image data matrix V has P rows of data as one scanning period, and the position matrix Y is calculated according to the scanning period. If Y(i,j) is the tth image element of the Tth period, then the row number i=(T-1)P+t.

[0016] Y(i,j)=(t-0.5(P+1))α(j)+0.5(P+1)+(T-1)P (1)

[0017] Wherein, the coefficient α(j) is the distance magnification factor, calculated as follows:

[0018] α(j)=L(j) / L0 (2)

[0019] Where L0 is the line-of-sight distance corresponding to the center of the detector when the camera observes from below the satellite; L(j) is the line-of-sight distance corresponding to the center of the detector at the imaging time of the j-th column of data.

[0020] Preferably, step S2 involves processing each column of the position matrix Y [Y(1,j),Y(2,j),……,Y(M,j)] T Sort the data in ascending order and store the sorted index [B(1,j),B(2,j),……,B(M,j)] T Y(B(1,j),j)≤Y(B(2,j),j)≤……≤Y(B(M,j),j); where T denotes transpose; From the sort index of each column, a sort index matrix B of size M rows and N columns is obtained. The operation of reordering the data matrix V according to the sort index matrix B is denoted as the rearrangement function f. The matrix after reordering the original data matrix V is denoted as V0. * ;

[0021]

[0022] The rearranged matrix V * The operation of restoring to the original position is denoted by the restoration function f. -1 ;

[0023] V = f -1 (V * (4)

[0024] V(B(i,j),j)=V * (i,j) (5).

[0025] Preferably, step S3 employs the following method for variational calculation of the cross-track and along-track directions:

[0026]

[0027]

[0028] in, These are the gradient operators for the cross-track and along-track directions, respectively;

[0029]

[0030]

[0031] wherein the coefficient ε1, the coefficient ε2 are constant normal numbers; The formula 8 can be obtained by the same reason.

[0032] Preferably, the optimization problem adopts:

[0033]

[0034] wherein, represents the optimal solution; U represents the independent variable of the optimization problem; λ represents the weight coefficient; the function g x represents the image variation in the cross-track direction, the function g y represents the image variation in the along-track direction; U * represents the matrix U after rearrangement according to the sorting index matrix B.

[0035] Preferably, the weight coefficient λ is used to balance the fidelity of the image in the cross-track direction and the smoothness in the along-track direction; the weight coefficient λ can be calculated by the following method:

[0036] λ=g y (V * ) / g x (V) (11).

[0037] Preferably, the optimization problem solving adopts the following iterative calculation method, the initial value k=0, U0=V,

[0038]

[0039] wherein, k is the iteration step number, U k is the result obtained by the (k-1)th iteration, is the image after rearrangement of U k , when |U k+1 -U k | is less than a set threshold, the iteration is terminated.

[0040] According to the strip noise suppression system of the cross-track wide swath scanning camera provided by the application, the system comprises:

[0041] Module M1: based on the original image data matrix, the position of each pixel in the along-track direction is calculated to form a position matrix;

[0042] Module M2: the sorting index matrix is calculated according to the position matrix, and the rearrangement function and the restoration function are constructed based on the sorting index matrix;

[0043] Module M3: based on the cross-track direction variation and the rearranged variation data in the along-track direction, the optimization problem is constructed and iteratively solved.

[0044] Preferably, the module M1 employs: calculating the position Y(i,j) of each pixel V(i,j) in the along-track direction based on the original image data matrix V with size of M rows and N columns, where i represents row number, i = 1, 2, …, M; j represents column number, j = 1, 2, …, N; forming the position matrix Y with size of M rows and N columns;

[0045] The module M1 employs:

[0046] When the cross-track wide swath scanning camera linear array contains P pixels in total, the original image data matrix V takes P rows of data as one scanning period, and the calculation method of the position matrix Y is calculated according to the scanning period, if Y(i,j) is the tthpixel in the Tthperiod, then the row number i = (T-1)P + t;

[0047] Y(i,j) = (t-0.5(P+1))a(j) + 0.5(P+1) + (T-1)P (1)

[0048] Wherein, the coefficient a(j) is a distance magnification factor, and the calculation method is

[0049] a(j) = L(j) / L0 (2)

[0050] Wherein, L0 is the line-of-sight distance corresponding to the center of the detector when the camera is observed at the subspace point; L(j) is the line-of-sight distance corresponding to the center of the detector at the imaging moment of the jthcolumn of data;

[0051] The module M2 employs: sorting each column [Y(1,j), Y(2,j), …, Y(M,j)] of the position matrix Y in ascending order, and storing the sorting index [B(1,j), B(2,j), …, B(M,j)] T T Y(B(1,j),j)≤Y(B(2,j),j)≤……≤Y(B(M,j),j); Wherein, T represents transposition; obtaining the sorting index matrix B with size of M rows and N columns from the sorting index of each column, and the operation of reordering the data matrix V according to the sorting index matrix B is recorded as a rearrangement function f, and the matrix of the reordered original data matrix V is recorded as V * ;

[0052]

[0053] The reordered matrix V * is restored to the original position, and the operation is recorded as a restoration function f -1 ;

[0054] V = f -1 (V * ) (4) ​

[0055] V(B(i,j),j) = V * (i,j) (5);

[0056] The module M3 employs the following method for the transversal and along-track direction variational calculation:

[0057]

[0058]

[0059] wherein, are the transversal and along-track direction gradient operators, respectively;

[0060]

[0061]

[0062] wherein, the coefficients ε1 and ε2 are constant normal numbers; which can be obtained by the same reasoning as formula 8;

[0063] The optimization problem employs:

[0064]

[0065] wherein, represents the optimal solution; U represents the independent variable of the optimization problem; λ represents the weight coefficient; the function g x represents the transversal direction image variation; the function g y represents the along-track direction image variation; U * represents the matrix U after rearrangement according to the sorting index matrix B;

[0066] The weight coefficient λ is used to balance the fidelity of the image in the transversal direction and the smoothness in the along-track direction; the weight coefficient λ can be calculated by the following method:

[0067] λ = g y (V * ) / g x (V) (11);

[0068] The optimization problem solving employs the following iterative calculation method, the initial value k = 0, U0 = V,

[0069]

[0070] wherein, k is the iteration step number, U k is the result obtained by the (k-1)th iteration, is the image after rearrangement of U k , when |U k+1 -Uk | less than the set threshold, terminate iteration.

[0071] Compared with the prior art, the present application has the following beneficial effects:

[0072] 1. The present application is aimed at the strip noise of the cross-track wide swath scanning camera, utilizes the spatial position relationship, removes the strip noise while retaining the image discontinuity caused by the earth curvature, can suppress the strip noise caused by the inconsistent response of the detector, improves the image quality and does not affect the quantitative application;

[0073] 2. The method of the present application is reasonable, simple in calculation and easy to implement, and can be universally applied to the strip noise suppression of the cross-track wide swath scanning camera;

[0074] 3. The present application realizes the suppression of the strip noise by combining the geographical spatial position with the along-track smoothing, improves the image quality of the cross-track wide swath scanning camera, and realizes the input for the high-precision quantitative application of the image. BRIEF DESCRIPTION OF DRAWINGS

[0075] Other features, objects and advantages of the present application will become more apparent from the following detailed description of non-limiting embodiments with reference to the following drawings:

[0076] Figure 1 is a flowchart of the present application.

[0077] Figure 2 is a schematic diagram of the spatial position characteristics of the cross-track wide swath scanning camera image.

[0078] Figure 3 is a first-level image of a certain satellite cross-track wide swath scanning camera.

[0079] Figure 4 is a remote sensing image after the strip noise suppression of Figure 3 by the method of the present application.

[0080] Figure 5 is a comparison of the pixel values of one column of the image edges before and after the strip noise suppression. DETAILED DESCRIPTION

[0081] The present application will be described in detail below with reference to specific embodiments. The following embodiments will help those skilled in the art to further understand the present application, but do not limit the present application in any form. It should be noted that those skilled in the art can make several changes and improvements without departing from the concept of the present application. These all belong to the protection scope of the present application.

[0082] Example 1

[0083] The technical problem to be solved by the present application is to provide a strip noise suppression method for a cross-track wide swath scanning camera, to smooth the image along the track direction by using the Bowtie effect, and to ensure the preservation of the details of the original image as much as possible.

[0084] Due to the non-uniformity of the linear array detector and the change in the on-track detector response, the image acquired by the cross-track wide swath scanning camera has strip noise, which is distributed along the scanning direction. Figure 1 The strip noise suppression method for the cross-track wide swath scanning camera according to the present application is shown in the flowchart

[0085]

[0086] Among them, is the optimal solution, U is the independent variable of the optimization problem, and λ is the weight coefficient. x is the image variation in the cross-track direction, and g y is the image variation in the along-track direction. U * is the matrix U rearranged according to the sorting index matrix B.

[0087] The camera image obtained by the linear array for cross-track wide swath scanning is affected by the curvature of the earth, and the image of one scanning period is in the shape of Bowtie (bowtie), and multiple periods are arranged in the area deviating from the sub-satellite point, that is, there is periodic image discontinuity, as shown in Figure 2 When suppressing the strip noise caused by the inconsistency of the detector, the image discontinuity caused by the Bowtie effect needs to be preserved.

[0088] If the linear array of the cross-track wide swath scanning camera contains P pixels in total, the image V takes P rows of data as one scanning period, and the calculation method of the position matrix Y in step 1 is calculated according to the scanning period, if Y(i,j) is the tth pixel of the Tth period, that is, the row number i=(T-1)P+t,

[0089] Y(i,j)=(t-0.5(P+1))α(j)+0.5(P+1)+(T-1)P (Formula 2)

[0090] Among them, the coefficient α(j) is the distance magnification factor, and the calculation method is

[0091] α(j)=L(j) / L0 (Formula 3)

[0092] L0 is the line of sight distance corresponding to the detector center when the camera is at the sub-space point, and L(j) is the line of sight distance corresponding to the detector center when the jth column of data is imaged.

[0093] For each column of the position matrix Y [Y(1,j), Y(2,j), …, Y(M,j)] T , sort them in ascending order and store the sorting index [B(1,j), B(2,j), …, B(M,j)] T , i.e. Y(B(1,j),j)≤Y(B(2,j),j)≤…≤Y(B(M,j),j). From the sorting index of each column, a sorting index matrix B with M rows and N columns is obtained. The operation of rearranging the data matrix according to the sorting index matrix B is denoted as a rearrangement function f, and the rearranged matrix of the original data matrix V is denoted as V * .

[0094]

[0095] The rearranged matrix V * is restored to the original position, and the operation is denoted as a restoration function f -1 , V = f -1 (V * ), i.e.

[0096] V(B(i,j),j) = V * (i,j) (Equation 5)

[0097] The original image has the characteristics of continuity in the cross-track direction and gradually discontinuity in the along-track direction away from the sub-space point in the spatial position. Therefore, the variational calculation in the cross-track and along-track directions adopts the following method:

[0098]

[0099]

[0100] are the gradient operators in the cross-track and along-track directions, respectively.

[0101]

[0102]

[0103] The coefficients ε1 and ε2 are constant numbers, which are used to ensure that the optimization problem described in (Equation 1) is differentiable in the defined domain. can be obtained by (Equation 8). Without loss of generality, we can take ε1 = 1 × 10 -6 ; ε2 = 1 × 10 -6 .

[0104] The weight coefficient λ in the formula (1) is used to balance the fidelity of the image in the cross-track direction and the smoothness in the along-track direction. If λ is too large, the image may lose details, and if λ is too small, the image may not be suppressed enough. When the intensity of the strip noise in the image is not uniform, the weight coefficient λ needs to be adaptively changed. Therefore, the weight coefficient λ can be calculated by the following method:

[0105] λ = g y (V * ) / g x (V) (Formula 10)

[0106] g y (V * ) can be obtained by (Formula 7), (Formula 9), and g x (V) can be obtained by (Formula 6), (Formula 8).

[0107] The optimization problem in the formula (1) is solved by using the gradient descent algorithm, and the kth iteration process is as follows,

[0108]

[0109] The initial value k = 0, U0 = V, U k is the result obtained by the (k-1)th iteration, is the image after rearrangement of U k , and each iteration judges the difference between U k+1 and U k , and when |U k+1 -U k | is less than a set threshold, the iteration is terminated.

[0110] The strip noise suppression system of the cross-track wide swath scanning camera provided by the present application can be realized by the step flow in the strip noise suppression method of the cross-track wide swath scanning camera provided by the present application. Those skilled in the art can understand that the strip noise suppression method of the cross-track wide swath scanning camera is a preferred example of the strip noise suppression system of the cross-track wide swath scanning camera.

[0111] The method of the present application is verified by a certain satellite remote sensing image, Figure 3 The image is a first-level remote sensing data image of a certain satellite cross-track wide swath scanning camera. The first-level data has been calibrated absolutely, but due to the existence of non-ideal factors, there is still strip noise in the image. Figure 4 The remote sensing image after strip noise suppression of Figure 3 by the method of the present application. Figure 4This indicates that stripe noise has been suppressed, and in the regions on the left and right sides of the image that are far from the nadir point, discontinuities related to the scan period have been preserved. To further examine the suppression of stripe noise on the left and right sides of the image, [the following parameters were used] respectively. Figure 3 and Figure 4 After selecting the edge column (column 5), rearranging it, and comparing the results, the following is shown: Figure 5 As shown. Figure 5 This indicates that, along the track direction, the image after noise suppression and rearrangement has good continuity and requires less correction to the pixel values ​​of the original image, which is beneficial for the quantitative application of this first-level remote sensing data image.

[0112] Those skilled in the art will understand that, in addition to implementing the system, apparatus, and their modules provided by this invention in purely computer-readable program code, the same program can be implemented in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers by logically programming the method steps. Therefore, the system, apparatus, and their modules provided by this invention can be considered a hardware component, and the modules included therein for implementing various programs can also be considered structures within the hardware component; alternatively, modules for implementing various functions can be considered both software programs implementing the method and structures within the hardware component.

[0113] Specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art can make various changes or modifications within the scope of the claims, which do not affect the essence of the present invention. Unless otherwise specified, the embodiments and features described in this application can be arbitrarily combined with each other.

Claims

1. A method of strip noise suppression for a cross-track swath scanning camera, the method comprising: Comprising: Step S1: calculating the position of each pixel in the along-track direction based on the original image data matrix to form a position matrix; Step S2: calculating a sorting index matrix according to the position matrix, and constructing a rearrangement function and a restoration function based on the sorting index matrix; Step S3: constructing an optimization problem and iteratively solving it based on the cross-track direction variation and the rearranged variation data in the along-track direction; The step S3 adopts: the cross-track and along-track direction variation calculation adopts the following method: wherein, respectively are the cross-rail and along-rail gradient operators, V represents the original image data matrix with size of M rows and N columns, i represents the row number, i = 1, 2, …, M; j represents the column number, j = 1, 2, …, N; wherein the coefficient ε1, the coefficient ε2 are constant normal numbers; By the same reasoning from equation 8; The optimization problem adopts: wherein, represents the optimal solution; U represents the argument of the optimization problem; λ represents a weight coefficient; function g x represents the image variation across the track direction, function g y represents the image variation along the track direction; U * represents the matrix U after being rearranged according to the ordering index matrix B; The weight coefficient λ is used to balance the fidelity of the image in the cross-track direction and the smoothness in the along-track direction; the weight coefficient λ is calculated by the following method: λ = g y (V * ) / g x (V) (11) The optimization problem is solved by using the following iterative calculation method, initial value k = 0, U0 = V, V * represents the reordered matrix of the original data matrix V; where k is the iteration step number, U k is the result of the (k-1)th iteration, is the image obtained by rearranging U k when |U k+1 -U k is less than a set threshold value.

2. The method of strip noise suppression for a cross-track wide swath scanning camera of claim 1, wherein, The step S1 adopts: based on the original image data matrix V with a size of M rows and N columns, the position Y(i,j) of each pixel V(i,j) in the along-track direction is calculated to form a position matrix Y with a size of M rows and N columns.

3. The method of strip noise suppression for a cross-rail wide-swath scanning camera of claim 2, wherein, The step S1 adopts: When the cross-track wide swath scanning camera linear array contains P pixels in total, the original image data matrix V takes P rows of data as one scanning period, and the calculation method of the position matrix Y is calculated according to the scanning period; if Y(i,j) is the tth pixel in the Tth period, then the row number i=(T-1)P+t; Y(i,j)=(t-0.5(P+1))α(j)+0.5(P+1)+(T-1)P (1) Wherein, the coefficient α(j) is a distance magnification factor, and the calculation method is α(j)=L(j) / L0 (2) Wherein, L0 is the line-of-sight distance corresponding to the center of the detector when the camera is observed at the subspace point; L(j) is the line-of-sight distance corresponding to the center of the detector at the imaging time of the jth column of data.

4. The method of strip noise suppression for a cross-rail wide-swath scanning camera of claim 1, wherein, The step S2 adopts: for each column of the position matrix Y [Y(1, j), Y(2, j), …, Y(M, j)] T , sorting from small to large, and storing the sorting indexes [B(1, j), B(2, j), …, B(M, j)] T , Y(B(1, j), j)≤Y(B(2, j), j)≤……≤Y(B(M, j), j); wherein T represents transposition; the sorting index matrix B with M rows and N columns is obtained from the sorting indexes of each column, the operation of reordering the data matrix V according to the sorting index matrix B is recorded as a rearrangement function f, and the reordered matrix of the original data matrix V is recorded as V * ; The rearranged matrix V * The operation to restore to the original position is denoted as a restore function f -1 ; V = f -1 (V * )(4) V(B(i,j),j) = V * (i,j)(5).

5. A system for strip noise suppression across a wide-swath scanning camera, the system comprising: Comprising: Module M1: calculating the position of each pixel in the along-track direction based on the original image data matrix to form a position matrix; Module M2: calculating a sorting index matrix according to the position matrix, and constructing a rearrangement function and a restoration function based on the sorting index matrix; Module M3: constructing an optimization problem and iteratively solving it based on the cross-track direction variation and the rearranged variation data in the along-track direction; The module M3 adopts: the cross-track and along-track direction variation calculation adopts the following method: wherein, are the cross-rail and along-rail gradient operators, respectively, V represents the original image data matrix of size M rows by N columns, i represents the row number, i = 1, 2,..., M; j represents the column number, j = 1, 2,..., N; wherein the coefficient ε1, the coefficient ε2 are constant normal numbers; By the same reasoning from equation 8; The optimization problem adopts: wherein, denotes the optimal solution; U denotes the argument of the optimization problem; λ denotes a weight coefficient; g x denotes the image variation across the track direction, g y denotes the image variation along the track direction; U * denotes the matrix U after reordering according to the ordering index matrix B; The weight coefficient λ is used to balance the fidelity of the image in the cross-track direction and the smoothness in the along-track direction; the weight coefficient λ is calculated by the following method: λ = g y (V * ) / g x (V)(11) The optimization problem is solved by using the following iterative calculation method, initial value k = 0, U0 = V, V * represents the reordered matrix of the original data matrix V; where k is the iteration step number, U k is the result of the (k-1)th iteration, is the image obtained by rearranging U k when |U k+1 -U k | is less than a set threshold.

6. The system for strip noise suppression of a cross-track wide swath scanning camera of claim 5, wherein, The module M1 adopts: based on the original image data matrix V with a size of M rows and N columns, the position Y(i,j) of each pixel V(i,j) in the along-track direction is calculated to form a position matrix Y with a size of M rows and N columns. The module M1 adopts: When the cross-track wide swath scanning camera linear array contains P pixels in total, the original image data matrix V takes P rows of data as one scanning period, and the calculation method of the position matrix Y is calculated according to the scanning period; if Y(i,j) is the tth pixel in the Tth period, then the row number i=(T-1)P+t; Y(i,j)=(t-0.5(P+1))α(j)+0.5(P+1)+(T-1)P (1) Wherein, the coefficient α(j) is a distance magnification factor, and the calculation method is α(j)=L(j) / L0 (2) Wherein, L0 is the line-of-sight distance corresponding to the center of the detector when the camera is observed at the subspace point; L(j) is the line-of-sight distance corresponding to the center of the detector at the imaging time of the jth column of data. Wherein, L0 is the line-of-sight distance corresponding to the detector center when the camera is observing the subspace point; L(j) is the line-of-sight distance corresponding to the detector center when the jth column of data is imaging; The module M2 adopts: for each column of the position matrix Y [Y(1, j), Y(2, j), …, Y(M, j)] T , sorting from small to large, and storing the sorting indexes [B(1, j), B(2, j), …, B(M, j)] T , Y(B(1, j), j)≤Y(B(2, j), j)≤……≤Y(B(M, j), j); wherein T represents transposition; the sorting index matrix B with M rows and N columns is obtained from the sorting indexes of each column, the operation of reordering the data matrix V according to the sorting index matrix B is recorded as a rearrangement function f, and the reordered matrix of the original data matrix V is recorded as V * ; The rearranged matrix V * The operation to restore to the original position is denoted as a restore function f -1 ; V = f -1 (V * )(4) V(B(i,j),j) = V * (i,j)(5).

Images