Method for extracting one-thousandth pixel of corner reflector
通过构建距离-多普勒模型和傅里叶变换技术,实现了角反射器的千分之一像素级精确提取,解决了现有技术中精度不足的问题,提高了卫星成像质量和定位准确性 , suitable for high-precision application scenarios.
Patent Information
- Application Number
- CN202410989130.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-23
- Publication Date
- 2024-11-12
- Estimated Expiration
- 2044-07-23
AI Technical Summary
The prior art is difficult to achieve accurate extraction of one thousandth of pixel level of angular reflectors, resulting in limited satellite imaging accuracy, especially in high-precision terrain mapping, which may lead to significant solution errors.
By constructing a distance-Doppler model, combining Newton's iterative method and Fourier transform technology, the initial pixel coordinates of the angle reflector are gradually solved, and the pixel coordinates of the angle reflector are accurately positioned through multiple upsampling and reconstruction matrix processing. , achieves one thousandth of pixel level accuracy.
It greatly improves the accuracy of angular reflector positioning, improves the imaging quality of SAR satellites, reduces the impact of errors, and is suitable for high-precision application scenarios such as geometry calibration and deformation monitoring, improving data processing efficiency.
Smart Images

Figure CN118938219A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of satellite measurement, and in particular relates to a method for extracting one thousandth of a pixel from a corner reflector. Background Art
[0002] Synthetic Aperture Radar (SAR) satellites play an irreplaceable role in natural resource management, emergency disaster reduction and other fields with their unique all-day and all-weather imaging capabilities. In order to ensure the high quality and accuracy of SAR satellite imaging, ground corner reflectors (CR) are required to perform geometric calibration after the satellite is launched, so that the satellite's plane positioning accuracy reaches an extremely high level. Among them, the precise extraction of corner reflector coordinates in SAR images is the basis for research such as geometric calibration, interferometric calibration and deformation monitoring.
[0003] Corner reflectors, with their unique geometric structure and strong echo characteristics, appear as bright spots with high signal-to-noise ratio in SAR images, making them easy to identify and locate. However, despite certain progress in the design and research of corner reflectors, such as the selection of reflector surface materials, design processes, manufacturing tolerances, and angle accuracy, research on precise corner reflector extraction technology is still insufficient, especially the lack of extraction technology that can achieve one thousandth of a pixel level accuracy.
[0004] Currently, the extraction of corner reflectors mainly relies on the contrast between the strong echo signals they generate in the image and the weaker echoes of the surrounding terrain. Common methods include identifying and matching corner reflectors in the image through window templates and calculating correlation coefficients to determine their precise positions. Although this method is computationally efficient, its accuracy is usually limited to the pixel level. Another technique identifies reflector points by calculating the coherence coefficient and amplitude difference within a window around the corner reflector. Although this method has high accuracy, it is costly to process large amounts of data.
[0005] The rough identification method of corner reflectors based on the mathematical model of satellite-ground position, combined with the strong echo characteristics of corner reflectors, can determine the approximate range of corner reflectors. However, the errors existing in the imaging process of the SAR system will cause the rough identification offset to be too large, and the influence of surrounding objects will increase, limiting the accuracy of corner reflector point extraction.
[0006] In addition, existing technologies generally lack effective quality assessment and verification methods, making it difficult to accurately evaluate the accuracy of extraction results, which may lead to significant solution errors in high-precision topographic mapping.
[0007] Therefore, in order to overcome the shortcomings of the above-mentioned technologies, the present invention proposes a method for extracting one-thousandth pixel of a corner reflector, aiming to provide the geodetic coordinates of any corner reflector, and solve the one-thousandth pixel coordinates of the corner reflector through technology, thereby providing strong technical support for the precise extraction, geometric calibration and interferometric measurement calibration of the phase center of the corner reflector. Summary of the invention
[0008] In order to solve the above technical problems, the present invention provides a method for extracting one thousandth of a pixel of a corner reflector, which can achieve a pixel positioning accuracy of one thousandth of a pixel for the corner reflector.
[0009] In order to solve the above technical problems, the present invention is achieved through the following technical solutions:
[0010] The present invention provides a method for extracting one thousandth of a pixel of a corner reflector, comprising the following steps:
[0011] Step S1: acquiring raw data, wherein the raw data includes the geodetic coordinates of the corner reflector point, the SAR image and the imaging parameters of the SAR image;
[0012] Step S2: constructing a range-Doppler (RD) model, and using the acquired raw data to calculate the initial pixel coordinates (x0, y0) of the corner reflector point according to the RD model;
[0013] Step S3, based on the initial pixel coordinates (x0, y0) of the corner reflector point obtained in step S2, the pixel coordinates (x3, y3) of the corner reflector point in the SAR image are calculated.
[0014] As a preferred technical solution of the present invention, step S2 specifically includes the following sub-steps:
[0015] Step S2.1: converting the geodetic coordinates of the corner reflector point into a geocentric rectangular coordinate system;
[0016] Step S2.2: fitting the satellite orbit through the SAR satellite state vector to obtain the SAR satellite state vector at any time;
[0017] Step S2.3: Calculate the azimuth imaging time t of the corner reflector point on the image, specifically according to the start time and time sampling interval and the row number of the corner reflector point on the SAR image, the formula is: t = t0 + t i ·i;
[0018] Step S2.4: construct RD model;
[0019] Step S2.5: Solve the RD model using the Newton iteration method to obtain the initial pixel coordinates (x0, y0) of the corner reflector point.
[0020] As a preferred technical solution of the present invention, the conversion formula used for the conversion of geodetic coordinates to geocentric rectangular coordinates in step S2.1 is:
[0021]
[0022] As a preferred technical solution of the present invention, the satellite orbit fitting in step S2.2 adopts polynomial fitting. Specifically, the polynomial is:
[0023]
[0024] As a preferred technical solution of the present invention, the RD model in step S2.4 is as follows:
[0025] Distance equation:
[0026] Doppler equation:
[0027] The equation of the Earth's ellipsoid:
[0028] As a preferred technical solution of the present invention, step S3 specifically includes the following sub-steps:
[0029] Step S3.1, taking the initial pixel coordinates (x0, y0) of the corner reflector point obtained in step S2 as the center, select a complex matrix with a window size of N1×N1;
[0030] Step S3.2, calculating the amplitude value of the N1×N1 complex matrix to obtain its amplitude value matrix; specifically, the amplitude value is calculated according to the following formula:
[0031]
[0032] Among them, Re(i,j) represents the real part of the pixel value (i,j), and Im(i,j) represents the imaginary part of the pixel value (i,j);
[0033] Step S3.3, finding the pixel coordinates (x1, y1) corresponding to the maximum amplitude value sig(i, j) described in step S3.2;
[0034] Step S3.4, taking the pixel coordinate (x1, y1) described in step S3.3 as the center, select a complex matrix with a window size of 2N2×2N2;
[0035] Step S3.5, upsampling the complex matrix with a window size of 2N2×2N2 by a factor of M1 (M1<1024) to obtain an upsampled and reconstructed complex matrix;
[0036] Step S3.6, calculating the amplitude value of the upsampled and reconstructed complex matrix to obtain its amplitude value matrix sig1(i,j);
[0037] Step S3.7, finding the pixel coordinates corresponding to the maximum amplitude value in step S3.6 is (x2, y2);
[0038] Step S3.8, upsampling the amplitude value of the row where the pixel coordinate x2 is located (i.e., the distance direction) by a factor of M2 to obtain a one-dimensional complex matrix Dx2, and upsampling the amplitude value of the column where the pixel coordinate y2 is located (i.e., the azimuth direction) by a factor of M2 to obtain a one-dimensional complex matrix Dy2;
[0039] Step S3.9, calculate the amplitude value matrix of the one-dimensional complex matrix Dx2 as The amplitude matrix of the one-dimensional complex matrix Dy2 is
[0040] Step S3.10, calculate the amplitude value matrix and The decibel (db) values are and Specifically, the decibel value is calculated according to the following formula:
[0041]
[0042] Among them, max(*) represents the maximum amplitude value in the amplitude value matrix;
[0043] Step S3.11, get the distance db value All db values from 3db on the left to 3db on the right form a new matrix
[0044] Step S3.12, according to the distance to db value The pixel coordinates and db values are fitted with trigonometric functions;
[0045] Step S3.13, find The coordinates corresponding to the peak point and the coordinates corresponding to the fitting value are determined, and the absolute value of the difference between the coordinate x3 corresponding to the peak point and the coordinate x'3 corresponding to the fitting value reaches the order of 1 / 1000, and the coordinate x3 corresponding to the distance to the peak point is output;
[0046] Step S3.14, azimuth db value Repeat S3.11-S3.13 and output the coordinate y3 corresponding to the azimuth peak point.
[0047] As a preferred technical solution of the present invention, step S3.5 specifically includes the following sub-steps:
[0048] Step S3.5.1, performing a two-dimensional fast Fourier transform on the complex matrix obtained in S3.4 to convert it into the frequency domain;
[0049] Step S3.5.2, upsampling the complex matrix transferred to the frequency domain by a factor of M1 to obtain an upsampled complex matrix
[0050] Step S3.5.3, reconstructing the upsampled complex matrix using the two-dimensional inverse Fourier transform The reconstructed complex matrix D2[M1×2N2,M1×2N2] is obtained.
[0051] As a preferred technical solution of the present invention, step S3.52 specifically includes the following sub-steps:
[0052] Step S3.5.2.1, calculate the amplitude value of the complex matrix D2[M1×2N2,M1×2N2] obtained in step S3.5.1 And calculate the sum of the amplitude values of each column to get the matrix tmp row (i,1);
[0053] Step S3.5.2.2, find step S3.5.2.1tmp row The row number i1 where the minimum value in (i,1) is located;
[0054] Step S3.5.2.3, construct a two-dimensional zero matrix O1[(M1-1)×2N2,2N2];
[0055] Step S3.5.2.4, the two-dimensional zero matrix obtained in step S3.5.2.3
[0056] O1[(M1-1)×2N2,2N2] insert sig D2 The i1+1th row in (i,j) and the matrix
[0057] Step S3.5.2.5, for step S3.5.2.4 matrix Calculate the sum of the amplitude values of each row to get the matrix tmp col (1,j);
[0058] Step S3.5.2.6, find the matrix of step S3.5.2.5 The column number j1 where the minimum value is located;
[0059] Step S3.5.2.7, construct a two-dimensional zero matrix O2[M1×2N2,(M1-1)×2N2];
[0060] Step S3.5.2.8, the two-dimensional zero matrix obtained in step S3.5.2.7
[0061] O2[M1×2N2,(M1-1)×2N2] insert The j1+1th column in the matrix The processing of the above eight sub-steps completes the first M1-fold upsampling.
[0062] As a preferred technical solution of the present invention, step S3.8 specifically includes the following sub-steps:
[0063] Step S3.8.1, determine the upsampling factor M2; specifically, determine M2 according to the following formula:
[0064] M2=1024 / M1
[0065] Step S3.8.2, upsample the amplitude value of the row where the pixel coordinate x2 is located (i.e., the distance direction) by M2 times to obtain a one-dimensional complex matrix Dx2, and upsample the amplitude value of the column where the pixel coordinate y2 is located (i.e., the azimuth direction) by M2 times to obtain a one-dimensional complex matrix Dy2.
[0066] As a preferred technical solution of the present invention, step S3.8.2 specifically includes the following sub-steps:
[0067] Step S3.8.2.1, construct a one-dimensional zero matrix and determine the size N3 of the one-dimensional zero matrix O; specifically, determine N3 according to formula (10):
[0068] N3=(2N2×M1)×M2
[0069] =1024*2N2
[0070] Step S3.8.2.2, perform spectrum centering processing on the amplitude value of the row where the pixel coordinate x2 is located (i.e., the distance direction), then perform fast Fourier transform on the amplitude value after spectrum centering processing, and finally perform spectrum centering processing again; this processing is to make the low-frequency part of the spectrum located at the beginning of the array and the high-frequency part located at the end of the array;
[0071] Step S3.8.2.3, assign the result obtained in step S3.8.2.2 to the zero matrix O3 Within the range, the assigned matrix Dx2 is obtained;
[0072] Step S3.8.2.4, first perform spectrum centering on Dx2 assigned in step S3.8.2.3, then perform inverse fast Fourier transform, and finally perform spectrum centering again; this operation ensures that the format of the final result is consistent with the original signal;
[0073] Step S3.8.2.5, the processing of the above four sub-steps completes the M2-fold upsampling of the amplitude value of the row where the pixel coordinate x2 is located (i.e., the distance direction); the amplitude value resampling in the azimuth direction repeats steps S3.8.2.1-step S3.8.2.5.
[0074] The present invention has the following beneficial effects:
[0075] 1. Greatly improve the positioning accuracy of corner reflectors;
[0076] 2. Improve SAR satellite imaging quality: By accurately extracting the coordinates of the corner reflectors, this technology can improve the imaging quality of SAR satellites, which is crucial for applications in natural resource management, emergency disaster reduction and other fields;
[0077] 3. Enhanced positioning accuracy: This method can provide higher positioning accuracy and reduce errors caused by manual puncture points through innovative algorithms and data processing technologies;
[0078] 4. Support high-precision applications: This technology is particularly suitable for application scenarios that require high-precision corner reflector positioning, such as satellite geometry calibration, interferometry calibration and deformation monitoring;
[0079] 5. Reduce the impact of errors: Compared with the prior art, this invention has significant advantages in reducing the impact of corner reflector extraction errors, especially in applications such as high-precision terrain mapping;
[0080] 6. Improve data processing efficiency: This method not only improves the accuracy of solution, but also optimizes the data processing efficiency, making it more efficient than existing technologies.
[0081] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS
[0082] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for describing the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.
[0083] Figure 1 It is a flow chart of the method for extracting one thousandth pixel of a mid-corner reflector;
[0084] Figure 2 It is a schematic diagram of the geometric relationship of synthetic aperture radar satellite positioning;
[0085] Figure 3 This is a schematic diagram of the imaging of a corner reflector on a SAR image. DETAILED DESCRIPTION
[0086] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0087] like Figure 1 As shown, the method for extracting one thousandth of a pixel of a corner reflector includes the following steps:
[0088] Step S1, acquiring raw data, wherein the raw data includes the coordinates of the corner reflector point, the SAR image and the imaging parameters of the SAR image;
[0089] Step S2, constructing a range-Doppler (RD) model, and using the acquired raw data to calculate the initial image coordinates (x0, y0) of the corner reflector point according to the RD model;
[0090] like Figure 2 As shown, specifically, step S2 specifically includes the following:
[0091] Step S2.1, converting the geodetic coordinates of the corner reflector point into a geocentric rectangular coordinate system; specifically, the conversion formula for converting the geodetic coordinates into a geocentric rectangular coordinate system is as follows:
[0092]
[0093] Among them, (X P , Y P , Z P ) is the geocentric rectangular coordinate of the corner reflector point, N is the radius of curvature of the yoke, (B, L, H) are the latitude, longitude and geodetic height of the geodetic coordinate of the corner reflector point, and e is the first eccentricity of the reference ellipsoid;
[0094] Step S2.2, fitting the satellite orbit through the SAR satellite state vector, thereby obtaining the SAR satellite state vector at any time, and the position vector and velocity vector of the satellite sensor at any time can be fitted by a polynomial; specifically, the polynomial is:
[0095]
[0096] Among them, (X s , Y s , Z s ) is the position vector to be fitted, (V xs , Yys , Z zs ) is the velocity vector to be fitted, is the coefficient matrix of the position fitting polynomial, is the coefficient matrix of the velocity fitting polynomial, t is the azimuth imaging time of the corner reflector point on the image;
[0097] Step S2.3, calculate the azimuth imaging time of the corner reflector point on the image; specifically, calculate t according to formula (4):
[0098] t=t0+t i ·i (4)
[0099] Among them, t0 is the starting time of the SAR image azimuth, t i is the time sampling interval of the SAR image in azimuth, and i is the row number of the corner reflector point on the SAR image.
[0100] Step S2.4, constructing an RD model, the RD model is as follows:
[0101] Distance equation:
[0102] Doppler equation:
[0103] The equation of the Earth's ellipsoid:
[0104] Among them, R gc =[X g Y g Z g ] T Represents the position vector of the ground point, R sc =[X s Y s Z s ] T represents the position vector of the satellite sensor, r0 is the starting slant distance of the SAR image range, m j is the range resolution of the SAR image, j is the column number of the corner reflector point on the SAR image, and f d is the Doppler frequency, R e represents the major axis of the Earth ellipsoid, R p H represents the minor semi-axis of the Earth ellipsoid. t is the elevation value of the ground point;
[0105] Step S2.5, using Newton iteration method to solve the RD model constructed in step S2.4 to solve the initial pixel coordinates of the corner reflector point;
[0106] Step S3, based on step S2, the initial image coordinates of the corner reflector point are obtained, and the pixel coordinates (x3, y3) of the corner reflector point in the SAR image are calculated using a one-thousandth pixel extraction method;
[0107] like Figure 3 As shown, specifically, step S3 specifically includes the following:
[0108] Step S3.1, taking the initial pixel coordinates (x0, y0) of the corner reflector point obtained in step S2 as the center, select a complex matrix with a window size of N1×N1;
[0109] Step S3.2, calculate the amplitude value of the N1×N1 complex matrix to obtain its amplitude value matrix; specifically, the amplitude value is calculated according to formula (8):
[0110]
[0111] Among them, Re(i,j) represents the real part of the pixel value (i,j), and Im(i,j) represents the imaginary part of the pixel value (i,j);
[0112] Step S3.3, finding the pixel coordinates (x1, y1) corresponding to the maximum amplitude value sig(i, j) described in step S3.2;
[0113] Step S3.4, taking the pixel coordinate (x1, y1) described in step S3.3 as the center, select a complex matrix D1[2N2, 2N2] with a window size of 2N2×2N2;
[0114] Step S3.5, upsampling the complex matrix with a window size of 2N2×2N2 by a factor of M1 (M1<1024), and obtaining an upsampled and reconstructed complex matrix D2[M1×2N2,M1×2N2];
[0115] Step S3.5.1, perform a two-dimensional fast Fourier transform on the complex matrix D2[M1×2N2,M1×2N2] obtained in S3.4 to convert it into the frequency domain;
[0116] Step S3.5.2, upsampling the complex matrix transferred to the frequency domain by a factor of M1 to obtain an upsampled complex matrix
[0117] Step S3.5.2.1, calculate the amplitude value of the complex matrix D2[M1×2N2,M1×2N2] obtained in step S3.5.1 And calculate the sum of the amplitude values of each column to get the matrix tmp row (i,1);
[0118] Step S3.5.2.2, find step S3.5.2.1tmp row The row number i1 where the minimum value in (i,1) is located;
[0119] Step S3.5.2.3, construct a two-dimensional zero matrix O1[(M1-1)×2N2,2N2];
[0120] Step S3.5.2.4, the two-dimensional zero matrix obtained in step S3.5.2.3
[0121] O1[(M1-1)×2N2,2N2] insert The i1+1th row in
[0122] Step S3.5.2.5, for step S3.5.2.4 matrix Calculate the sum of the amplitude values of each row to get the matrix tmp col (1,j);
[0123] Step S3.5.2.6, find the matrix of step S3.5.2.5 The column number j1 where the minimum value is located;
[0124] Step S3.5.2.7, construct a two-dimensional zero matrix O2[M1×2N2,(M1-1)×2N2];
[0125] Step S3.5.2.8, insert the two-dimensional zero matrix O2[M1×2N2,(M1-1)×2N2] obtained in step S3.5.2.7 into The j1+1th column in the matrix The processing of the above eight sub-steps completes the first M1-fold upsampling;
[0126] Step S3.5.3, the matrix obtained in step S3.5.2 Using the two-dimensional inverse Fourier transform, we get the complex matrix D2[M1×2N2,M1×2N2];
[0127] Step S3.6, calculate the amplitude value of the upsampled complex matrix D2[M1×2N2,M1×2N2] to obtain its amplitude value matrix sig1(i,j); specifically, calculate its amplitude value according to formula (8);
[0128] Step S3.7, finding the pixel coordinates (x2, y2) corresponding to the maximum amplitude value described in step S3.6;
[0129] Step S3.8, upsampling the amplitude value of the row where the pixel coordinate x2 is located (i.e., the distance direction) by a factor of M2 to obtain a one-dimensional complex matrix Dx2, and upsampling the amplitude value of the column where the pixel coordinate y2 is located (i.e., the azimuth direction) by a factor of M2 to obtain a one-dimensional complex matrix Dy2;
[0130] Step S3.8.1, determine the upsampling factor M2; specifically, determine M2 according to formula (9):
[0131] M2=1024 / M1 (9)
[0132] Step S3.8.2, upsampling the amplitude value of the row where the pixel coordinate x2 is located (i.e., the distance direction) by a factor of M2 to obtain a one-dimensional complex matrix Dx2, and upsampling the amplitude value of the column where the pixel coordinate y2 is located (i.e., the azimuth direction) by a factor of M2 to obtain a one-dimensional complex matrix Dy2;
[0133] Step S3.8.2.1, construct a one-dimensional zero matrix and determine the size N3 of the one-dimensional zero matrix O; specifically, determine N3 according to formula (10):
[0134]
[0135] Step S3.8.2.2, perform spectrum centering processing on the amplitude value of the row where the pixel coordinate x2 is located (i.e., the distance direction), then perform fast Fourier transform on the amplitude value after spectrum centering processing, and finally perform spectrum centering processing again; this processing is to make the low-frequency part of the spectrum located at the beginning of the array and the high-frequency part located at the end of the array;
[0136] Step S3.8.2.3, assign the result obtained in step S3.8.2.2 to the zero matrix O3 Within the range, the assigned matrix Dx2 is obtained;
[0137] Step S3.8.2.4, first perform spectrum centering on Dx2 assigned in step S3.8.2.3, then perform inverse fast Fourier transform, and finally perform spectrum centering again; this operation ensures that the format of the final result is consistent with the original signal;
[0138] Step S3.8.2.5, the processing of the above four sub-steps completes the M2-fold upsampling of the amplitude value of the row where the pixel coordinate x2 is located (i.e., the range direction); the resampling of the amplitude value in the azimuth direction repeats steps S3.8.2.1 to S3.8.2.5;
[0139] Step S3.9, calculate the amplitude value matrix of the one-dimensional complex matrix Dx2 as The amplitude matrix of the one-dimensional complex matrix Dy2 is
[0140] Step S3.10, calculate the amplitude value matrix and The decibel (db) values are and Specifically, the decibel value is calculated according to the following formula:
[0141]
[0142] Among them, max(*) represents the maximum amplitude value in the amplitude value matrix;
[0143] Step S3.11, get the distance db value All db values from 3db on the left to 3db on the right form a new matrix
[0144] Step S3.12, according to the distance to db value The pixel coordinates and db values are fitted with trigonometric functions;
[0145] Step S3.13, find The coordinates corresponding to the peak point and the coordinates corresponding to the fitting value are determined, and the absolute value of the difference between the coordinate x3 corresponding to the peak point and the coordinate x'3 corresponding to the fitting value reaches the order of 1 / 1000, and the output distance is the coordinate x3 corresponding to the actual value of the peak point;
[0146] Step S3.14, azimuth db value Repeat S3.11-S3.13 and output the coordinate y3 corresponding to the azimuth peak point.
[0147] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.
[0148] The preferred embodiments of the present invention disclosed above are only used to help illustrate the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.
Claims
1. A method for extracting one thousandth of a pixel from a corner reflector, characterized in that: The following steps are involved: Step S1: acquiring raw data, wherein the raw data includes the geodetic coordinates of the corner reflector point, the SAR image and the imaging parameters of the SAR image; Step S2: constructing a range-Doppler (RD) model, and using the acquired raw data to calculate the initial pixel coordinates (x0, y0) of the corner reflector point according to the RD model; Step S3, based on the initial pixel coordinates (x0, y0) of the corner reflector point obtained in step S2, the pixel coordinates (x3, y3) of the corner reflector point in the SAR image are calculated.
2. The method for extracting one thousandth pixel of a corner reflector according to claim 1, characterized in that: The step S2 specifically includes the following sub-steps: Step S2.1: converting the geodetic coordinates of the corner reflector point into a geocentric rectangular coordinate system; Step S2.2: fitting the satellite orbit through the SAR satellite state vector to obtain the SAR satellite state vector at any time; Step S2.3: Calculate the azimuth imaging time t of the corner reflector point on the image, specifically according to the start time and time sampling interval and the row number of the corner reflector point on the SAR image, the formula is: t = t0 + t i ·i; Step S2.4: construct RD model; Step S2.5: Solve the RD model using the Newton iteration method to obtain the initial pixel coordinates (x0, y0) of the corner reflector point.
3. The method for extracting one thousandth pixel of a corner reflector according to claim 1, characterized in that: The conversion formula used for the conversion of geodetic coordinates to geocentric rectangular coordinates in step S2.1 is:
4. The method for extracting one thousandth pixel of a corner reflector according to claim 1, characterized in that: The satellite orbit fitting in step S2.2 adopts polynomial fitting. Specifically, the polynomial is:
5. The method for extracting one thousandth pixel of a corner reflector according to claim 1, characterized in that: The RD model in step S2.4 is as follows: Distance equation: Doppler equation: The equation of the Earth's ellipsoid:
6. The method for extracting one thousandth pixel of a corner reflector according to claim 1, characterized in that: Step S3 specifically includes the following sub-steps: Step S3.1, taking the initial pixel coordinates (x0, y0) of the corner reflector point obtained in step S2 as the center, select a complex matrix with a window size of N1×N1; Step S3.2, calculating the amplitude value of the N1×N1 complex matrix to obtain its amplitude value matrix; specifically, the amplitude value is calculated according to the following formula: Among them, Re(i,j) represents the real part of the pixel value (i,j), and Im(i,j) represents the imaginary part of the pixel value (i,j); Step S3.3, finding the pixel coordinates (x1, y1) corresponding to the maximum amplitude value sig(i, j) described in step S3.2; Step S3.4, taking the pixel coordinate (x1, y1) described in step S3.3 as the center, select a complex matrix with a window size of 2N2×2N2; Step S3.5, upsampling the complex matrix with a window size of 2N2×2N2 by a factor of M1 (M1<1024) to obtain an upsampled and reconstructed complex matrix; Step S3.6, calculating the amplitude value of the upsampled and reconstructed complex matrix to obtain its amplitude value matrix sig1(i,j); Step S3.7, finding the pixel coordinates corresponding to the maximum amplitude value in step S3.6 is (x2, y2); Step S3.8, upsampling the amplitude value of the row where the pixel coordinate x2 is located (i.e., the distance direction) by a factor of M2 to obtain a one-dimensional complex matrix Dx2, and upsampling the amplitude value of the column where the pixel coordinate y2 is located (i.e., the azimuth direction) by a factor of M2 to obtain a one-dimensional complex matrix Dy2; Step S3.9, calculate the amplitude value matrix of the one-dimensional complex matrix Dx2 as The amplitude value matrix of the one-dimensional complex matrix Dy2 is Step S3.10, calculate the amplitude value matrix and The decibel (db) values are and Specifically, the decibel value is calculated according to the following formula: Among them, max(*) represents the maximum amplitude value in the amplitude value matrix; Step S3.11, get the distance db value All db values from 3db on the left to 3db on the right form a new matrix Step S3.12, according to the distance to db value The pixel coordinates and db values are fitted with trigonometric functions; Step S3.13, find The coordinates corresponding to the peak point and the coordinates corresponding to the fitting value are determined, and the absolute value of the difference between the coordinate x3 corresponding to the peak point and the coordinate x'3 corresponding to the fitting value reaches the order of 1 / 1000, and the coordinate x3 corresponding to the distance to the peak point is output; Step S3.14, azimuth db value Repeat S3.11-S3.13 and output the coordinate y3 corresponding to the azimuth peak point.
7. The method for extracting one thousandth pixel of a corner reflector according to claim 1, characterized in that: The step S3.5 specifically includes the following sub-steps: Step S3.5.1, performing a two-dimensional fast Fourier transform on the complex matrix obtained in S3.4 to convert it into the frequency domain; Step S3.5.2, upsampling the complex matrix transferred to the frequency domain by a factor of M1 to obtain an upsampled complex matrix Step S3.5.3, reconstructing the upsampled complex matrix using the two-dimensional inverse Fourier transform The reconstructed complex matrix D2[M1×2N2,M1×2N2] is obtained.
8. The method for extracting one thousandth pixel of a corner reflector according to claim 1, characterized in that: The step S3.5.2 specifically includes the following sub-steps: Step S3.5.2.1, calculate the amplitude value of the complex matrix D2[M1×2N2,M1×2N2] obtained in step S3.5.1 And calculate the sum of the amplitude values of each column to get the matrix tmp row (i,1); Step S3.5.2.2, find step S3.5.2.1tmp row The row number i1 where the minimum value in (i,1) is located; Step S3.5.2.3, construct a two-dimensional zero matrix O1[(M1-1)×2N2,2N2]; Step S3.5.2.4, insert the two-dimensional zero matrix O1[(M1-1)×2N2,2N2] obtained in step S3.5.2.3 into The i1+1th row in Step S3.5.2.5, for step S3.5.2.4 matrix Calculate the sum of the amplitude values of each row to get the matrix tmp col (1,j); Step S3.5.2.6, find the matrix of step S3.5.2.5 The column number j1 where the minimum value is located; Step S3.5.2.7, construct a two-dimensional zero matrix O2[M1×2N2,(M1-1)×2N2]; Step S3.5.2.8, insert the two-dimensional zero matrix O2[M1×2N2,(M1-1)×2N2] obtained in step S3.5.2.7 into The j1+1th column in the matrix The processing of the above eight sub-steps completes the first M1-fold upsampling.
9. The method for extracting one thousandth pixel of a corner reflector according to claim 1, characterized in that: The step S3.8 specifically includes the following sub-steps: Step S3.8.1, determine the upsampling factor M2; specifically, determine M2 according to the following formula: M2=1024 / M1 Step S3.8.2, upsample the amplitude value of the row where the pixel coordinate x2 is located (i.e., the distance direction) by M2 times to obtain a one-dimensional complex matrix Dx2, and upsample the amplitude value of the column where the pixel coordinate y2 is located (i.e., the azimuth direction) by M2 times to obtain a one-dimensional complex matrix Dy2.
10. The method for extracting one thousandth pixel of a corner reflector according to claim 1, characterized in that: The step S3.8.2 specifically includes the following sub-steps: Step S3.8.2.1, construct a one-dimensional zero matrix and determine the size N3 of the one-dimensional zero matrix O; specifically, determine N3 according to formula (10): N3=(2N2×M1)×M2 =1024*2N2 Step S3.8.2.2, perform spectrum centering processing on the amplitude value of the row where the pixel coordinate x2 is located (i.e., the distance direction), then perform fast Fourier transform on the amplitude value after spectrum centering processing, and finally perform spectrum centering processing again; this processing is to make the low-frequency part of the spectrum located at the beginning of the array and the high-frequency part located at the end of the array; Step S3.8.2.3, assign the result obtained in step S3.8.2.2 to the zero matrix O3 Within the range, the assigned matrix Dx2 is obtained; Step S3.8.2.4, first perform spectrum centering on Dx2 assigned in step S3.8.2.3, then perform inverse fast Fourier transform, and finally perform spectrum centering again; this operation ensures that the format of the final result is consistent with the original signal; Step S3.8.2.5, the processing of the above four sub-steps completes the M2-fold upsampling of the amplitude value of the row where the pixel coordinate x2 is located (i.e., the distance direction); the amplitude value resampling in the azimuth direction repeats steps S3.8.2.1-step S3.8.2.5.
Citation Information
Patent Citations
Sub-pixel-level corner reflector automatic positioning method based on SAR rigorous imaging model
CN106526593A
High-precision spaceborne SAR geometric calibration method
CN107367716A
Corner reflector coordinate positioning method for shipborne SAR image
CN111896954A
Synthetic aperture radar imaging quality index evaluation system
CN115079109A
Interference SAR external calibration signal automatic extraction and application method
CN118091566A
