Azimuth ambiguity suppression method for stripe-mode spaceborne SAR (Synthetic Aperture Radar) image

Abstract

The invention discloses an azimuth ambiguity suppression method for a strip-mode spaceborne SAR image, and the method comprises the steps: 1, reading related parameters and echo data, removing the amplitude information of original echoes on the basis of the echo data, and only keeping the phase information of the original echoes, and 2, carrying out the precise distance compression of ambiguity, the method comprises the following steps of 1, detecting a fuzzy signal migration line in data obtained after the step 2 is completed, 4, carrying out fuzzy suppression in an original echo signal without amplitude information elimination based on the migration line screened out in the step 3, and 5, carrying out inverse operation of precise distance compression on the signal obtained in the step 4, and thus, original echoes are recovered for imaging. According to the method, the migration line characteristics of the fuzzy signals are utilized, fuzzy detection and suppression are carried out in the range Doppler domain, the azimuth fuzzy signals in the satellite-borne SAR image are finally eliminated, the calculation amount is reduced, the efficiency is improved, and meanwhile the original image resolution can be maintained to the maximum degree.

Description

Technical Field

[0001] The invention belongs to the field of image processing and relates to a method for suppressing azimuth ambiguity of space-borne SAR images, in particular to a high-efficiency method for suppressing azimuth ambiguity of space-borne SAR images in a stripe mode. Background Art

[0002] Synthetic aperture radar (SAR) satellites have developed rapidly in recent years. Since SAR satellites are not restricted by weather, geography, time and other factors, they can observe the earth all day long and have a certain degree of penetration. Therefore, they are widely used in military reconnaissance, terrain mapping, resource exploration, ocean observation, ecological monitoring, natural disaster monitoring, rapid rescue and other aspects.

[0003] With the development of spaceborne SAR, the quality requirements for spaceborne SAR images are becoming increasingly higher. However, due to the specific characteristics of the SAR system, special artifacts are prone to appear on SAR images, among which the azimuth ambiguity problem is particularly prominent. The limited pulse repetition frequency (PRF) and the non-ideal antenna pattern will cause azimuth ambiguity, causing the image to produce "ghosting", that is, strong targets outside the main beam of the antenna are superimposed on the imaging scene area because the signal sampling frequency is less than the Doppler frequency. Although most SAR systems are carefully designed to reduce azimuth ambiguity, it is difficult to completely eliminate it when facing targets with extremely high scattering intensity. In addition, azimuth ambiguity will also occur during the orbit debugging phase of spaceborne SAR due to the failure to optimize the system parameters, affecting the image quality. Therefore, it is extremely urgent to use signal processing to suppress azimuth ambiguity without changing the system design. For spaceborne SAR images, the commonly used azimuth ambiguity suppression methods can basically be divided into three categories:

[0004] 1. Azimuth ambiguity suppression method based on spectrum processing

[0005] The azimuth ambiguity suppression method based on spectrum processing is an effective method for SAR image azimuth ambiguity suppression. Its core idea is to suppress azimuth ambiguity by reducing azimuth spectrum aliasing. The commonly used method is to use frequency domain filtering in the azimuth frequency domain to eliminate azimuth ambiguity, but its ambiguity suppression effect is limited and it will also reduce the azimuth resolution of the image.

[0006] 2. Fuzzy echo reconstruction method based on fuzzy source

[0007] The core idea of ​​the fuzzy echo reconstruction method based on blur sources is to reconstruct the blurred echo based on the true target of the blur source and then subtract the reconstructed blurred signal from the original image to obtain a blur-free image. However, this method requires a precise understanding of the blur source, but in most cases, it is difficult to obtain an echo containing the blur source. Even if an echo containing the blur source is obtained, it cannot be used to accurately locate the specific location of the blur source.

[0008] 3. Blur suppression method based on fuzzy focusing

[0009] The core idea of ​​the blur suppression method based on blur focusing is to first focus the blurred area and detect and identify the position and order of the blurred signal in the focused image; then refocus the original image to the position of the blurred signal of the corresponding order and eliminate the signal; finally, perform the inverse operation (refocus back to the real image) to obtain the image with suppressed blur.

[0010] Existing SAR azimuth ambiguity suppression methods are mostly based on the idea of ​​selecting less blurred spectra for imaging or sparse reconstruction. However, due to the scalability of SAR antenna patterns, severely blurred spectra may exist within the wider frequency bands of the main region, making it difficult to find less blurred areas, resulting in limited applications. Even if aliased spectra in some blurred regions are suppressed through spectral filtering, they will still affect the main region spectrum, reducing bandwidth and azimuth resolution, ultimately affecting the image. Furthermore, sparse reconstruction methods are not only inefficient but also tend to zero out weak targets in the scene, significantly deviating from traditional SAR image products and potentially affecting some applications.

[0011] Meanwhile, the aforementioned azimuth ambiguity reduction methods, such as those based on spectral processing, ambiguity source-based echo reconstruction, and ambiguity focusing, while each with its own advantages, also suffer from numerous challenges, such as reduced azimuth resolution and the need for precise knowledge of the ambiguity source, which is difficult to obtain or estimate in advance. In short, existing azimuth ambiguity reduction methods all have limitations, and more comprehensive and effective solutions are urgently needed. Summary of the Invention

[0012] To address the above problems, the present invention proposes an azimuth ambiguity suppression method for strip-mode spaceborne SAR images. This method utilizes the migration line characteristics of the ambiguity signal to perform ambiguity detection and suppression in the range Doppler domain, ultimately eliminating the azimuth ambiguity signal in the spaceborne SAR image.

[0013] A method for suppressing azimuth ambiguity in stripe-mode spaceborne SAR images includes the following steps:

[0014] Step 1: Read in relevant parameters and echo data. Based on the echo data, remove the amplitude information of the original echo and only retain its phase information.

[0015] 1) Read in the relevant parameters for fuzzy suppression, including: scene center reference slant distance R ref , satellite equivalent speed V r , carrier frequency f0, wavelength λ, azimuth ambiguity order k, pulse repetition frequency f p , signal modulation frequency K r , range sampling rate f s , transmit pulse bandwidth B w , the angle θ between the beam and the normal of the satellite flight directionc , the relative speed V between the satellite and the ground target s 、Antenna length L a , the Doppler center frequency f of the main scene signal pointed by the beam center d0 .

[0016] 2) Read in the satellite-borne SAR image echo data and SAR image size; the SAR image size includes the total number of azimuth data points N a And the total number of sampling data points N in the distance direction r ; For the data point (i, j) in the spaceborne SAR image, i is the row number, corresponding to the azimuth, i = 0, 1, ..., N a -1; j is the number of columns, corresponding to the distance direction, j = 0, 1, ..., N r -1.

[0017] 3) Eliminate the amplitude information of the original echo and only retain its phase information.

[0018] Step 2: Perform accurate range compression on the fuzzy signal in the echo data.

[0019] (1) Perform a two-dimensional Fourier transform on the echo data without amplitude information to convert it into a two-dimensional frequency domain;

[0020] (2) Phase multiplication of echo data and reference function is performed in the two-dimensional frequency domain, and range compression and secondary range compression (SRC) of k-order fuzzy signals are simultaneously completed;

[0021] (3) The data is transformed into the range-Doppler domain through the range-inverse Fourier transform.

[0022] Step 3: Detect the fuzzy signal migration line in the data after completing step 2.

[0023] A. Perform distance interpolation on complex data;

[0024] B. Perform distance radiation correction to eliminate possible gain changes of the antenna in the distance direction;

[0025] C. Take the modulus and square (i.e., take the power) of the processed complex data to facilitate subsequent migration line detection;

[0026] D. Use sinc 2 The convolution kernel performs matched filtering on the blurred migration line to enhance the migration line of the blurred signal;

[0027] E. Binarize the matched filtered data to highlight the migration line;

[0028] F. Perform probabilistic Hough transform on the image after threshold detection, detect and identify the migration line in the parameter space, and thus obtain its position in the image.

[0029] Step 4: Based on the migration lines selected in step 3, fuzzy suppression is performed on the original echo data without removing the amplitude information.

[0030] Ⅰ. Perform the operation in step 2 on the original echo without removing the amplitude information;

[0031] II. In the original echo after completing the operation of sub-step I, the real part and the imaginary part of the complex data corresponding to the migration line area obtained in step 3 are set to zero, thereby removing the blurred lines.

[0032] Step 5: Perform an inverse operation of the precise range compression on the signal obtained in step 4 to restore the original echo for imaging.

[0033] ① Perform range-to-Fourier transform on the echo data obtained in step 4 to convert it into the two-dimensional frequency domain;

[0034] ② In the two-dimensional frequency domain, phase multiplication is performed with the new reference function to complete the inverse operation of precise distance compression, and the original echo data after fuzzy suppression processing in the two-dimensional frequency domain is obtained. Normal imaging can then be performed through the parameters of the main scene area.

[0035] The advantages of the present invention are:

[0036] (1) This method uses a unique approach to ambiguity detection and suppression in the range-Doppler domain. It first uses the Doppler parameters of the sidelobe azimuth ambiguity for precise range compression, then focuses the image after detecting and suppressing the ambiguity line and performing an inverse operation. This effectively avoids many of the problems of existing azimuth ambiguity suppression methods, such as reduced azimuth resolution and the need for precise knowledge of the ambiguity source, which is difficult to obtain or estimate in advance. It enhances adaptability to different scenarios and eliminates the need for manual parameter adjustment.

[0037] (2) This method can reduce the amount of calculation and improve efficiency while maintaining the original image resolution to the greatest extent without causing a major impact, ensuring that the image quality is not damaged, and taking into account both processing efficiency and imaging effects. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 is a flow chart of the method of the present invention;

[0039] Figure 2 This is a flow chart of the present invention for detecting and suppressing the migratory line of an azimuthally ambiguous signal;

[0040] Figure 3 Schematic diagram of the convolution kernel of the present invention;

[0041] Figure 4is the antenna pattern simulated in an embodiment of the present invention;

[0042] Figure 5 It is an image directly imaged from the simulation data of the present invention;

[0043] Figure 6 This is the image obtained after the azimuth blur suppression of the present invention. DETAILED DESCRIPTION

[0044] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0045] The present invention is a method for suppressing azimuth ambiguity of spaceborne SAR based on range Doppler domain. Figure 1 As shown, the following steps are included:

[0046] Step 1: Read in relevant parameters and echo data. Based on the echo data, remove the amplitude information of the original echo and only retain its phase information.

[0047] 1) Read in the relevant parameters for blur suppression, including: scene center reference slant distance R ref , satellite equivalent speed V r , carrier frequency f0, wavelength λ, azimuth ambiguity order k, pulse repetition frequency f p , signal modulation frequency K r , range sampling rate f s , transmit pulse bandwidth B w , the angle θ between the beam and the normal of the satellite flight direction c , the relative speed V between the satellite and the ground target s 、Antenna length L a , the Doppler center frequency f of the main scene signal pointed by the beam center d0 .

[0048] 2) Read in the satellite-borne SAR image echo data and SAR image size; the SAR image size includes the total number of azimuth data points N a And the total number of sampling data points N in the distance direction r ; For the data point (i, j) in the spaceborne SAR image, i is the row number, corresponding to the azimuth, i = 0, 1, ..., N a -1; j is the number of columns, corresponding to the distance direction, j = 0, 1, ..., N r -1.

[0049] 3) Eliminate the amplitude information of the original echo and retain only its phase information. The specific method is as follows:

[0050] The expression of the read-in original echo data is as follows:

[0051] Sr =S r .re+j·S r .im

[0052] Among them, S r Represents echo data, S r .re and S r .im represents the real part and imaginary part of the echo data respectively. Substituting the read original echo data (i, j) into the following formula in sequence can obtain the pure phase echo without the amplitude information of the original echo:

[0053]

[0054] Step 2: Perform accurate range compression on the k-order azimuth ambiguity in the echo data.

[0055] (1) Perform a two-dimensional Fourier transform on the echo data after removing the amplitude information, and convert it into a two-dimensional frequency domain.

[0056] First, perform Fourier transform in azimuth direction, and for the echo data (i, j) without amplitude information, traverse in range direction (column), take one column of data each time, and perform N a_fft The Fast Fourier Transform (FFT) of the points is then rearranged so that the zero-frequency component (DC component) is located at the center of the spectrum, which is a translation operation. After traversing and processing all columns, they are reassembled into a two-dimensional array to obtain the result of the azimuthal Fourier transform. a_fft is not less than N a The smallest power of 2.

[0057] Secondly, perform distance Fourier transform, and traverse the data of the azimuth Fourier transform in the azimuth direction (row), taking one row of data each time and performing N r_fft The Fast Fourier Transform (FFT) of the point is then rearranged so that the zero-frequency component (DC component) is located at the center of the spectrum, which is a translation operation. After traversing and processing all rows, they are reassembled into a two-dimensional array to obtain the result of the distance Fourier transform. r_fft is not less than N r The smallest power of 2.

[0058] (2) Phase multiplication of the echo data and the reference function is performed in the two-dimensional frequency domain, and range compression and secondary range compression (SRC) of the k-order ambiguous signal are simultaneously performed;

[0059] The reference function expression is as follows:

[0060]

[0061] Where Km (f η ,R ref ) is the range modulation frequency corrected in the range Doppler domain, expressed as follows:

[0062]

[0063] Where c is the speed of light, which is 3.0e8 (m / s), D (f η ,V r ) is the migration factor, which is expressed as follows:

[0064]

[0065] For the above expression, f τ and f η Represent the range frequency and azimuth frequency respectively. When the zero frequency component is located at the center of the spectrum after FFT, the calculation method is as follows:

[0066]

[0067]

[0068] Where i r and i a They refer to the distance point position and azimuth point position in the two-dimensional frequency domain respectively.

[0069] The echo data of step (1) is stored in a two-dimensional array, and each data position (i a ,i r ), calculate the reference function corresponding to that position using the aforementioned reference function expression, and then perform a complex multiplication on the reference function by the data at the corresponding position. After traversing and processing all positions, reassemble them into a two-dimensional array to obtain the result of phase multiplication.

[0070] (3) The data is transformed into the range-Doppler domain through the range-inverse Fourier transform.

[0071] For the echo data that has completed sub-step (2), traverse it in azimuth (rows), take one row of data each time, first perform a translation operation so that the zero frequency component (DC component) is located at the starting point of the spectrum again, and then perform N r_fft After traversing and processing all rows, they are reassembled into a two-dimensional array to obtain the result of the inverse Fourier transform of the distance.

[0072] Step 3: Detect the fuzzy signal migration line in the echo data after completing step 2, such as Figure 2 shown.

[0073] A. Perform range interpolation on the echo data (complex data) after the operation in step 2;

[0074] Traverse in direction (row), take one row of data each time, and perform Factor on the row of data r The interpolation method may include but is not limited to linear interpolation, polynomial interpolation or spline interpolation. After traversing and processing all rows, recombining them into a two-dimensional array, the result of distance interpolation can be obtained. r It is the distance interpolation multiplier.

[0075] B. Perform distance radiation correction to eliminate possible gain changes of the antenna in the distance direction;

[0076] After completing sub-step A, the echo data is traversed in the distance direction (column), and a column of complex data is taken each time, and the modulus value of the column of complex data is accumulated. After traversing and processing all columns, the obtained modulus value data is stored in the distance direction (column) as a one-dimensional modulus value array Amplitude[·], with an array size of N r_fft The modulus value is calculated as follows:

[0077]

[0078] Wherein, Re and Im represent the real part and imaginary part of the complex data, respectively, and A is the modulus value of the complex data.

[0079] Find the mean of the array Amplitude[·], calculated as follows:

[0080]

[0081] Among them, Mean is the mean of the modulus value array.

[0082] After obtaining the mean of the modulus array, you can perform distance radiation correction. The correction method is as follows:

[0083]

[0084] Among them, S r_correction [·][·] are the echo data after correction, S r [·][·] is the original data after completing sub-step A, i r and i a They are the current distance point position and azimuth point position respectively.

[0085] C. Take the modulus and square (i.e., take the power) of the processed complex data to facilitate subsequent migration line detection;

[0086] D. Use sinc 2The convolution kernel performs matched filtering on the blurred migration line to enhance the migration line of the blurred signal;

[0087] sinc 2 Convolution kernel Figure 3 As shown, the distance profile of the convolution kernel is sinc 2 function, and sinc 2 The main lobe width of the function is set to f s / B w , sinc at different positions in the convolution kernel 2 The vertical offset angle of the peak position line of the function is θ. 2 The position of the function peak and the size of the convolution kernel need to be designed. The specific design method is as follows:

[0088] First, the size of the convolution kernel must be determined. The appropriate convolution kernel size should be equivalent to the length of the migration line. For strip-mode spaceborne SAR, the formula for calculating the migration line length is as follows:

[0089]

[0090] Among them, L RML is the migration line length, and the convolution kernel size M is generally set to be slightly smaller than L RML and set it to an odd number.

[0091] Secondly, determine the sinc in the convolution kernel 2 The position of the function peak, that is, the offset angle θ, for the migration line of the k-order ambiguity signal, its range migration in the range Doppler domain can be approximated as:

[0092]

[0093] Among them, the distance migration RM(i a ) and the azimuth point position i a related.

[0094] Since range migration is usually a curve, the slope k of the migration line at each azimuth position can be calculated slope , set the positive direction of the azimuth to the x-axis and the positive direction of the distance to the y-axis. The specific solution is as follows:

[0095] a) Traverse i a The value of RM(i a );

[0096] b) The slope k of two adjacent azimuth positions slope The calculation formula is as follows:

[0097] k slope=RM(n+1)-RM(n),n=0,1,2,...,N a_fft -2

[0098] c) Solve in sequence to obtain the maximum value k of the slope slope_max and minimum value k slope_min , then relative to Figure 3 The maximum and minimum values ​​of the angle (offset angle relative to the vertical (azimuth) direction) of the black dotted line in the figure are θ max and θ max , the solution is as follows:

[0099] θ max =arctan(k slope_max )

[0100] θ min =arctan(k slope_min )

[0101] Then the offset angle θ can be taken as the mean of the two angles obtained in sub-step c), that is,

[0102] In determining the convolution kernel, sinc 2 After determining the position of the function peak and the size of the convolution kernel M, the convolution kernel can be obtained. The calculation formula for each element Conv[i][j] in its matrix is ​​as follows:

[0103]

[0104] Among them, D(i,j) represents the sinc value of the element in row i and column j relative to this row. 2 The position of the function peak is calculated as follows:

[0105]

[0106] in, Represents the sinc of the i-th row in the convolution kernel 2 The location of the function peak.

[0107] Finally, the obtained convolution kernel can be used to perform a convolution operation on the echo data after completing the operation in sub-step (c) to perform matched filtering on it.

[0108] E) Binarize the matched filtered data to highlight the migration lines.

[0109] The image is binarized by setting the threshold T. The empirical setting method of the threshold T is as follows:

[0110] T=μ+3δ

[0111] Wherein, μ represents the mean value of the power data after completing the operation of sub-step D), and δ represents the standard deviation of the power data after completing the operation of sub-step D).

[0112] After the threshold T is obtained, the power data after completing the operation of sub-step D) can be binarized using the threshold T, with the value of 255 being used for data greater than the threshold T and 0 being used for data less than the threshold T.

[0113] F) Performing a probabilistic Hough transform on the image after threshold detection, detecting and identifying the migration lines in the parameter space, and thus obtaining their positions in the image.

[0114] The specific operation can be done using the cv2.HoughLinesP function in OpenCV. The return value of this function is a list containing the coordinates of the start and end points of the line. In this case, the migration line can be approximated as a straight line. Therefore, the position of the migration line can be obtained based on the returned start and end point coordinates. The function prototype is as follows:

[0115] HoughLinesP(image,rho,theta,threshold,minLineLength=None,maxLineGap=None)

[0116] Wherein, image represents the input binary image, which in this patent is the data after completing the operation of sub-step E); rho is the pixel resolution, and its empirical value is 1; theta is the angular resolution, and its empirical value is 0.1;

[0117] threshold is the minimum number of votes, minLineLength is the minimum line length, and maxLineGap is the maximum line gap. The minimum line length minLineLength and the minimum number of votes threshold need to be adjusted according to the minimum length of the migratory line in the image. The setting of the maximum line gap maxLineGap is related to the sidelobe spacing of the antenna radiation pattern. The accuracy of migratory line detection can be improved by continuously adjusting the parameters.

[0118] Step 4: Based on the migration line identified in step 3, fuzzy suppression is performed on the original echo data without removing the amplitude information, such as Figure 2 As shown, the specific steps are:

[0119] Ⅰ. Perform the operation in step 2 on the original echo without removing the amplitude information, and perform the same distance interpolation process as in step 3 sub-step A); the specific operation is: perform the operation in step 2 on the original echo without removing the amplitude information according to the same parameters, and perform Factor r Multiple distance interpolation processing.

[0120] II. In the original echo after completing sub-step I, set both the real and imaginary parts of the complex data corresponding to the migration line region obtained in step 3 to zero, thereby removing the blurred lines in the original echo data without removing the amplitude information. This operation can be performed using the cv2.Line function in OpenCV. This function is used to draw a straight line on an image using the starting and ending coordinates. Each call to this function directly modifies the original image, setting the real and imaginary parts of the data to zero and obtaining the data with the blurred lines removed. The function prototype is as follows:

[0121] cv2.line(img,pt1,pt2,color,thickness=None,lineType=None,shift=None)

[0122] Among them, img is the image to be drawn, which in this patent is the echo data after completing the operation of sub-step I of step 4, including the real part and the imaginary part; pt1 and pt2 are the starting point and end point of the straight line respectively, which in this patent are the endpoints of the migration line obtained in step 3, and the color value is 0.

[0123] For line thickness, line type, and decimal point precision shift, select the function's default values.

[0124] Step 5: Perform an inverse operation of the precise range compression on the echo data obtained in step 4 after removing the blurred lines, thereby restoring the original echo for imaging.

[0125] ① Perform range-direction Fourier transform on the echo data obtained in step 4 after removing the blurred lines, and transform it into the two-dimensional frequency domain;

[0126] Traverse in the direction (row), take one row of data each time, and perform N r_fft The Fast Fourier Transform (FFT) of the points is then performed, and the FFT output is rearranged so that the zero-frequency component (DC component) is located at the center of the spectrum. This is a translation operation. After traversing and processing all rows, they are reassembled into a two-dimensional array to obtain the result of the range Fourier transform.

[0127] ② In the two-dimensional frequency domain, the transformed echo data is phase-multiplied with the new reference function to complete the inverse operation of precise distance compression, and the original echo data after fuzzy suppression processing in the two-dimensional frequency domain is obtained. Normal imaging can then be performed through the parameters of the main scene area.

[0128] The new reference function expression is as follows:

[0129]

[0130] Its calculation method and phase multiplication method are consistent with sub-step (2) of step 2.

[0131] In order to illustrate the effectiveness of the method of the present invention, the following simulation test is carried out. The simulation parameters are shown in Table 1. The number of simulated points is a single point target. The simulated antenna pattern is as follows: Figure 4 As shown in Figure 1, this antenna pattern has high-gain sidelobe azimuth angles, so there is azimuth ambiguity. The images before and after azimuth ambiguity suppression are shown in Figure 1. Figure 5 、 Figure 6 The red lines represent the parts before and after suppression. The azimuth resolution, peak sidelobe ratio, and azimuth ambiguity ratio before and after correction are shown in Table 2.

[0132] Table 1 Simulation parameters

[0133] Simulation parameters Selected parameter value Number of distance sampling points 4096 Azimuth simulation pulse number 16384 Signal sampling rate (Hz) 333000000.0 Signal bandwidth (Hz) 300000000.0 Pulse duration (microseconds) 10.36 Reference slant distance (km) 555.0 Carrier frequency (Hz) 13500000000.0 Doppler center frequency (Hz) 0.0 Satellite equivalent speed (m / s) 7080.0 Main lobe antenna length (m) 4.236 High gain sidelobe azimuth (degrees) 0.95

[0134] Table 2 Analysis of azimuth resolution, peak sidelobe ratio and azimuth ambiguity ratio before and after correction

[0135]

[0136] from Figure 5 、 Figure 6 As can be seen, before correction, the portion selected by the red box has high energy, indicating a strong azimuthally blurred signal. After correction, the energy of the portion selected by the red box is significantly weakened, effectively suppressing azimuth blur. Table 2 shows the azimuth blur ratio, azimuth resolution, and peak sidelobe ratio before and after correction. It can be seen that the azimuth blur ratio changes significantly, while the azimuth resolution and peak sidelobe ratio do not change significantly. These results demonstrate that azimuth blur suppression is achieved effectively while maintaining the resolution of the original image.

Claims

1. A method for azimuth ambiguity suppression of stripe-mode spaceborne SAR images, characterized by: The steps are: Step 1: Read in relevant parameters and echo data. Based on the echo data, remove the amplitude information of the original echo and only retain its phase information. 1) Read in the relevant parameters for fuzzy suppression, including: scene center reference slant distance R ref , satellite equivalent speed V r , carrier frequency f0, wavelength λ, azimuth ambiguity order k, pulse repetition frequency f p , signal modulation frequency K r , range sampling rate f s , transmit pulse bandwidth B w , the angle θ between the beam and the normal of the satellite flight direction c , the relative speed V between the satellite and the ground target s 、Antenna length L a , the Doppler center frequency f of the main scene signal pointed by the beam center d0 ; 2) Read in the satellite-borne SAR image echo data and SAR image size; the SAR image size includes the total number of azimuth data points N a And the total number of sampling data points N in the distance direction r ; For the data point (i, j) in the spaceborne SAR image, i is the row number, corresponding to the azimuth direction, i = 0, 1, ..., N a -1; j is the number of columns, corresponding to the distance direction, j = 0, 1, ..., N r -1; 3) Eliminate the amplitude information of the original echo and retain only its phase information; Step 2: Perform accurate range compression on the fuzzy signal in the echo data; (1) Perform a two-dimensional Fourier transform on the echo data without amplitude information to convert it into a two-dimensional frequency domain; (2) Phase multiplication of echo data and reference function expression is performed in the two-dimensional frequency domain, and range compression and secondary range compression of k-order fuzzy signals are simultaneously completed; (3) Transform the data into the range-Doppler domain through the range-inverse Fourier transform; Step 3: Detect the fuzzy signal migration line in the data after completing step 2; A. Perform distance interpolation on complex data; B. Perform distance radiation correction to eliminate possible gain changes of the antenna in the distance direction; C. Modulo and square the processed complex data to facilitate subsequent migration line detection; D. Use sinc 2 The convolution kernel performs matched filtering on the blurred migration line to enhance the migration line of the blurred signal; E. Binarize the matched filtered data to highlight the migration line; F. Performing a probabilistic Hough transform on the threshold-detected image to detect and identify the migration lines in the parameter space, thereby obtaining their positions in the image. Step 4: Based on the migration lines selected in step 3, fuzzy suppression is performed on the original echo data without removing the amplitude information; I. Perform the operation in step 2 on the original echo without removing the amplitude information, and perform the same range interpolation process as in sub-step A of step 3; II. In the original echo after completing operation A, the real and imaginary parts of the complex data corresponding to the migration line area obtained in step 3 are set to zero to remove the fuzzy lines; Step 5: Perform an inverse operation of the precise range compression on the signal obtained in step 4 to restore the original echo for imaging; ① Perform range-to-Fourier transform on the echo data obtained in step 4 to convert it into the two-dimensional frequency domain; ② In the two-dimensional frequency domain, phase multiplication is performed with the new reference function to complete the inverse operation of precise distance compression, and the original echo data after fuzzy suppression processing in the two-dimensional frequency domain is obtained. Normal imaging can then be performed through the parameters of the main scene area.

2. The method for suppressing azimuth ambiguity in stripe-mode spaceborne SAR images according to claim 1, characterized in that: The method for removing the amplitude information of the original echo in sub-step 3) of step 1 is: The expression of the read-in original echo data is: S r =S r .re+j·S r .im Among them, S r Represents echo data, S r .re and S r .im represents the real part and imaginary part of the echo data respectively. Substituting the read original echo data (i, j) into the following formula in sequence can obtain the pure phase echo without the amplitude information of the original echo:

3. The method for suppressing azimuth ambiguity in stripe-mode spaceborne SAR images according to claim 1, characterized in that: The two-dimensional Fourier transform method of the echo data with amplitude information removed in sub-step (1) of step 2 is: First, perform Fourier transform in azimuth direction, and for the echo data (i, j) without amplitude information, traverse in range direction, taking one column of data each time and performing N a_fft The fast Fourier transform of the point is then rearranged so that the zero-frequency component is located at the center of the spectrum, which is a translation operation. After traversing and processing all columns, they are reassembled into a two-dimensional array to obtain the result of the azimuthal Fourier transform; where N a_fft is not less than N a The smallest power of 2; Secondly, perform Fourier transform in distance direction, and traverse the data of the completed Fourier transform in azimuth direction by azimuth direction, taking one row of data each time and performing N r_fft The fast Fourier transform of the point is performed, and the output of the fast Fourier transform is rearranged so that the zero-frequency component is located at the center of the spectrum; after traversing and processing all rows, it is recombined into a two-dimensional array to obtain the result of the distance Fourier transform; where N r_fft is not less than N r The smallest power of 2.

4. The method for suppressing azimuth ambiguity in stripe-mode spaceborne SAR images according to claim 1, wherein: The reference function in sub-step (2) of step 2 is designed as: Where K m (f η ,R ref ) is the range modulation frequency corrected in the range Doppler domain, expressed as: Where c is the speed of light, which is 3.0e8 (m / s), D (f η ,V r ) is the migration factor, expressed as: For the above expression, f τ and f η Represent the range frequency and azimuth frequency respectively. When the zero frequency component is located at the center of the spectrum after FFT, the calculation method is as follows: Where i r and i a Respectively refer to the distance point position and azimuth point position in the two-dimensional frequency domain; N a_fft is not less than N a The smallest power of 2; The echo data of step (1) is stored in a two-dimensional array, and each data position (i a ,i r ), solve for each data position (i a ,i r ) corresponding to the reference function; further multiply the solved reference function by the data at the corresponding position; after traversing and processing all positions, recombining them into a two-dimensional array to obtain the result of phase multiplication.

5. The method for suppressing azimuth ambiguity in stripe-mode spaceborne SAR images according to claim 1, wherein: The range radiation correction method of sub-step (B) in step 3 is: After completing sub-step A, the echo data is traversed in the distance direction, and a column of complex data is taken each time, and the modulus values ​​of the column of complex data are accumulated. After traversing and processing all columns, the obtained modulus value data is stored in the one-dimensional modulus value array Amplitude[·] in the distance direction, and the array size is N r_fft ; The modulus value is calculated as follows: Where Re and Im represent the real and imaginary parts of the complex data, respectively, and A is the modulus of the complex data; Find the mean of the array Amplitude[·], calculated as follows: Among them, Mean is the mean of the modulus array; After obtaining the mean of the modulus array, you can perform distance radiation correction. The correction method is as follows: Among them, S r_correction [·][·] are the echo data after correction, S r [·][·] is the original data after completing sub-step A, i r and i a They are the current distance point position and azimuth point position respectively.

6. The method for suppressing azimuth ambiguity in stripe-mode spaceborne SAR images according to claim 1, characterized in that: In step 3, in sub-step (D), the convolution kernel sinc 2 The position of the function peak and the size of the convolution kernel are designed as follows: First, we need to determine the size of the convolution kernel, which is equivalent to the length of the migration line. The length of the migration line is: Among them, L RML is the migration line length, and the convolution kernel size M is generally set to be slightly smaller than L RML The value of , and set to an odd number; Secondly, determine the sinc in the convolution kernel 2 The position of the function peak, that is, the offset angle θ, and its range migration in the range Doppler domain are approximately: Among them, the distance migration RM(i a ) and the azimuth point position i a related; Calculate the slope k of each azimuth position of the migration line slope , setting the positive direction of the azimuth to the x-axis and the positive direction of the distance to the y-axis, the solution is: a) Traverse i a The value of RM(i a ); b) The slope k of two adjacent azimuth positions slope The calculation formula is as follows: k slope =RM(n+1)-RM(n),n=0,1,2,…,N a_fft -2 c) Solve in sequence to obtain the maximum value k of the slope slope_max and minimum value k slope_min , and the maximum and minimum values ​​of the offset angle relative to the vertical direction are θ max and θ max , the solution is as follows: θ max =arctan(k slope_max ) θ min =arctan(k slope_min ) Then the offset angle θ can be taken as the mean of the two angles obtained in sub-step c), that is, The sinc in the convolution kernel determined above 2 After calculating the position of the function peak and the size of the convolution kernel M, the convolution kernel is obtained. The calculation formula for each element Conv[i][j] in its matrix is: Among them, D(i,j) represents the sinc value of the element in row i and column j relative to this row. 2 The position of the function peak is calculated as follows: in, Represents the sinc of the i-th row in the convolution kernel 2 The location of the function peak.

7. The method for suppressing azimuth ambiguity in strip-mode spaceborne SAR images according to claim 1, characterized in that: Step 3, sub-step E) Binarization specific method is: The image is binarized by setting the threshold T. The empirical setting method of the threshold T is as follows: T=μ+3δ Wherein, μ represents the mean value of the power data after completing the operation of sub-step D), and δ represents the standard deviation of the power data after completing the operation of sub-step D); After the threshold T is obtained, the power data after the operation in sub-step D) can be binarized using the threshold T, with the value of 255 being used for data greater than the threshold T and 0 being used for data less than the threshold T.

8. The method for suppressing azimuth ambiguity in stripe-mode spaceborne SAR images according to claim 1, wherein: In step 3, sub-step F): use the cv2.HoughLinesP function in OpenCV; the return value of the cv2.HoughLinesP function is a list containing the coordinates of the start and end points of the line. The migration line is approximated as a straight line, and the position of the migration line can be obtained based on the returned start and end point coordinates.

9. The method for suppressing azimuth ambiguity in stripe-mode spaceborne SAR images according to claim 1, characterized in that: Step 4, sub-step II is implemented using the cv2.Line function in OpenCV; the cv2.Line function is used to draw a straight line on the image with the starting and ending coordinates. At the same time, each call to the function will directly modify the original image, complete the operation of setting the real and imaginary parts of the data to 0, and obtain the data without the blurred line.

10. The method for suppressing azimuth ambiguity of stripe-mode spaceborne SAR images according to claim 1, characterized in that: In step 5, sub-step ②, the new reference function expression is: Where K m (f η ,R ref ) is the range modulation frequency corrected in the range Doppler domain, expressed as: Where c is the speed of light, which is 3.0e8 (m / s), D (f η ,V r ) is the migration factor, expressed as: For the above expression, f τ and f η Represent the range frequency and azimuth frequency respectively. When the zero frequency component is located at the center of the spectrum after FFT, the calculation method is as follows: Where i r and i a Respectively refer to the distance point position and azimuth point position in the two-dimensional frequency domain; N a_fft is not less than N a The smallest power of 2.

Images