CN114780911A - Ocean wide swath distance ambiguity solving method based on deep learning - Google Patents

Abstract

The invention discloses a deep learning-based method for de-ambiguating the distance of a wide oceanic swath. First, an echo signal of a spaceborne SAR system is acquired, a reference signal vector is constructed according to radar parameters, and the echo signal points are multiplied by the conjugate of the reference signal vector. , get the data matrix after distance pulse pressure; then perform inverse Fourier transform and Fourier transform; then construct a Doppler compensation function, and perform dot product with the Fourier transformed matrix; finally construct a deep neural network , the obtained vector is used as input and label, and the distance deblurring of ocean wide swath is realized after training. The invention can effectively solve the problem of deblurring the distance of the swath of the spaceborne SAR system, and obtain the information of the wide swath of the ocean without range ambiguity, so as to achieve the purpose of SAR imaging of moving targets on the sea surface without range ambiguity.

Description

A deep learning-based method for distance deblurring of ocean wide swaths

technical field

The invention belongs to the technical field of radar signal processing, and in particular relates to a distance deblurring method for an ocean wide swath.

Background technique

For the spaceborne SAR system, when high-resolution imaging of a scene is required, it can be achieved by increasing the revisit period of the satellite. However, the increase in satellite revisit cycles comes with significant costs. In order to reduce the revisit period of the satellite, the spaceborne SAR system can be used to transmit a high pulse repetition frequency (PRF) chirp signal. Since the system transmits a high PRF signal, more echo information of the scene target can be obtained, thereby realizing high-resolution imaging of the spaceborne SAR radar.

However, transmitting a high PRF signal will also bring about the problem of distance ambiguity. As shown in Figure 1, when the echo delay difference of different imaging scenes in the swath is equal to an integer multiple of the pulse repetition period (PRT), the received wide The swath echoes produce distance blur.

For wide swath solution range ambiguity, the most commonly used methods are look-up table method and transmitting orthogonal coded signal method. For the method of transmitting quadrature coded signals, there are commonly used quadrature phase numbered signals and quadrature frequency coded signals. It is difficult to achieve the resolution requirement by using the quadrature phase coded signal to deblur the accuracy, while the quadrature frequency coded signal can meet the resolution requirement, but the quadrature coded signal is more sensitive to Doppler. However, these methods all need to meet the condition of high echo signal-to-noise ratio.

In practical situations, the signal-to-noise ratio of echoes for marine targets is relatively low, generally 0dB. At the same time, most spaceborne SAR systems acquire wide swath information by transmitting high PRF signals, so they face the problem of distance ambiguity of echoes with low SNR in wide swaths.

SUMMARY OF THE INVENTION

In order to overcome the deficiencies of the prior art, the present invention provides a deep learning-based ocean wide swath range de-ambiguity method. First, the echo signal of the spaceborne SAR system is obtained, a reference signal vector is constructed according to the radar parameters, and the echo signal is Dot multiply the conjugate of the reference signal vector to obtain the data matrix after the distance pulse pressure; then perform inverse Fourier transform and Fourier transform; then construct a Doppler compensation function, and perform point with the Fourier transformed matrix Finally, a deep neural network is constructed, and the obtained vector is used as the input and label, and after training, the distance deblurring of the ocean wide swath is realized. The invention can effectively solve the problem of deblurring the distance of the swath of the spaceborne SAR system, and obtain the information of the wide swath of the ocean without range ambiguity, so as to achieve the purpose of SAR imaging of moving targets on the sea surface without range ambiguity.

The technical scheme adopted by the present invention to solve the technical problem comprises the following steps:

Step 1: Obtain the echo signal of the spaceborne SAR system, which is an n _r ×n _a -dimensional matrix, perform Fourier transform processing on the echo signal matrix by column, and save the result in the matrix s(n1 , n2); _{n r} represents the number of points in the distance direction, and na represents the number of points in the azimuth direction;

s_r(n1)为n_r×1向量；其中，k_r表示调频率，k_r＝B/T_p，B表示发射信号带宽，T_p表示发射脉冲宽度，f_r表示为距离向频域坐标，

Δf为距离频域间隔，

n1＝0,1,...,n_r-1；Step 2: According to the radar parameters, construct the reference signal vector

s_r(n1) is an n _r ×1 vector; among them, k _r represents the frequency modulation, k _r =B/T _p , B represents the transmission signal bandwidth, T _p represents the transmission pulse width, and _fr represents the range-to-frequency domain coordinates,

Δf is the distance frequency domain interval,

n1=0,1,..., _nr -1;

L表示为合成孔径长度，n2＝0,1,...,n_a-1；Step 3: Multiply each column of s(n1, n2) with the conjugate of the reference signal vector s_r(n1) to obtain the data matrix s(f _r , x _n2 ) after the distance pulse pressure; wherein, x _n2 represents Azimuth time domain coordinates,

L represents the synthetic aperture length, n2=0,1,...,n _a -1;

Step 4: Perform inverse Fourier transform processing on the matrix s(f _r , x _n2 ) by column, and store the result in the matrix s(n _r , n _a );

PRF为方位采用频率，Δf_a为方位频域间隔，

Step 5: Fourier transform is performed on the matrix s(n _r ,n _a ) by row, and the result is stored in the s(n _r ,f _a ) matrix; among them, f _a represents the azimuth frequency domain coordinate,

PRF is the azimuth adopted frequency, Δf _a is the azimuth frequency domain interval,

s_h(n1)为n_r×1的向量；其中，n_r表示距离向点数，v为动目标径向速度，V为雷达平台运动速度，θ为斜距平面雷达视线的方位角，λ为雷达的工作波长，g表示重力加速度；Step 6: Construct the Doppler compensation function according to the radar parameters

s_h(n1) is a vector of n _r × 1; among them, n _r represents the number of points in the range direction, v is the radial velocity of the moving target, V is the moving speed of the radar platform, θ is the azimuth angle of the slant range plane radar line of sight, and λ is the radar The working wavelength of , g is the acceleration of gravity;

Step 7: Do point multiplication with the conjugate of s(n _r , f _a ) and the Doppler compensation function s_h(n1) vector, and perform inverse FFT processing on the multiplication result row by row to obtain the post-pulse Doppler compensation. The data matrix ss(n _r ,n _a );

Step 8: According to the spatial filtering theory, construct a matrix Ω composed of target scattering coefficients, where Ω is an n _r ×n _a -dimensional matrix;

Step 9: Divide the matrix ss(n _r , na ) into two channel data ss_real( _{n r} , na ) and _{ss_imag} ( _{n r} , na ) according to the real part and the imaginary part; The imaginary part is the two channel data Ω_real(n _r ,n _a ) and Ω_i mag(n _r ,n _a );

Step 10: Construct a deep learning network based on two-channel input and two-channel output of a convolutional neural network; use the matrix ss_real(n _r ,n _a ) and the matrix ss_i mag(n _r ,n _a ) as the input data of the deep learning network, and set the The matrix Ω_real(n _r ,n _a ) and the matrix Ω_i mag(n _r ,n _a ) are used as training label data to train the deep learning network;

Step 11: Input the data to be deblurred into the deep learning network trained in step 10, and obtain the data matrix y_real(n _r ,n _a ) and matrix y_i mag(n _r ,n _a ) after distance deblurring;

Step 12: Combine the matrix y_real(n _r ,n _a ) and the matrix y_i mag(n _r ,n _a ) into the matrix y(n _r ,n _a ) according to the relationship between the real and imaginary parts;

Step 13: Perform Doppler parameter estimation on the matrix y( _{n r} , na ) to obtain the estimated Doppler parameter

Step 14: Perform range migration correction on the matrix y( _{n r} , na ) according to the Doppler parameter f _dc , and store the result in the matrix y′( _{n r} , na ).

s_a(n2)为1×n_a向量；其中，f_c表示雷达发射信号的载频，c为电磁波的传播速度，V为卫星速度，θ为卫星的斜视角，t_n2为方位慢时间；Step 15: According to the radar parameters, construct the reference signal vector

s_a(n2) is _a 1×na vector; among them, f _c represents the carrier frequency of the radar transmission signal, c is the propagation speed of the electromagnetic wave, V is the satellite speed, θ is the oblique angle of the satellite, and t _n2 is the azimuth slow time;

Step 16: Perform FFT processing on the matrix y'(n _r , n _a ) by row to obtain y(n _r , f _a ); take out each row of y(n _r , f _a ), and multiply the average point by the reference signal vector s_a( The conjugate of n2), the data matrix y(n _r , f _a ) after distance pulse pressure is obtained;

Step 17: Perform inverse FFT processing on the matrix y(n _r , f _a ) by row to obtain the SAR focusing result matrix z( _{n r} , na ).

Further, the specific steps of step 8 are as follows:

The spaceborne SAR system transmits encoded signals of multiple code patterns, and the echo signal received by the multi-antenna digital receiver is expressed as A=[s(1),...,s(m),...,s(M)], m=1,2,...,M, where M is the number of swaths, and s(m) represents the echo signal of the mth swath received by the multi-antenna digital receiver; The scattering coefficient of the target is recovered from the fuzzy echo signal, that is, the corresponding scattering coefficient is 1 when there is a target, and 0 when there is no target; the weight vector is Ω=[Ω ₁ ,…,Ω _{m , .} ^_ _{_ _ _} (A) Y _m , the whole defuzzification process is completed by solving Ω.

The beneficial effects of the present invention are as follows:

The invention can effectively solve the problem of deblurring the distance of the swath of the spaceborne SAR system, and obtain the information of the wide swath of the ocean without range ambiguity, so as to achieve the purpose of SAR imaging of moving targets on the sea surface without range ambiguity.

Description of drawings

FIG. 1 is a schematic diagram of the distance blurring of a wide swath of the present invention.

Fig. 2 The results of the wide swath range deblurring of the spaceborne SAR system according to the embodiment of the present invention, wherein, (a) 100 iterations of deblurring results; (b) 200 iterations of deblurring results; (c) 300 iterations of deblurring results; (d) 500 iterations of defuzzification results; (e) 900 iterations of defuzzification results; (f) 1500 iterations of defuzzification results.

Fig. 3 Imaging results after distance blurring according to an embodiment of the present invention, wherein (a) imaging results of three moving targets; (b) distance profile diagram of the first point; (c) azimuth profile diagram of the first point; (d) ) the distance profile of the second point; (e) the azimuth profile of the second point; (f) the distance profile of the third point; (g) the azimuth profile of the third point.

Detailed ways

The present invention will be further described below with reference to the accompanying drawings and embodiments.

Aiming at the distance ambiguity of the ocean wide swath of the spaceborne SAR system, on the basis of the existing distance deblurring algorithm, the purpose of the present invention is to combine the discrete frequency coded signal and the deep learning technology, and propose a deep learning-based ocean wide A swath range deblurring method, which can achieve range deblurring of echoes with low signal-to-noise ratios.

A deep learning-based method for deblurring the distance of a wide ocean swath, comprising the following steps:

Step 1: Obtain the echo signal of the spaceborne SAR system, which is an n _r ×n _a -dimensional matrix, perform Fourier transform processing on the echo signal matrix by column, and save the result in the matrix s(n1 , n2); _{n r} represents the number of points in the distance direction, and na represents the number of points in the azimuth direction;

s_r(n1)为n_r×1向量；其中，k_r表示调频率，k_r＝B/T_p，B表示发射信号带宽，T_p表示发射脉冲宽度，f_r表示为距离向频域坐标，

Δf为距离频域间隔，

n1＝0,1,...,n_r-1；Step 2: According to the radar parameters, construct the reference signal vector

s_r(n1) is an n _r ×1 vector; among them, k _r represents the frequency modulation, k _r =B/T _p , B represents the transmission signal bandwidth, T _p represents the transmission pulse width, and _fr represents the range-to-frequency domain coordinates,

Δf is the distance frequency domain interval,

n1=0,1,..., _nr -1;

L表示为合成孔径长度，n2＝0,1,...,n_a-1；Step 3: Multiply each column of s(n1, n2) with the conjugate of the reference signal vector s_r(n1) to obtain the data matrix s(f _r , x _n2 ) after the distance pulse pressure; wherein, x _n2 represents Azimuth time domain coordinates,

L represents the synthetic aperture length, n2=0,1,...,n _a -1;

Step 4: Perform inverse Fourier transform processing on the matrix s(f _r , x _n2 ) by column, and store the result in the matrix s(n _r , n _a );

PRF为方位采用频率，Δf_a为方位频域间隔，

Step 5: Fourier transform is performed on the matrix s(n _r ,n _a ) by row, and the result is stored in the s(n _r ,f _a ) matrix; among them, f _a represents the azimuth frequency domain coordinate,

PRF is the azimuth adopted frequency, Δf _a is the azimuth frequency domain interval,

Step 6: Construct a Doppler compensation function s_h(n1) according to the radar parameters, where s_h(n1) is a vector of n _r × 1;

Step 7: Do point multiplication with the conjugate of s(n _r , f _a ) and the Doppler compensation function s_h(n1) vector, and perform inverse FFT processing on the multiplication result row by row to obtain the post-pulse Doppler compensation. The data matrix ss(n _r ,n _a );

Step 8: According to the spatial filtering theory, construct a matrix Ω composed of target scattering coefficients, where Ω is an n _r ×n _a -dimensional matrix;

Step 9: Divide the matrix ss(n _r , na ) into two channel data ss_real( _{n r} , na ) and _{ss_imag} ( _{n r} , na ) according to the real part and the imaginary part; The imaginary part is the two channel data Ω_real(n _r ,n _a ) and Ω_i mag(n _r ,n _a );

Step 10: Construct a deep learning network based on two-channel input and two-channel output of a convolutional neural network; use the matrix ss_real(n _r ,n _a ) and the matrix ss_i mag(n _r ,n _a ) as the input data of the deep learning network, and set the The matrix Ω_real(n _r ,n _a ) and the matrix Ω_i mag(n _r ,n _a ) are used as training label data to train the deep learning network;

Step 11: Input the data to be deblurred into the deep learning network trained in step 10, and obtain the data matrix y_real(n _r ,n _a ) and matrix y_i mag(n _r ,n _a ) after distance deblurring;

Step 12: Combine the matrix y_real(n _r ,n _a ) and the matrix y_i mag(n _r ,n _a ) into the matrix y(n _r ,n _a ) according to the relationship between the real and imaginary parts;

Step 13: Perform Doppler parameter estimation on the matrix y( _{n r} , na ) to obtain the estimated Doppler parameter

Step 14: Perform range migration correction on the matrix y( _{n r} , na ) according to the Doppler parameter f _dc , and store the result in the matrix y′( _{n r} , na ).

s_a(n2)为1×n_a向量；其中，f_c表示雷达发射信号的载频，c为电磁波的传播速度，V为卫星速度，θ为卫星的斜视角，t_n2为方位慢时间；Step 15: According to the radar parameters, construct the reference signal vector

s_a(n2) is _a 1×na vector; among them, f _c represents the carrier frequency of the radar transmission signal, c is the propagation speed of the electromagnetic wave, V is the satellite speed, θ is the oblique angle of the satellite, and t _n2 is the azimuth slow time;

Step 16: Perform FFT processing on the matrix y'(n _r , n _a ) by row to obtain y(n _r , f _a ); take out each row of y(n _r , f _a ), and multiply the average point by the reference signal vector s_a( The conjugate of n2), the data matrix y(n _r , f _a ) after distance pulse pressure is obtained;

Step 17: Perform inverse FFT processing on the matrix y(n _r , f _a ) by row to obtain the SAR focusing result matrix z( _{n r} , na ).

Further, the specific steps of step 8 are as follows:

The spaceborne SAR system transmits encoded signals of multiple code patterns, and the echo signal received by the multi-antenna digital receiver is expressed as A=[s(1),...,s(m),...,s(M)], m=1,2,...,M, where M is the number of swaths; using the spatial filtering vector to filter the signal in the spatial domain, the scattering coefficient of the target can be recovered from the blurred echo signal, that is, the scattering corresponding to the presence of the target The coefficient is 1, and the corresponding scattering coefficient is 0 when there is no target; the weight vector is Ω=[Ω ₁ ,…,Ω _m ,…,Ω _M ]; for the mth mapping with A ^T Ω _m =Y _m , Among them, Y _m =[0,...,1,...0]; the matrix composed of the target scattering coefficients is Ω _m =inv(A)Y _m , and the whole deblurring process is completed by solving Ω.

Specific examples:

The effectiveness of the present invention is further verified through simulation experimental data below.

(1) Simulation experiment

1. Measured parameters

In order to verify the effectiveness of the method of the present invention, the measured data parameters in Table 1 are given here.

Table 1 Simulation data parameters

carrier frequency 5.3GHz Sampling frequency 200MHz Velocity of spaceborne SAR systems 7100m/s Signal bandwidth 160MHz Scene centerline distance 850km Double the blur distance 75km Pulse repetition frequency (PRF) 2000Hz Number of swaths 3 Number of moving targets 3 moving target speed -8m/s,4m/s,10m/s

2. Experiment content

FIG. 2 illustrates a simulation data processing result obtained by using a deep learning-based ocean wide swath distance deblurring algorithm proposed by the present invention. Among them, (a) 100 iterations of defuzzification results; (b) 200 iterations of defuzzification results; (c) 300 iterations of defuzzification results; (d) 500 iterations of defuzzification results; (e) 900 iterations of defuzzification Results; (f) 1500 iterations of deblurring results. It can be seen from the figure that the distance deblurring effect of the method of the present invention is used for training with a deep learning network. When the number of iterations reaches 300 times, a better distance deblurring result can be obtained.

3 shows the results of SAR imaging using the data after distance deblurring according to the present invention, wherein (a) the imaging results of three moving targets; (b) the distance profile of the first point; (c) the first point The azimuth profile of ; (d) the distance profile of the second point; (e) the azimuth profile of the second point; (f) the distance profile of the third point; (g) the azimuth profile of the third point. It can be seen from the figure that the method of the present invention can realize two-dimensional high-resolution SAR imaging of ocean moving targets at different speeds by de-blurring the data.

Therefore, the method of the present invention can effectively solve the problem of de-ambiguity of the wide swath of the spaceborne SAR system.

Claims (2)

1. a kind of ocean wide swath distance deblurring method based on deep learning, is characterized in that, comprises the steps: Step 1: Obtain the echo signal of the spaceborne SAR system, which is an n _r ×n _a -dimensional matrix, perform Fourier transform processing on the echo signal matrix by column, and save the result in the matrix s(n1 , n2); _{n r} represents the number of points in the distance direction, and na represents the number of points in the azimuth direction; Step 2: According to the radar parameters, construct the reference signal vector

s_r(n1) is an n _r ×1 vector; among them, k _r represents the frequency modulation, k _r =B/T _p , B represents the transmission signal bandwidth, T _p represents the transmission pulse width, and _fr represents the range-to-frequency domain coordinates,

Δf is the distance frequency domain interval,

Step 3: Multiply each column of s(n1, n2) with the conjugate of the reference signal vector s_r(n1) to obtain the data matrix s(f _r , x _n2 ) after the distance pulse pressure; wherein, x _n2 represents Azimuth time domain coordinates,

L is expressed as the synthetic aperture length, n2=0, 1, . . . , n _a -1; Step 4: Perform inverse Fourier transform processing on the matrix s(f _r , x _n2 ) by column, and store the result in the matrix s(n _r , n _a ); Step 5: Fourier transform is performed on the matrix s( _{n r} , na ) by row, and the result is stored in the s(n _r , _{f a} ) matrix;

PRF is the azimuth adopted frequency, Δf _a is the azimuth frequency domain interval,

Step 6: Construct the Doppler compensation function according to the radar parameters

s_h(n1) is a vector of n _r × 1; among them, n _r represents the number of points in the range direction, v is the radial velocity of the moving target, V is the moving speed of the radar platform, θ is the azimuth angle of the slant range plane radar line of sight, and λ is the radar The working wavelength of , g is the acceleration of gravity; Step 7: Do point multiplication with the conjugate of s(n _r , f _a ) and the Doppler compensation function s_h(n1) vector, and perform inverse FFT processing on the multiplication result row by row to obtain the intrapulse Doppler compensation. The data matrix ss(n _r , n _a ); Step 8: According to the spatial filtering theory, construct a matrix Ω composed of target scattering coefficients, where Ω is an n _r ×n _a -dimensional matrix; Step 9: Divide the matrix ss(n _r , na ) into two channel data ss_real( _{n r} , na ) and _{ss_imag} ( _{n r} , na ) according to the real and imaginary parts; divide the matrix Ω according to the real and imaginary parts The part is two channel data Ω_real( _{n r} , na ) and Ω_imag( _{n r} , na ); Step 10: Construct a deep learning network based on two-channel input and two-channel output of a convolutional neural network; use the matrix ss_real( _{n r} , na ) and the matrix ss_imag( _{n r} , na ) as the input data of the deep learning network, and the matrix Ω_real (n _r , n _a ) and the matrix Ω_imag (n _r , n _a ) are used as training label data to train the deep learning network; Step 11: Input the data that needs to be deblurred into the deep learning network trained in step 10, and obtain the data matrix y_real( _{n r} , na ) and matrix y_imag( _{n r} , na ) after distance deblurring; Step 12: Combine the matrix y_real( _{n r} , na ) and the matrix y_imag(n _r , na ) into the matrix _y ( _{n r} , na ) according to the relationship between the real part and the imaginary part; Step 13: Perform Doppler parameter estimation on the matrix y( _{n r} , na ) to obtain the estimated Doppler parameter

Step 14: Perform range migration correction on the matrix y( _{n r} , na ) according to the Doppler parameter f _dc , and store the result in the matrix y′( _{n r} , na ). Step 15: According to the radar parameters, construct the reference signal vector

s_a(n2) is _a 1×na vector; among them, f _c represents the carrier frequency of the radar transmission signal, c is the propagation speed of the electromagnetic wave, V is the satellite speed, θ is the oblique angle of the satellite, and t _n2 is the azimuth slow time; Step 16: Perform FFT processing on the matrix y'( _{n r} , na ) by row to obtain y(n _r , f _a ); take out each row of y(n _r , f _a ), and multiply the reference signal vector s_a( The conjugate of n2), the data matrix y(n _r , f _a ) after distance pulse pressure is obtained; Step 17: Perform inverse FFT processing on the matrix y(n _r , f _a ) by row to obtain the SAR focusing result matrix z( _{n r} , na ). 2. a kind of ocean wide swath distance deblurring method based on deep learning according to claim 1, is characterized in that, described step 8 concrete steps are as follows: The spaceborne SAR system transmits encoded signals of multiple code patterns, and the echo signal received by the multi-antenna digital receiver is expressed as A=[s(1),...,s(m),...,s(M)], m=1, 2, ..., M, M is the number of swaths, and s(m) represents the echo signal of the mth swath received by the multi-antenna digital receiver; The scattering coefficient of the target is recovered from the fuzzy echo signal, that is, the corresponding scattering coefficient is 1 when there is a target, and 0 when there is no target; the weight vector is Ω=[Ω ₁ ,...,Ω _{m , .} ^_ _{_ _ _} (A) Y _m , the whole defuzzification process is completed by solving Ω.

Images