A method to improve the phase synchronization accuracy of bistatic SAR based on compressed sensing
Through the compression perception method, the high-precision imaging phase structure prototype and prior information of the dual-based SAR system are learned, and combined with the maximum posterior probability estimation and orthogonal matching tracking algorithm, the noise pollution problem of phase synchronization error in the dual-based SAR system is solved, and the noise pollution problem of phase synchronization error is realized. Higher precision phase synchronization and imaging effects.
Patent Information
- Application Number
- CN202210727007.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-06-24
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2042-06-24
AI Technical Summary
In the dual-based SAR system, the phase synchronization error caused by frequency source deviation is contaminated by noise, resulting in a decrease in synchronization accuracy. The existing technology such as the Kalman filtering method has a time lag limitation, making it difficult to further improve synchronization accuracy.
By acquiring imaging phase difference, synchronous compensation phase and Doppler compensation phase, the structural prototype and prior information of high-precision imaging phase are learned using compression perception methods, and the noise component is eliminated using the maximum posterior probability estimate, combined with the orthogonal matching tracking algorithm and The K-SVD algorithm performs sparse encoding and denoising processing to improve synchronization accuracy.
It effectively reduces the noise component in the synchronization compensation phase, avoids time lag problems, improves the synchronization accuracy of the dual-base SAR phase, and realizes higher accuracy imaging and interference processing.
Smart Images

Figure CN115032638B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of radar technology, and in particular to a method for improving the phase synchronization accuracy of a bistatic SAR (Synthetic Aperture Radar, SAR) based on compressed sensing. Background Art
[0002] The bistatic synthetic aperture radar (SAR) system is a microwave imaging system with separate transmitter and receiver. Compared with the traditional monostatic SAR, bistatic SAR has attracted much attention due to its unique advantages such as rich observation modes, diverse system configurations, flexible baseline configurations, good electromagnetic concealment and wide application fields. It has become a key development direction in the field of microwave remote sensing and earth observation.
[0003] However, while the split-transmitter-receiver configuration gives the bistatic SAR system the above advantages, it also brings new difficulties and challenges, such as phase synchronization. Since different platforms use different frequency sources, the deviation of the frequency source will produce additional residual modulation phase in the echo signal received by the auxiliary satellite. This modulation term will cause problems such as target defocusing and imaging position offset. Therefore, phase synchronization is a key problem that needs to be solved in the bistatic SAR system. The current advanced phase synchronization scheme is the non-interrupt phase synchronization scheme adopted by my country's Land Exploration No. 1 (LT-1) satellite. The synchronization signal is exchanged in the time slot between the end of the radar signal receiving window and the start of the next pulse repetition interval, thereby avoiding interruption of normal data collection and greatly improving the synchronization frequency.
[0004] Ideally, after phase compensation, the phase error caused by system frequency deviation can be eliminated. But in fact, the synchronization signal will be affected by thermal noise, free space attenuation, multipath effect, etc. in the link, and the synchronization phase will be contaminated by noise, resulting in reduced synchronization accuracy.
[0005] In the framework of the non-interrupt phase synchronization scheme, in order to reduce the interference of noise, the synchronization accuracy can be further improved through coherent accumulation. However, when the synchronization frequency is low, the synchronization accuracy will decrease as the signal loses coherence. In addition, there is a trade-off between the coherence and the average of the signal, and the selection of the average needs to be quite cautious. Existing phase denoising schemes such as the Kalman filtering method have improved the accuracy of phase synchronization to a certain extent. However, the Kalman filter estimates the current state by combining the observed value and the predicted value, in which the predicted value is derived from the estimation of the previous state. Therefore, the estimated value of the current state has a certain time lag, which limits the further improvement of the synchronization accuracy. Learning the prior information in the high-precision imaging phase through the compressed sensing method and then using the maximum a posteriori probability estimation to eliminate the noise component is a reliable solution to improve the phase synchronization accuracy of bistatic SAR. Summary of the invention
[0006] In view of this, the present invention provides a method for improving the phase synchronization accuracy of a bistatic SAR based on compressed sensing, which can reduce the noise component in the synchronization compensation phase and effectively improve the phase synchronization accuracy of the bistatic SAR.
[0007] To achieve the above object, the technical solution of the present invention is achieved as follows:
[0008] A method for improving phase synchronization accuracy of bistatic SAR based on compressed sensing includes the following steps:
[0009] Step 1, obtaining imaging phase difference, synchronization compensation phase, internal calibration compensation phase and Doppler compensation phase, and estimating the signal-to-noise ratio of the synchronization signal;
[0010] Step 2: Downsample the imaging phase difference and extract its high-order terms by the least square method; use the high-order term phase to train an overcomplete dictionary to learn the structural prototype and prior information of the high-precision imaging phase;
[0011] Step 3: Separate the linear term and the high-order term of the synchronization compensation phase by the least square method, and use the orthogonal matching pursuit algorithm to sparsely encode the high-order term of the synchronization compensation phase;
[0012] Step 4: Use the signal-to-noise ratio of the synchronization signal to determine the relevant parameters, use the maximum a posteriori probability estimation to denoise the high-order terms of the synchronization compensation phase, and sum them with the linear terms of the synchronization compensation phase to obtain the denoised synchronization compensation phase;
[0013] Step 5: After upsampling the denoised synchronous compensation phase, the internal calibration compensation phase and the Doppler compensation phase are compensated to obtain a high-precision synchronous compensation phase.
[0014] Furthermore, the step 2 comprises:
[0015] The imaging phase difference is downsampled using the cubic spline interpolation method, and a polynomial fitting is performed on the downsampled imaging phase difference using the least squares method. The fitted linear phase is subtracted from the imaging phase difference to extract the high-order terms.
[0016] Furthermore, in step 2, the high-order phase is used to train an overcomplete dictionary to learn the structural prototype and prior information of the high-precision imaging phase, specifically:
[0017] The high-order phase term is input into the K-SVD algorithm, and an over-complete dictionary containing high-precision imaging phase structure prototypes and prior information is trained through an alternating optimization strategy.
[0018] Furthermore, the step 4 comprises:
[0019] The signal-to-noise ratio of the synchronization signal is used to determine the approximation weight, and the approximation weight, the over-complete dictionary and the sparse coefficients of the synchronization compensation phase are input into the maximum a posteriori probability estimation model to eliminate the noise component in the high-order terms of the synchronization compensation phase. The noise component is then summed with the linear terms of the synchronization compensation phase to obtain the denoised synchronization compensation phase.
[0020] Furthermore, the step 5 comprises:
[0021] The denoised synchronous compensation phase is upsampled using the cubic spline interpolation method, and then the time-varying internal compensation phase and Doppler compensation phase are calibrated to obtain a high-precision synchronous compensation phase, which is used to accurately compensate the phase of bistatic SAR data, thereby achieving precise imaging and interferometric processing.
[0022] Beneficial effects:
[0023] Compared with the existing coherent accumulation method, the present invention does not need to consider the problems of decoherence and sensitivity of average selection. Compared with the Kalman method, the present invention reconstructs the synchronous compensation phase from the phase structure prototype, thereby avoiding the problem of time lag. In addition, the present invention also obtains the prior information in the imaging phase difference by learning the dictionary, and obtains a higher precision phase synchronization effect. BRIEF DESCRIPTION OF THE DRAWINGS
[0024] Figure 1 The method flow for improving the phase synchronization accuracy of bistatic SAR based on compressed sensing;
[0025] Figure 2 This is a schematic diagram of the hardware structure of the ground verification system for the Land Exploration No. 1 non-interruption phase synchronization scheme;
[0026] Figure 3 This is a photo of the four-arm helical antenna installed on the Land Exploration No. 1 satellite;
[0027] Figure 4a , Figure 4b is the residual phase error and its statistical histogram without synchronous compensation phase denoising; where, Figure 4a is the residual phase error diagram, Figure 4b is a statistical histogram;
[0028] Figure 5a , Figure 5b is the residual phase error and its statistical histogram after synchronous compensation phase denoising, where: Figure 5a is the residual phase error diagram, Figure 5b It is a statistical histogram. DETAILED DESCRIPTION
[0029] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all 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 protection scope of the present invention.
[0030] According to an embodiment of the present invention, Figure 1 As shown, a method for improving the phase synchronization accuracy of a bistatic SAR based on compressed sensing of the present invention comprises the following steps:
[0031] Step 101: Acquire imaging phase difference, synchronization compensation phase, internal calibration compensation phase and Doppler compensation phase, and estimate the signal-to-noise ratio of the synchronization signal.
[0032] In the ground verification system of the non-interrupted phase synchronization scheme, the bistatic SAR radar signal is transmitted through an optical delay line. After pulse compression of the imaging echo, its phase difference can be expressed as:
[0033]
[0034] in, and They represent the peak phases of the pulse pressure of the imaging echo of the primary and auxiliary satellites respectively.
[0035] The main and auxiliary satellites exchange synchronization signals through the time slot between the end of the radar signal receiving window and the start of the next pulse repetition interval. After the two sets of synchronization signals are pulse compressed, their signal-to-noise ratio (SNR) is estimated, and then the phase at the peak position is extracted and expressed as and and The difference is used to obtain the synchronous compensation phase:
[0036]
[0037] Extract the phase error introduced by the hardware system from the radar receiving loop CR, reference loop RE, synchronous transmitting loop ST and synchronous receiving loop SR, that is, the internal calibration compensation phase:
[0038]
[0039] in, and Respectively represent the phases of loops CR, RE, ST, and SR. a and{·} b It means that the phases collect the main star and auxiliary star respectively.
[0040] When in orbit, the satellite is in high-speed motion, and the Doppler frequency between satellites is f d (t) = f0v ab (t) / c, where v ab (t) represents the relative speed between the primary and secondary satellites, and c represents the speed of light. Assume that the transmission time of the inter-satellite synchronization signal is τ sys , then the Doppler phase error can be expressed as:
[0041]
[0042] Step 102: downsample the imaging phase difference, extract its high-order terms by least square method, use the phase of the high-order terms to train an over-complete dictionary, and learn the structural prototype and prior information of the high-precision imaging phase.
[0043] Since the synchronization frequency is less than the pulse repetition frequency (PRF), the imaging phase difference is downsampled to the synchronization frequency by uniform / non-uniform interpolation using the cubic spline interpolation method before dictionary training. can be decomposed into linear terms and higher-order terms
[0044]
[0045] Among them, △f0 represents the USO frequency difference of the dual-base system, which can usually be considered to be constant over a short period of time. Therefore, the phase characteristics are mainly determined by the high-order terms reflect.
[0046] Fitting by Least Squares Method (LSM) Then separate the higher-order terms Original phase As input data, the Ramanujan Sums (RS) transformation matrix is selected as the initial basis, and the K-SVD algorithm with an alternating optimization strategy is used to complete the training of the overcomplete dictionary. After training with the K-SVD algorithm, the overcomplete dictionary D contains the relevant features of the clean phase, thereby obtaining the prior information of the ideal synchronization compensation phase that is not contaminated by noise.
[0047] Step 103: Separate the linear term and the high-order term of the synchronization compensation phase by the least square method, and perform sparse coding on the high-order term of the synchronization compensation phase by using the orthogonal matching pursuit algorithm.
[0048] Separation of synchronous compensation phase by least squares method The linear term and higher-order terms Note the higher-order terms To measure the phase Now take the synchronous phase segment y of length n from Y i =R i Y, where the matrix is the matrix for extracting the i-th phase from the measured phase. Assuming the overcomplete dictionary It is known that k represents the number of atoms in the dictionary, and k>n. According to the theory of compressed sensing, each segment in the original phase can be sparsely represented by the dictionary D, that is:
[0049]
[0050] Among them, ||·||1 and ||·||2 represent the L of the vector respectively. 1 Norm and L 2 Norm. is the vector y i Sparse representation under precision ε. The orthogonal matching pursuit (OMP) algorithm is used to solve the sparse solution of the above problem. When the solution is sparse enough, the OMP algorithm is very simple and efficient, and can obtain a fairly accurate solution.
[0051] Step 104: Use the signal-to-noise ratio of the synchronization signal to determine the relevant parameters, use the maximum a posteriori probability estimation to denoise the high-order terms of the synchronization compensation phase, and sum them with the linear terms of the synchronization compensation phase to obtain the denoised synchronization compensation phase.
[0052] Through the maximum a posteriori probability (MAP) estimation, the denoising of the measured phase Y is equivalent to the energy minimization problem:
[0053]
[0054] The first term requires the approximation between the measured phase Y and the denoised phase X. λ represents the weight of the approximation. The larger its value, the better the estimated phase The closer it is to the measured phase Y. The second term requires that each phase segment can be represented by the dictionary D and the sparse coefficient α within a bounded error. i The last term is to express that the coefficients of each phase segment are as sparse as possible, μ i is the weight of the corresponding phase segment.
[0055] In order to obtain all the sparse coefficients α in equation (6) i After that, formula (7) is transformed into the following optimization problem:
[0056]
[0057] The above formula is a typical L 2 Norm optimization problem, with analytical solution:
[0058]
[0059] The weight λ of the approximation is an important parameter of the algorithm and is negatively correlated with the standard deviation of the noise V. The higher the noise level, the lower the The similarity between λ and Y is used to obtain a better denoising effect. According to experimental experience, λ = 0.01 / σ. In fact, the standard deviation of noise σ is affected by the signal-to-noise ratio error, interpolation error, and aliasing error. When the synchronization frequency is high, the interpolation error and aliasing error can be ignored, and σ is mainly determined by the signal-to-noise ratio error:
[0060]
[0061] Among them, f syn is the synchronization frequency, H syn (f) is the noise spectrum. For additive white Gaussian noise, H syn (f) = 1. az (f) is the azimuth transfer function used for imaging processing. SNR can be estimated from the synchronization signal after pulse compression. Therefore, the weight of the approximation can refer to formula (10).
[0062] Step 105: After upsampling the de-noised synchronous compensation phase, the internal calibration compensation phase and the Doppler compensation phase are compensated to obtain a high-precision synchronous compensation phase.
[0063] After denoising, the synchronous compensation phase is upsampled to the same length as the imaging phase difference using cubic strip interpolation. In addition to noise, the Doppler effect between the hardware system and the satellite will also introduce phase errors, which require internal calibration compensation phase and Doppler compensation phase The final accurate synchronous compensation phase can be expressed as:
[0064]
[0065] in, represents the synchronous compensation phase after denoising, which is the high-order term after denoising With linear terms The sum of the two. Obtain high-precision synchronous compensation phase After that, it can be used to accurately compensate the phase of bistatic SAR data, thereby achieving subsequent precise imaging and interferometric processing.
[0066] Example 1
[0067] The embodiment selects the measured data of the ground verification system of the Land Exploration No. 1 mission for processing. The hardware structure of the Land Exploration No. 1 synchronization system is as follows: Figure 2 The imaging signal is transmitted in an optical delay line rather than in free space to maintain a stable high SNR. The synchronization signal is used as Figure 3The four pairs of four-arm spiral synchronous antennas shown in the figure realize inter-satellite transmission. The antenna pairs are reasonably selected according to the principle of maximum signal-to-noise ratio to achieve 360° full coverage. The synchronization signal selects a linear frequency modulation signal with the same carrier frequency as the radar signal. After demodulation and pulse compression, the corresponding phase is extracted from the peak position.
[0068] Figure 4a , Figure 4b The residual phase error and its statistical histogram without synchronous compensation phase denoising are shown. As can be seen from the figure, the probability density function shows that the residual phase error follows a Gaussian distribution, and the standard deviation of the residual error before denoising is 0.6163°. Figure 5a , Figure 5b The residual phase and its distribution after denoising by the present invention are shown, the standard deviation of the noise is reduced to 0.2273°, and the accuracy of the synchronized phase is improved by 63%. The above experimental results verify the effectiveness of the method for improving the phase synchronization accuracy of the bistatic SAR based on compressed sensing, and have the potential to achieve high-precision phase synchronization of the Land Exploration No. 1 bistatic SAR system on orbit.
[0069] The above descriptions are only some embodiments of the present invention. The present invention is still applicable in other cases and is not intended to limit the protection scope of the present invention.
Claims
1. A method for improving the phase synchronization accuracy of bistatic SAR based on compressed sensing, characterized in that: The method includes: Step 1, obtaining imaging phase difference, synchronization compensation phase, internal calibration compensation phase and Doppler compensation phase, and estimating the signal-to-noise ratio of the synchronization signal; Step 2: Downsample the imaging phase difference and extract its high-order terms by the least square method; use the high-order term phase to train an overcomplete dictionary to learn the structural prototype and prior information of the high-precision imaging phase; Step 3: Separate the linear term and the high-order term of the synchronization compensation phase by the least square method, and use the orthogonal matching pursuit algorithm to sparsely encode the high-order term of the synchronization compensation phase; Step 4: Use the signal-to-noise ratio of the synchronization signal to determine the relevant parameters, use the maximum a posteriori probability estimation to denoise the high-order terms of the synchronization compensation phase, and sum them with the linear terms of the synchronization compensation phase to obtain the denoised synchronization compensation phase; Step 5: After upsampling the denoised synchronous compensation phase, the internal calibration compensation phase and the Doppler compensation phase are compensated to obtain a high-precision synchronous compensation phase.
2. The method for improving the phase synchronization accuracy of bistatic SAR based on compressed sensing according to claim 1, characterized in that: The step 2 comprises: The imaging phase difference is downsampled using the cubic spline interpolation method, and a polynomial fitting is performed on the downsampled imaging phase difference using the least squares method. The fitted linear phase is subtracted from the imaging phase difference to extract the high-order terms.
3. The method for improving the phase synchronization accuracy of bistatic SAR based on compressed sensing according to claim 2, characterized in that: In step 2, the high-order phase is used to train an over-complete dictionary to learn the structural prototype and prior information of the high-precision imaging phase, specifically: The high-order phase term is input into the K-SVD algorithm, and an over-complete dictionary containing high-precision imaging phase structure prototypes and prior information is trained through an alternating optimization strategy.
4. The method for improving the phase synchronization accuracy of bistatic SAR based on compressed sensing according to claim 3, characterized in that: The step 4 comprises: The signal-to-noise ratio of the synchronization signal is used to determine the approximation weight, and the approximation weight, the over-complete dictionary and the sparse coefficients of the synchronization compensation phase are input into the maximum a posteriori probability estimation model to eliminate the noise component in the high-order terms of the synchronization compensation phase. The noise component is then summed with the linear terms of the synchronization compensation phase to obtain the denoised synchronization compensation phase.
5. The method for improving the phase synchronization accuracy of bistatic SAR based on compressed sensing according to claim 4, characterized in that: The step 5 comprises: The denoised synchronous compensation phase is upsampled using the cubic spline interpolation method, and then the time-varying internal compensation phase and Doppler compensation phase are calibrated to obtain a high-precision synchronous compensation phase, which is used to accurately compensate the phase of bistatic SAR data, thereby achieving precise imaging and interferometric processing.
Citation Information
Patent Citations
Doppler compensation method for synthetic aperture underwater acoustic communication and system
CN103888168A
Phase difference-based inverse synthetic radar (ISAR) translational motion compensation method under condition of low signal-noise ratio
CN109031299A
Adaptive compensation of doppler shift in a mobile communication system
CN1215512A
Method of non-uniform doppler compensation for wideband orthogonal frequency division multiplexed signals
US20090067514A1
