Phase compensation algorithm for low signal-to-noise ratio of synthetic aperture radar (ISAR) imaging of sea targets

Abstract

The present invention relates to the field of radar imaging technology, and discloses a low signal-to-noise ratio phase compensation algorithm for synthetic aperture radar (ISAR) imaging of sea targets, including: transmitting a radar signal to a sea target through a synthetic aperture radar system to obtain a signal to be processed; determining the signal-to-noise ratio to determine a compensation scheme for the signal to be processed; when the signal-to-noise ratio is lower than a signal-to-noise ratio threshold, extracting a characteristic index and determining a constant index group and a variable index group; obtaining a characteristic vector of the signal to be processed and comparing it with a historical data group to determine a compensation value; when the similarities are all less than or equal to the similarity threshold, determining a first compensation value based on a maximum likelihood phase error, and determining a second compensation value based on a convolutional neural network model, and determining a compensation value based on the first compensation value and the second compensation value. The present application improves the phase compensation accuracy under low signal-to-noise ratio conditions through a multi-level, dynamically adaptive phase compensation mechanism.

Description

Technical Field

[0001] The present invention relates to the field of radar imaging technology, and in particular to a low signal-to-noise ratio phase compensation algorithm for synthetic aperture radar (ISAR) imaging of sea targets. Background Art

[0002] With the widespread application of Synthetic Aperture Radar (SAR) technology, especially in the fields of ocean monitoring, national defense security and disaster warning, the radar system's requirements for target imaging accuracy are gradually increasing. In these application scenarios, ISAR (Inverse Synthetic Aperture Radar) imaging, as an important imaging technology, can effectively perform high-resolution imaging of moving marine targets. Compared with traditional SAR, ISAR can use the relative motion of the target to improve the imaging resolution, especially in the sea or other complex environments, and has extremely high application value for target tracking and identification.

[0003] However, ISAR imaging of maritime targets often faces serious challenges, especially in low signal-to-noise ratio (SNR) conditions. Due to the influence of environmental noise and signal attenuation, the noise and interference in the echo signal are significant, affecting the imaging quality and the accuracy of target detection. Traditional phase compensation methods usually adopt static solutions, lack of adaptability, and cannot adjust the compensation strategy according to the actual situation under different signal-to-noise ratio conditions.

[0004] Therefore, it is necessary to design a low signal-to-noise ratio phase compensation algorithm for synthetic aperture radar ISAR imaging of sea targets to solve the problems existing in current technology. Summary of the invention

[0005] In view of this, the present invention proposes a low signal-to-noise ratio phase compensation algorithm for synthetic aperture radar (ISAR) imaging of sea targets, aiming to solve the problem that current phase compensation methods usually adopt static solutions, lack of adaptability and cannot adjust the compensation strategy according to actual conditions.

[0006] The present invention proposes a low signal-to-noise ratio phase compensation algorithm for synthetic aperture radar (ISAR) imaging of sea targets, comprising:

[0007] Transmitting a radar signal to a sea target through a synthetic aperture radar system, cyclically collecting echo signals within a preset time period at least three times when the sea target moves in the same direction, selecting three sections of the echo signals for noise processing to obtain a signal to be processed;

[0008] Determining a signal-to-noise ratio according to the signal to be processed, and determining a compensation scheme for the signal to be processed according to the signal-to-noise ratio;

[0009] When the signal-to-noise ratio is lower than the signal-to-noise ratio threshold, extracting characteristic indexes of each echo signal in the signal to be processed, comparing one set of characteristic indexes with the other two sets, and determining a constant index group and a variable index group;

[0010] Obtaining a characteristic vector of the signal to be processed according to the constant index group and the variable index group, comparing the characteristic vector with a historical data group, and determining a compensation value according to the comparison result;

[0011] When the similarity between the historical feature vector in the historical data group and the feature vector is less than or equal to the similarity threshold, a first compensation value is determined based on the maximum likelihood phase error, and a second compensation value is determined based on the convolutional neural network model, and the compensation value is determined according to the first compensation value and the second compensation value.

[0012] Further, when performing noise processing on the echo signal to obtain a signal to be processed, it includes:

[0013] The Daubechies 4 wavelet is used for discrete wavelet transform, and the echo signal is decomposed by 4 layers of wavelets to obtain the approximation coefficient and detail coefficient:

[0014] The soft threshold method is used to perform threshold denoising on the detail coefficients, and the threshold is set to 1.5 times the standard deviation of the signal noise;

[0015] The denoised coefficients are used to perform inverse wavelet transform, and the denoised signal is reconstructed to obtain the signal to be processed.

[0016] Further, when determining a signal-to-noise ratio according to the signal to be processed, and determining a compensation scheme for the signal to be processed according to the signal-to-noise ratio, it includes:

[0017] Calculating the square mean of the signal to be processed to obtain the power of the signal to be processed;

[0018] Collecting echo signals from a target-free silent area, and calculating the power of the echo signals from the target-free silent area to obtain noise power;

[0019] The signal-to-noise ratio is obtained using the following formula:

[0020]

[0021] Among them, SNR represents signal-to-noise ratio, P represents the power of the signal to be processed, and P0 represents the noise power;

[0022] Comparing the signal-to-noise ratio with a signal-to-noise ratio threshold, and determining a compensation scheme for the signal to be processed according to the comparison result;

[0023] When the signal-to-noise ratio is higher than or equal to the signal-to-noise ratio threshold, using Doppler frequency for phase compensation;

[0024] When the signal-to-noise ratio is lower than the signal-to-noise ratio threshold, the characteristic index of each echo signal in the signal to be processed is extracted, one group of characteristic indexes is compared with the other two groups, a constant index group and a variable index group are determined, a characteristic vector of the signal to be processed is obtained according to the constant index group and the variable index group, and a compensation value is determined according to the characteristic vector.

[0025] Furthermore, when the Doppler frequency is used for phase compensation, it includes:

[0026] Determine the relative speed of the sea target and the operating frequency of the radar to obtain a Doppler frequency, and calculate a phase error according to the Doppler frequency;

[0027] The phase error is used as the compensation value, and the compensation value is obtained by calculating the following formula:

[0028]

[0029] in, represents the compensation value, V represents the relative speed of the sea target, represents the wavelength of the radar signal, represents the center frequency of the radar, and t represents the signal sampling time.

[0030] Further, when determining the compensation value according to the characteristic vector, it includes:

[0031] When there is data in the historical data group whose similarity with the feature vector is greater than a similarity threshold, determining the compensation value according to the historical compensation values ​​in the historical data group;

[0032] When the similarity between the historical feature vector in the historical data group and the feature vector is less than or equal to the similarity threshold, a first compensation value is determined based on the maximum likelihood phase error, and a second compensation value is determined based on the convolutional neural network model, and the compensation value is determined according to the first compensation value and the second compensation value.

[0033] Further, when determining the compensation value according to the historical compensation value in the historical data group, it includes:

[0034] Constructing data in the historical data group whose similarity to the feature vector is greater than a similarity threshold into a similarity data set; the similarity data set includes a historical feature vector and a historical compensation value;

[0035] Acquire a mean value of historical compensation values ​​according to the similar data set, and split the similar data set according to the mean value of historical compensation values;

[0036] Classifying the data whose historical compensation value is greater than the average historical compensation value as a first similar data set;

[0037] Classifying the data whose historical compensation value is equal to the average of the historical compensation values ​​as a second similar data set;

[0038] Classifying the data whose historical compensation value is less than the average historical compensation value as a third similar data set;

[0039] The compensation value is determined according to the historical compensation value average and the first similar data set and the third similar data set.

[0040] Further, when the compensation value is determined according to the historical compensation value average and the first similar data set and the third similar data set, the compensation value is calculated by the following formula:

[0041]

[0042] in, represents the compensation value, N is the number of historical compensation values ​​in the first similar data set, is the i-th historical compensation value in the first similar data set, represents the mean value of historical compensation values, M is the number of historical compensation values ​​in the third similar data set, is the mth historical compensation value in the third similar data set.

[0043] Further, when the first compensation value is determined based on the maximum likelihood phase error, the expression is:

[0044]

[0045] in, represents the first compensation value, represents a given phase error, Indicates a signal to be processed. represents an ideal echo signal without phase error, and j represents an imaginary unit.

[0046] Further, when determining the compensation value according to the first compensation value and the second compensation value, it includes:

[0047] Obtaining a compensation value difference according to the first compensation value and the second compensation value, wherein the compensation value difference is an absolute value of a difference between the first compensation value and the second compensation value, and determining the compensation value according to the compensation value difference;

[0048] When the compensation value difference is less than a% of the first compensation value, and the compensation value difference is less than a% of the second compensation value, obtaining a mean compensation value according to the first compensation value and the second compensation value, and using the mean compensation value as the compensation value;

[0049] When the compensation value difference is greater than or equal to a% of the first compensation value, or the compensation value difference is greater than or equal to a% of the second compensation value, the first compensation value is adjusted according to an adjustment coefficient determined according to the compensation value difference to determine the compensation value.

[0050] Further, the first compensation value is adjusted by determining an adjustment coefficient according to the compensation value difference, and when determining the compensation value, it includes:

[0051] When the first compensation value is greater than the second compensation value, the first compensation value is adjusted by determining a reduction coefficient according to the compensation value difference to determine the compensation value, and the reduction coefficient is inversely proportional to the compensation value difference;

[0052] When the first compensation value is less than the second compensation value, an increase coefficient is determined according to the compensation value difference to adjust the first compensation value to determine the compensation value, and the increase coefficient is proportional to the compensation value difference.

[0053] Compared with the prior art, the beneficial effects of the present invention are: dynamically selecting different compensation strategies through signal-to-noise ratio evaluation ensures that the traditional Doppler frequency compensation method can be quickly adopted in the case of high signal-to-noise ratio, and accurately compensated by extracting signal features, comparing with historical data, and combining maximum likelihood estimation and convolutional neural network model in the case of low signal-to-noise ratio. Automatically adjusting the compensation strategy according to different environments and signal characteristics enhances the adaptability and robustness, and improves the phase compensation accuracy in the case of low signal-to-noise ratio, which is conducive to obtaining clearer and more accurate target information in the imaging of complex sea surfaces and dynamic targets. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Moreover, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:

[0055] Figure 1 A flowchart of a low signal-to-noise ratio phase compensation algorithm for synthetic aperture radar (ISAR) imaging of sea targets provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0056] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art. It should be noted that, in the absence of conflict, the embodiments of the present invention and the features described in the embodiments can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0057] In some embodiments of the present application, see Figure 1 As shown, a low signal-to-noise ratio phase compensation algorithm for synthetic aperture radar ISAR imaging of sea targets includes:

[0058] S100: transmitting a radar signal to a sea target through a synthetic aperture radar system, cyclically collecting echo signals within a preset time period at least three times when the sea target moves in the same direction, and selecting three echo signals for noise processing to obtain a signal to be processed.

[0059] S200: Determine a signal-to-noise ratio according to the signal to be processed, and determine a compensation scheme for the signal to be processed according to the signal-to-noise ratio.

[0060] S300: When the signal-to-noise ratio is lower than the signal-to-noise ratio threshold, extract the characteristic index of each echo signal in the signal to be processed, compare one group of characteristic indexes with the other two groups, and determine a constant index group and a variable index group.

[0061] S400: Obtain a feature vector of the signal to be processed according to the constant index group and the variable index group, compare the feature vector with the historical data group, and determine a compensation value according to the comparison result.

[0062] When the similarity between the historical feature vector and the feature vector in the historical data group is less than or equal to the similarity threshold, the first compensation value is determined based on the maximum likelihood phase error, and the second compensation value is determined based on the convolutional neural network model, and the compensation value is determined according to the first compensation value and the second compensation value.

[0063] Specifically, in S100, a radar signal is transmitted to the sea target through a synthetic aperture radar system, and echo signals within at least three preset time periods are collected when the sea target moves in the same direction. The same direction movement means that the sea target does not change the direction of movement, and the echo signals are collected multiple times for noise processing. The signals to be processed containing target information are obtained, and these signals can better reflect the real characteristics of the target after noise removal. Noise processing usually includes technical means such as signal filtering and denoising, the purpose of which is to reduce random noise in the echo signal and ensure that the signal to be processed can express the movement and position of the target as accurately as possible. In S200, the signal-to-noise ratio is calculated according to the signal to be processed after noise processing. The signal-to-noise ratio is an important indicator for evaluating signal quality, which reflects the ratio between the effective information of the signal and the noise. By measuring the signal-to-noise ratio, the current signal quality is judged, so as to decide whether to use a simpler compensation method or a more complex compensation strategy. The calculation of the signal-to-noise ratio is achieved by comparing the signal strength and the variance of the noise. When the signal-to-noise ratio is low, it means that the noise has a greater impact and a complex compensation method needs to be used. When the signal-to-noise ratio is high, a simpler compensation method, such as Doppler compensation, can be used. In S300, when the signal-to-noise ratio is lower than the set threshold, the characteristic index of each echo signal is further extracted. The characteristic index is a measure that describes the signal characteristics and is used to represent the time-frequency characteristics, phase information, etc. of the signal. Then, a set of characteristic indexes is compared with the other two sets of characteristic indexes to determine the constant index group and the variable index group. The constant index group represents those features that remain unchanged in different echo signals, while the variable index group reflects the changing part of the signal, which helps to further distinguish noise and target information. The characteristic index extraction can be based on technical means such as time-frequency analysis, Fourier transform, and waveform matching. This extracts key information from complex echo signals and helps identify stable and changing signal parts. In S400, a characteristic vector of the signal to be processed is generated based on the constant index group and the variable index group. The characteristic vector is a mathematical representation of the signal characteristics and may include information such as frequency, phase, and amplitude. The characteristic vector is compared with the characteristic vector in the historical data group in order to determine whether the compensation value can be determined based on the empirical value of the historical data by comparing the similarity. Similarity measurement methods, such as Euclidean distance and cosine similarity, determine the matching degree between the current signal and the historical data, so as to select the most suitable compensation strategy. If the similarity between the feature vector in the historical data group and the feature vector of the current signal is greater than the preset similarity threshold, the compensation value in the historical data is directly used. If the similarity is lower than the threshold, the next step is the maximum likelihood estimation and convolutional neural network model prediction.

[0064] Specifically, the historical data set includes historical feature vectors and historical compensation values. When the historical feature vectors in the historical data set have a low similarity with the feature vectors of the current signal, the maximum likelihood estimation is used to determine the first compensation value. Maximum likelihood estimation is a statistical method that estimates the most likely compensation value by minimizing the phase error. At the same time, a convolutional neural network model is used to learn the compensation law of the signal from historical data, and the second compensation value is output according to the deep learning model.

[0065] It is understandable that the phase error problem in ISAR imaging of maritime targets under low signal-to-noise ratio conditions is effectively solved through a multi-level, adaptive compensation mechanism. In the case of high signal-to-noise ratio, the traditional Doppler frequency compensation method is used to quickly complete the compensation. In the low signal-to-noise ratio environment, an adaptive compensation strategy is provided by extracting signal features, comparing with historical data, and combining maximum likelihood estimation and convolutional neural network models. It enhances the robustness and adaptability under complex sea surface conditions, and also improves the imaging quality under low signal-to-noise ratio conditions, making target imaging clearer and able to accurately identify and locate maritime targets.

[0066] In some embodiments of the present application, when performing noise processing on the echo signal to obtain the signal to be processed, it includes: using Daubechies 4 wavelets to perform discrete wavelet transform, performing 4-layer wavelet decomposition on the echo signal, and obtaining approximation coefficients and detail coefficients:

[0067] The soft threshold method is used to perform threshold denoising on the detail coefficients, and the threshold is set to 1.5 times the standard deviation of the signal noise.

[0068] The denoised coefficients are used to perform inverse wavelet transform, reconstruct the denoised signal, and obtain the signal to be processed.

[0069] Specifically, the wavelet transform decomposes the signal into multiple layers to obtain a set of coefficients, including approximation coefficients (low-frequency components) and detail coefficients (high-frequency components). The approximation coefficients capture the low-frequency information of the signal, while the detail coefficients capture the high-frequency information of the signal (related to noise). During the wavelet decomposition process, the detail coefficients of the signal usually contain noise information, so the detail coefficients need to be processed by denoising methods. Each detail coefficient is thresholded, and the threshold is set to 1.5 times the standard deviation of the signal noise. The noise components smaller than the threshold are effectively removed while retaining the larger coefficients (i.e., the true characteristics of the signal). The standard deviation of the noise can be estimated by the statistical characteristics of the detail coefficients of the first few wavelet decomposition layers.

[0070] It is understandable that the wavelet denoising method does not require prior knowledge of the specific characteristics of the signal, but performs adaptive processing based on the local characteristics of the signal itself, so it is suitable for echo signal denoising in various complex environments. The soft threshold denoising method can effectively retain signal information while removing noise components. The threshold is set to 1.5 times the standard deviation of the noise, which can significantly reduce the impact of noise without losing important signal features.

[0071] In some embodiments of the present application, when determining a signal-to-noise ratio according to a signal to be processed, and determining a compensation scheme for the signal to be processed according to the signal-to-noise ratio, the method includes:

[0072] The square mean of the signal to be processed is calculated to obtain the power of the signal to be processed.

[0073] The echo signal of the target-free silent area is collected, and the power of the echo signal of the target-free silent area is calculated to obtain the noise power.

[0074] The signal-to-noise ratio is obtained using the following formula:

[0075]

[0076] Wherein, SNR represents signal-to-noise ratio, P represents the power of the signal to be processed, and P0 represents the noise power.

[0077] Specifically, the signal-to-noise ratio is compared with the signal-to-noise ratio threshold, and the compensation scheme of the signal to be processed is determined according to the comparison result. When the signal-to-noise ratio is higher than or equal to the signal-to-noise ratio threshold, the Doppler frequency is used for phase compensation. When the signal-to-noise ratio is lower than the signal-to-noise ratio threshold, the characteristic index of each echo signal in the signal to be processed is extracted, and one group of characteristic indexes is compared with the other two groups to determine the constant index group and the variable index group, and the characteristic vector of the signal to be processed is obtained according to the constant index group and the variable index group, and the compensation value is determined according to the characteristic vector.

[0078] It is understandable that by calculating the signal-to-noise ratio and determining the compensation scheme based on the signal-to-noise ratio, the scheme can dynamically adjust the compensation strategy according to the actual signal quality, thereby achieving more optimized phase compensation under different signal-to-noise ratio conditions. For situations with high signal-to-noise ratios, the Doppler frequency compensation method is used to quickly complete the compensation operation and ensure imaging accuracy. For situations with low signal-to-noise ratios, the method of feature extraction and historical data comparison is combined with maximum likelihood estimation and convolutional neural network models to accurately calculate the compensation value in a noisy environment and reduce the impact of noise on imaging quality. By adaptively selecting compensation schemes, it is possible to effectively cope with noise interference in complex environments and improve imaging quality and target detection accuracy under low signal-to-noise ratios.

[0079] In some embodiments of the present application, when the Doppler frequency is used for phase compensation, it includes: determining the relative speed of the sea target and the working frequency of the radar to obtain the Doppler frequency, and calculating the phase error according to the Doppler frequency.

[0080] The phase error is used as the compensation value, and the compensation value is calculated by the following formula:

[0081]

[0082] in, represents the compensation value, V represents the relative speed of the sea target, represents the wavelength of the radar signal, represents the center frequency of the radar, and t represents the signal sampling time.

[0083] In some embodiments of the present application, when determining the compensation value according to the feature vector, it includes: when there is data in the historical data group whose similarity with the feature vector is greater than the similarity threshold, determining the compensation value according to the historical compensation value in the historical data group. When the similarity between the historical feature vector in the historical data group and the feature vector is less than or equal to the similarity threshold, determining the first compensation value based on the maximum likelihood phase error, determining the second compensation value based on the convolutional neural network model, and determining the compensation value according to the first compensation value and the second compensation value.

[0084] In some embodiments of the present application, when determining the compensation value according to the historical compensation value in the historical data group, it includes: constructing the data in the historical data group whose similarity with the feature vector is greater than the similarity threshold as a similarity data set. The similarity data set includes the historical feature vector and the historical compensation value. Obtain the mean of the historical compensation value according to the similarity data set, and split the similarity data set according to the mean of the historical compensation value.

[0085] Specifically, data whose historical compensation value is greater than the mean historical compensation value is classified as the first similar data set. Data whose historical compensation value is equal to the mean historical compensation value is classified as the second similar data set. Data whose historical compensation value is less than the mean historical compensation value is classified as the third similar data set. The compensation value is determined based on the mean historical compensation value and the first similar data set and the third similar data set.

[0086] In some embodiments of the present application, when the compensation value is determined according to the historical compensation value average and the first similar data set and the third similar data set, the compensation value is calculated by the following formula:

[0087]

[0088] in, represents the compensation value, N is the number of historical compensation values ​​in the first similar data set, is the i-th historical compensation value in the first similar data set, represents the mean value of historical compensation values, M is the number of historical compensation values ​​in the third similar data set, is the mth historical compensation value in the third similar data set.

[0089] It is understandable that by calculating the similarity and grouping the historical data, the compensation value of the current signal can be inferred based on the historical compensation value, avoiding the limitation of relying solely on the current signal. The compensation accuracy is improved and the adaptive ability is also improved. The compensation value is calculated by weighted average of similar data sets, making the calculation of the compensation value more robust. In the case of low signal-to-noise ratio, the compensation value can be effectively calculated using historical data to reduce errors caused by noise or interference.

[0090] In some embodiments of the present application, when the first compensation value is determined based on the maximum likelihood phase error, the expression is:

[0091]

[0092] in, represents the first compensation value, represents a given phase error, Indicates a signal to be processed. represents an ideal echo signal without phase error, and j represents an imaginary unit.

[0093] Specifically, the phase compensation is determined by calculating the phase error between the signal to be processed and the ideal echo signal. The ideal echo signal is the signal when it is assumed that there is no phase error, while the signal to be processed contains phase errors introduced by factors such as motion and noise. Given the difference between the signal to be processed and the ideal echo signal without phase error, we can define the maximum likelihood phase error. Since the signal is usually represented in complex form in the frequency domain, the phase error compensation is also calculated in the complex domain. Specifically, the difference between the signal to be processed and the ideal echo signal is measured in complex form. By minimizing the quantization loss function of the phase error (signal difference), a numerical optimization method (such as gradient descent or direct search) is used to obtain the phase compensation value.

[0094] It can be understood that the maximum likelihood estimation can estimate the phase error optimally in a statistical sense. Compared with the traditional phase estimation method, it can more accurately compensate for the phase error in the echo signal and reduce the impact of the error. The phase compensation is calculated based on the actual received echo signal, which has strong adaptability and can cope with the phase error under different signal environments and noise conditions. By converting the phase error into an optimization problem of minimizing the signal error, it can effectively handle the complex noise influence and provide more stable phase compensation. Through accurate phase compensation, the quality of the signal to be processed can be significantly improved, thereby improving the resolution and accuracy of ISAR imaging.

[0095] It can be understood that the convolutional neural network is trained using the echo signals and their corresponding compensation values ​​in the historical data set, so that the network can automatically extract useful information from the features of the echo signals and learn the mapping relationship between the echo signals and the compensation values. After the training is completed, given the feature vector of the signal to be processed, the convolutional neural network model inputs it into the network through forward propagation and generates the corresponding compensation value output. This process can not only identify complex signal patterns, but also effectively deal with a variety of noise interference situations and provide prediction values ​​for phase compensation.

[0096] In some embodiments of the present application, when determining the compensation value based on the first compensation value and the second compensation value, it includes: obtaining a compensation value difference based on the first compensation value and the second compensation value, the compensation value difference is the absolute value of the difference between the first compensation value and the second compensation value, and determining the compensation value based on the compensation value difference.

[0097] Specifically, when the compensation value difference is less than a% of the first compensation value, and the compensation value difference is less than a% of the second compensation value, the mean compensation value is obtained according to the first compensation value and the second compensation value, and the mean compensation value is used as the compensation value. When the compensation value difference is greater than or equal to a% of the first compensation value, or the compensation value difference is greater than or equal to a% of the second compensation value, the adjustment coefficient is determined according to the compensation value difference to adjust the first compensation value and determine the compensation value.

[0098] In some embodiments of the present application, the adjustment coefficient is determined according to the compensation value difference to adjust the first compensation value, and when the compensation value is determined, it includes: when the first compensation value is greater than the second compensation value, the adjustment coefficient is determined according to the compensation value difference to adjust the first compensation value, and the compensation value is determined, and the adjustment coefficient is inversely proportional to the compensation value difference. When the first compensation value is less than the second compensation value, the adjustment coefficient is determined according to the compensation value difference to adjust the first compensation value, and the compensation value is determined, and the adjustment coefficient is directly proportional to the compensation value difference.

[0099] Specifically, for example, we perform ISAR imaging of sea targets in a synthetic aperture radar (SAR) system. In the compensation value determination stage of the processing process, the first compensation value and the second compensation value have been calculated according to the maximum likelihood estimation and convolutional neural network model. The first compensation value is 3 radians, and the second compensation value is 2.5 radians. The difference threshold a is 10% (indicating 10% of the first compensation value and the second compensation value). The difference in compensation value is 0.5 radians, the a% of the first compensation value is 0.3, and the a% of the second compensation value is 0.25. The difference in compensation value is large, so an adjustment scheme is adopted. Because the first compensation value is greater than the second compensation value, a reduction coefficient is selected to adjust the first compensation value. The range of the reduction coefficient is [0.8, 1), and the range of the increase coefficient is (1, 1.2]. For example, when the difference in compensation is 0.5 radians, the reduction coefficient is selected as 0.9, so the compensation value is 3*0.9=2.7 radians.

[0100] It can be understood that by dynamically adjusting the compensation value based on the difference, whether the mean compensation is selected or the adjustment coefficient is used for correction, the accuracy and stability of the compensation result can be effectively improved. The calculation strategy is automatically adjusted according to the size of the compensation value difference, which enhances the adaptive ability under different signal-to-noise ratio conditions and reduces the impact of noise interference. By adjusting the compensation coefficient, the compensation value can be flexibly adjusted when the compensation value difference is large, making the entire compensation process more robust and reliable, which is suitable for ISAR imaging of sea targets in low signal-to-noise ratio environments.

[0101] In the above embodiment, different compensation strategies are dynamically selected through signal-to-noise ratio evaluation, ensuring that the traditional Doppler frequency compensation method can be quickly adopted in the case of high signal-to-noise ratio, and accurate compensation is performed by extracting signal features, comparing with historical data, and combining maximum likelihood estimation and convolutional neural network model in the case of low signal-to-noise ratio. The compensation strategy is automatically adjusted according to different environments and signal characteristics, which enhances the adaptability and robustness, improves the phase compensation accuracy in the case of low signal-to-noise ratio, and is conducive to obtaining clearer and more accurate target information in the imaging of complex sea surfaces and dynamic targets.

[0102] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0103] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0104] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0105] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0106] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A low signal-to-noise ratio phase compensation algorithm for synthetic aperture radar (ISAR) imaging of sea targets, characterized in that: include: Transmitting a radar signal to a sea target through a synthetic aperture radar system, cyclically collecting echo signals within a preset time period at least three times when the sea target moves in the same direction, selecting three sections of the echo signals for noise processing to obtain a signal to be processed; Determining a signal-to-noise ratio according to the signal to be processed, and determining a compensation scheme for the signal to be processed according to the signal-to-noise ratio; When the signal-to-noise ratio is lower than the signal-to-noise ratio threshold, extracting characteristic indexes of each echo signal in the signal to be processed, comparing one set of characteristic indexes with the other two sets, and determining a constant index group and a variable index group; Obtaining a characteristic vector of the signal to be processed according to the constant index group and the variable index group, comparing the characteristic vector with a historical data group, and determining a compensation value according to the comparison result; When the similarity between the historical feature vector in the historical data group and the feature vector is less than or equal to the similarity threshold, a first compensation value is determined based on the maximum likelihood phase error, a second compensation value is determined based on the convolutional neural network model, and the compensation value is determined according to the first compensation value and the second compensation value; The historical data set includes historical feature vectors and historical compensation values.

2. The low signal-to-noise ratio phase compensation algorithm for synthetic aperture radar (ISAR) imaging of sea targets according to claim 1 is characterized in that: When performing noise processing on the echo signal to obtain a signal to be processed, it includes: The Daubechies 4 wavelet is used for discrete wavelet transform, and the echo signal is decomposed by 4 layers of wavelets to obtain the approximation coefficient and detail coefficient: The soft threshold method is used to perform threshold denoising on the detail coefficients, and the threshold is set to 1.5 times the standard deviation of the signal noise; The denoised coefficients are used to perform inverse wavelet transform, and the denoised signal is reconstructed to obtain the signal to be processed.

3. The low signal-to-noise ratio phase compensation algorithm for synthetic aperture radar (ISAR) imaging of sea targets according to claim 1 is characterized in that: Determining a signal-to-noise ratio according to the signal to be processed, and determining a compensation scheme for the signal to be processed according to the signal-to-noise ratio, includes: Calculating the square mean of the signal to be processed to obtain the power of the signal to be processed; Collecting echo signals from a target-free silent area, and calculating the power of the echo signals from the target-free silent area to obtain noise power; The signal-to-noise ratio is obtained using the following formula: Among them, SNR represents signal-to-noise ratio, P represents the power of the signal to be processed, and P0 represents the noise power; Comparing the signal-to-noise ratio with a signal-to-noise ratio threshold, and determining a compensation scheme for the signal to be processed according to the comparison result; When the signal-to-noise ratio is higher than or equal to the signal-to-noise ratio threshold, using Doppler frequency for phase compensation; When the signal-to-noise ratio is lower than the signal-to-noise ratio threshold, the characteristic index of each echo signal in the signal to be processed is extracted, one group of characteristic indexes is compared with the other two groups, a constant index group and a variable index group are determined, a characteristic vector of the signal to be processed is obtained according to the constant index group and the variable index group, and a compensation value is determined according to the characteristic vector.

4. The low signal-to-noise ratio phase compensation algorithm for synthetic aperture radar (ISAR) imaging of sea targets according to claim 3 is characterized in that: When using Doppler frequency for phase compensation, it includes: Determine the relative speed of the sea target and the operating frequency of the radar to obtain a Doppler frequency, and calculate a phase error according to the Doppler frequency; The phase error is used as the compensation value, and the compensation value is obtained by calculating the following formula: in, represents the compensation value, V represents the relative speed of the sea target, represents the wavelength of the radar signal, represents the center frequency of the radar, and t represents the signal sampling time.

5. The low signal-to-noise ratio phase compensation algorithm for synthetic aperture radar (ISAR) imaging of sea targets according to claim 3 is characterized in that: When determining the compensation value according to the characteristic vector, it includes: When there is data in the historical data group whose similarity with the feature vector is greater than a similarity threshold, determining the compensation value according to the historical compensation values ​​in the historical data group; When the similarity between the historical feature vector in the historical data group and the feature vector is less than or equal to the similarity threshold, a first compensation value is determined based on the maximum likelihood phase error, and a second compensation value is determined based on the convolutional neural network model, and the compensation value is determined according to the first compensation value and the second compensation value.

6. The low signal-to-noise ratio phase compensation algorithm for synthetic aperture radar (ISAR) imaging of sea targets according to claim 5 is characterized in that: When the compensation value is determined according to the historical compensation value in the historical data group, it includes: Constructing data in the historical data group whose similarity to the feature vector is greater than a similarity threshold into a similarity data set; the similarity data set includes a historical feature vector and a historical compensation value; Acquire a mean value of historical compensation values ​​according to the similar data set, and split the similar data set according to the mean value of historical compensation values; Classifying the data whose historical compensation value is greater than the average historical compensation value as a first similar data set; Classifying the data whose historical compensation value is equal to the average of the historical compensation values ​​as a second similar data set; Classifying the data whose historical compensation value is less than the average historical compensation value as a third similar data set; The compensation value is determined according to the historical compensation value average and the first similar data set and the third similar data set.

7. The low signal-to-noise ratio phase compensation algorithm for synthetic aperture radar (ISAR) imaging of sea targets according to claim 6 is characterized in that: When the compensation value is determined according to the historical compensation value average and the first similar data set and the third similar data set, the compensation value is calculated by the following formula: in, represents the compensation value, N is the number of historical compensation values ​​in the first similar data set, is the i-th historical compensation value in the first similar data set, represents the mean value of historical compensation values, M is the number of historical compensation values ​​in the third similar data set, is the mth historical compensation value in the third similar data set.

8. The low signal-to-noise ratio phase compensation algorithm for synthetic aperture radar (ISAR) imaging of sea targets according to claim 5 is characterized in that: When the first compensation value is determined based on the maximum likelihood phase error, the expression is: in, represents the first compensation value, represents a given phase error, Indicates a signal to be processed. represents an ideal echo signal without phase error, and j represents an imaginary unit.

9. The low signal-to-noise ratio phase compensation algorithm for synthetic aperture radar (ISAR) imaging of sea targets according to claim 8 is characterized in that: When determining the compensation value according to the first compensation value and the second compensation value, it includes: Obtaining a compensation value difference according to the first compensation value and the second compensation value, wherein the compensation value difference is an absolute value of a difference between the first compensation value and the second compensation value, and determining the compensation value according to the compensation value difference; When the compensation value difference is less than a% of the first compensation value, and the compensation value difference is less than a% of the second compensation value, obtaining a mean compensation value according to the first compensation value and the second compensation value, and using the mean compensation value as the compensation value; When the compensation value difference is greater than or equal to a% of the first compensation value, or the compensation value difference is greater than or equal to a% of the second compensation value, the first compensation value is adjusted according to an adjustment coefficient determined according to the compensation value difference to determine the compensation value.

10. The low signal-to-noise ratio phase compensation algorithm for synthetic aperture radar (ISAR) imaging of sea targets according to claim 9 is characterized in that: Determining an adjustment coefficient according to the compensation value difference to adjust the first compensation value, and determining the compensation value includes: When the first compensation value is greater than the second compensation value, the first compensation value is adjusted by determining a reduction coefficient according to the compensation value difference to determine the compensation value, and the reduction coefficient is inversely proportional to the compensation value difference; When the first compensation value is less than the second compensation value, an increase coefficient is determined according to the compensation value difference to adjust the first compensation value to determine the compensation value, and the increase coefficient is proportional to the compensation value difference.

Images