Single-channel SAR deception jamming identification method and system based on feature graph learning
Through feature map learning method, SAR echo signals are processed, the characteristics of interference targets are extracted and enhanced, the real target information is suppressed, and the neural network is used to identify spoof interference targets, which solves the problem of difficulty in identifying near-range interference targets in the existing technology, and improves the problem of Identify accuracy.
Patent Information
- Application Number
- CN202111411377.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-11-25
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2041-11-25
AI Technical Summary
The prior art is difficult to effectively identify close-range fraud interference targets, resulting in the interference target being close to the real target features and difficulty in identifying them.
Using a method based on feature map learning, the distance-to-match filtering and orientation de-abercing are performed through the echo signal received by the receiver, real scenes and spoof interference features are extracted, normalized adjustments and feature enhancement suppression are performed, and RGB images are generated. The input neural network is trained to identify the interference target.
It enhances the characteristic obviousness of the interference target, suppresses the real target information, and improves the accuracy of spoofing interference identification, especially when the interference target is close to the jammer.
Smart Images

Figure CN114120118B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of computer processing technology, and in particular to a single-channel SAR deception interference identification method, system, terminal and computer-readable storage medium based on feature map learning. Background Art
[0002] SAR (Synthetic Aperture Radar, Synthetic Aperture Radar, is an active earth observation system that can be installed on aircraft, satellites, spacecraft and other flying platforms to observe the earth all day and all weather, and has a certain surface penetration ability) deception interference is to detect the key parameters of the SAR signal such as the center frequency, modulation frequency, bandwidth, etc., and interfere with the SAR system by means of interference modulation to simulate radar echoes or echo forwarding, and has the characteristics of strong flexibility, low interference power, and high fidelity. SAR deception interference makes the false interference signal and the real echo signal received by the real SAR system overlap with each other, and the real target and the set false target are overlapped in the imaging result, achieving the imaging effect of "false and real", making it difficult to identify the real target, and then making wrong decisions. Therefore, the identification of deception interference is particularly important.
[0003] With the improvement of deception jamming technology, the characteristics of jamming targets and real targets are getting closer and closer, which makes it more difficult to identify jamming targets. The recognition effect of deception jamming is related to the position of the jamming target in azimuth from the jammer. The closer the jamming target is to the jammer, the less obvious its characteristics are and the more difficult it is to identify.
[0004] Therefore, the prior art still needs to be improved and developed. Summary of the invention
[0005] The main purpose of the present invention is to provide a single-channel SAR deception interference identification method, system, terminal and computer-readable storage medium based on feature map learning, aiming to solve the problem in the prior art that the features of the interference target and the real target are getting closer and closer, and the identification of the interference target is becoming more and more difficult.
[0006] To achieve the above object, the present invention provides a single-channel SAR deception interference identification method based on feature map learning, and the single-channel SAR deception interference identification method based on feature map learning comprises the following steps:
[0007] The echo signals received by the receiver are processed by matched filtering in range and de-skewing in azimuth respectively to obtain all the features of the real scene features and the deceptive interference features;
[0008] Normalizing and adjusting the real scene feature and the deception interference feature;
[0009] Suppressing the deceptive interference feature, enhancing the real scene feature, and obtaining the real feature;
[0010] Enhance the deceptive interference feature and suppress the real scene feature to obtain an interference feature;
[0011] All the features obtained after normalization are multiplied by the first coefficient and input into the first feature channel, the real features are multiplied by the second coefficient and input into the second feature channel, the interference features are multiplied by the third coefficient and input into the third feature channel, and the first feature channel, the second feature channel and the third feature channel are respectively mapped to the R, G and B channels of the RGB color space to generate an RGB image;
[0012] The obtained RGB image is input into the neural network for training, and the weights obtained from the training are used to identify the interference target.
[0013] The single-channel SAR deception interference identification method based on feature graph learning, wherein the echo signal received by the receiver is respectively subjected to range matched filtering processing and azimuth de-skewing processing to obtain all the features of the real scene and the deception interference, specifically includes:
[0014] The received echo signal is processed by range matching filtering, and then the inverse Fourier transform is performed to obtain the processed signal s r (t r ,t a ), where t r is the distance to time, t a is the azimuth time;
[0015] For r (t r ,t a ) performs azimuth de-skewing processing, and then performs azimuth Fourier transform to obtain the signal S r (t r ,f a ), where f a is the Doppler frequency, and its expression is as follows:
[0016] S r (t r ,f a )=F{s r (t r ,t a )s0(t a )};
[0017] Among them, s0(t a ) is the de-skewed reference signal, F(·) represents the Fourier transform, S r (t r ,f a) represents all features including real scenes and deceptive interference.
[0018] The single-channel SAR deception interference identification method based on feature graph learning, wherein the normalization adjustment of the real scene features and the deception interference features specifically includes:
[0019] For S r (t r ,f a ) is normalized, the expression is as follows:
[0020]
[0021] Where abs[·] means taking the absolute value, max{·} means taking the maximum value, S TF (t r ,f a ) represents all the features of the real scene and deceptive interference after normalization.
[0022] The single-channel SAR deception interference identification method based on feature graph learning, wherein the deception interference feature is suppressed, the real scene feature is enhanced, and the real feature is obtained, specifically including:
[0023] The echo signal after the range-direction matched filtering is de-skewed in azimuth and windowed to suppress the real scene features. The expression is as follows:
[0024] S J (t r ,f a )=F{ω(t a )s r (t r ,t a )s0(t a )};
[0025] Among them, ω(t a ) represents a rectangular window, de-skewing the reference signal s0(t a ) covers the entire temporal distribution of all targets;
[0026] Suppress the deceptive interference features and enhance the real scene features to obtain the real features:
[0027]
[0028] Where α is a constant greater than 0, and the numerator in the formula is expressed as normalized and then raised to the power of α, S J (t r ,f a ) is the signal after suppressing the real scene, in which case the real scene feature is a strong feature and the deception interference feature is a weak feature;
[0029] When the exponent α∈(0,1), the deceptive interference feature is relatively enhanced, and the real scene feature is relatively weakened; when α∈(1,+∞), the real scene feature is relatively enhanced, and the deceptive interference feature is relatively weakened;
[0030] The normalized S r (t r ,f a ) is put into the denominator of the fraction so that the real scene features are enhanced.
[0031] The single-channel SAR deception interference identification method based on feature graph learning, wherein the enhancing the deception interference feature, suppressing the real scene feature, and obtaining the interference feature specifically includes:
[0032] A bandpass filter is added to the echo signal after the range-direction matched filtering in the Doppler domain, and then the azimuth de-skewing is performed to suppress the interference target. The expression is as follows:
[0033] S T (t r ,f a )=F{F -1 {F[s(t r ,t a )]H f (f a )}s0(t a )};
[0034] Among them, F -1 {·} represents the inverse Fourier transform, H f (f a ) is a bandpass filter;
[0035] Suppress the real scene features and enhance the deception interference features to obtain the interference features:
[0036]
[0037] Where β is a constant greater than 0, and the numerator in the formula represents the normalization and then taking the β power, S T (t r ,f a ) is the signal after suppressing the interference target, in which case the deception interference feature is a strong feature and the real scene feature is a weak feature;
[0038] When the exponent β∈(0,1), the real scene features are relatively enhanced, and the deception interference features are relatively weakened; when β∈(1,+∞), the deception interference features are relatively enhanced, and the real scene features are relatively weakened;
[0039] The normalized S r(t r ,f a ) is put into the denominator of the fraction so that the interference target features are enhanced.
[0040] The single-channel SAR deception interference identification method based on feature map learning, wherein all the features obtained after normalization are multiplied by the first coefficient and then input into the first feature channel, the real features are multiplied by the second coefficient and then input into the second feature channel, and the interference features are multiplied by the third coefficient and then input into the third feature channel, specifically includes:
[0041] S TF (t r ,f a ) multiplied by the first coefficient k1 to obtain: S R (t r ,f a )=k1·S TF (t r ,f a ), S R (t r ,f a ) is input to the first feature channel;
[0042] S GJ (t r ,f a ) multiplied by the second coefficient k2 gives: S G (t r ,f a )=k2·S GJ (t r ,f a ), S G (t r ,f a ) is input to the second feature channel;
[0043] S BT (t r ,f a ) multiplied by the third coefficient k3 to obtain: S B (t r ,f a )=k3·S BT (t r ,f a ), S B (t r ,f a ) is input to the third feature channel;
[0044] Among them, k1, k2 and k3 are the weights of the corresponding feature channels, which are used to adjust the strength of the features and are all constants ranging from 0 to 255.
[0045] The single-channel SAR deception interference identification method based on feature map learning, wherein the obtained RGB image is input into the neural network for training, specifically:
[0046] The simulated RGB image is input into the neural network for training using deep learning method.
[0047] In addition, to achieve the above-mentioned purpose, the present invention also provides a single-channel SAR deception interference identification system based on feature map learning, wherein the single-channel SAR deception interference identification system based on feature map learning includes:
[0048] The signal processing module is used to perform matched filtering processing in the range direction and de-skewing processing in the azimuth direction on the echo signal received by the receiver, so as to obtain all the features of the real scene features and the deception interference features;
[0049] A normalization processing module, used for normalizing and adjusting the real scene features and the deception interference features;
[0050] A first feature processing module, used to suppress the deceptive interference feature, enhance the real scene feature, and obtain the real feature;
[0051] A second feature processing module is used to enhance the deceptive interference feature, suppress the real scene feature, and obtain the interference feature;
[0052] A feature image generation module, which multiplies all features obtained after normalization by a first coefficient and inputs the result to a first feature channel, multiplies the real feature by a second coefficient and inputs the result to a second feature channel, multiplies the interference feature by a third coefficient and inputs the result to a third feature channel, and maps the first feature channel, the second feature channel and the third feature channel to the R, G and B channels of the RGB color space, respectively, to generate an RGB image;
[0053] The training recognition module is used to input the obtained RGB image into the neural network for training, and use the weights obtained from the training to identify the interference target.
[0054] In addition, to achieve the above-mentioned purpose, the present invention also provides a terminal, wherein the terminal includes: a memory, a processor, and a single-channel SAR deception interference identification program based on feature map learning stored in the memory and run on the processor, and when the single-channel SAR deception interference identification program based on feature map learning is executed by the processor, the steps of the single-channel SAR deception interference identification method based on feature map learning as described above are implemented.
[0055] In addition, to achieve the above-mentioned objectives, the present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a single-channel SAR deception interference identification program based on feature map learning, and when the single-channel SAR deception interference identification program based on feature map learning is executed by a processor, the steps of the single-channel SAR deception interference identification method based on feature map learning as described above are implemented.
[0056] The present invention obtains all features of real scene features and deception interference features by performing distance matching filtering and azimuth de-skewing processing on the echo signals received by the receiver respectively; normalizes and adjusts the real scene features and the deception interference features; suppresses the deception interference features, enhances the real scene features, and obtains real features; enhances the deception interference features, suppresses the real scene features, and obtains interference features; multiplies all features obtained after normalization by a first coefficient and inputs them into a first feature channel, multiplies the real features by a second coefficient and inputs them into a second feature channel, multiplies the interference features by a third coefficient and inputs them into a third feature channel, and maps the first feature channel, the second feature channel, and the third feature channel to R, G, and B channels of the RGB color space respectively to generate an RGB image; inputs the obtained RGB image into a neural network for training, and uses the weights obtained by the training to identify the interference target. The present invention not only enhances the features of the interference target, but also suppresses the real target information by performing different processing on the echo signals and putting them into different feature channels, so that the features of the interference target are more obvious, and finally identifies the interference target through a neural network to achieve the purpose of deception interference identification. BRIEF DESCRIPTION OF THE DRAWINGS
[0057] Figure 1 It is a flow chart of a preferred embodiment of a single-channel SAR deception interference identification method based on feature graph learning of the present invention;
[0058] Figure 2 It is a schematic diagram of the principle of the processing process of a preferred embodiment of the single-channel SAR deception interference identification method based on feature graph learning of the present invention;
[0059] Figure 3 It is an imaging result diagram of adding interference targets in a simulation experiment in a preferred embodiment of the single-channel SAR deception interference identification method based on feature graph learning of the present invention;
[0060] Figure 4 It is a recognition result diagram of the jammer's interference target being far away from the jammer in a preferred embodiment of the single-channel SAR deception jamming identification method based on feature graph learning of the present invention;
[0061] Figure 5It is the recognition result of the jammer target approaching the jammer in the preferred embodiment of the single-channel SAR deception jamming identification method based on feature graph learning of the present invention;
[0062] Figure 6 It is a statistical result diagram of a preferred embodiment of the single-channel SAR deception interference identification method based on feature graph learning of the present invention;
[0063] Figure 7 It is a schematic diagram of the principle of a preferred embodiment of a single-channel SAR deception interference identification system based on feature graph learning of the present invention;
[0064] Figure 8 Schematic diagram of the operating environment of a preferred embodiment of the terminal of the present invention. DETAILED DESCRIPTION
[0065] In order to make the purpose, technical solution and advantages of the present invention clearer and more specific, the present invention is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.
[0066] The single-channel SAR deception interference identification method based on feature graph learning described in the preferred embodiment of the present invention is as follows: Figure 1 and Figure 2 As shown, the single-channel SAR deception interference identification method based on feature graph learning includes the following steps:
[0067] Step S10: Perform range-wise matched filtering and azimuth-wise de-skewing processing on the echo signal received by the receiver to obtain all features of the real scene features and the deceptive interference features.
[0068] Specifically, the echo signal received by the receiver is processed by range matching filtering, and then the processed signal s is obtained by inverse Fourier transform. r (t r ,t a ), where t r is the distance time (fast time), t a is the azimuth time (slow time); for s r (t r ,t a ) performs azimuth de-skewing processing, and then performs azimuth Fourier transform to obtain the signal S r (t r ,f a ), where f a is the Doppler frequency, and its expression is as follows:
[0069] S r (t r ,f a )=F{sr (t r ,t a )s0(t a )};
[0070] Among them, s0(t a ) is the de-skewed reference signal, F(·) represents the Fourier transform, S r (t r ,f a ) represents all features including real scenes and deceptive interference.
[0071] Step S20: normalize and adjust the real scene features and the deception interference features.
[0072] Specifically, for S r (t r ,f a ) is normalized, the expression is as follows:
[0073]
[0074] Where abs[·] means taking the absolute value, max{·} means taking the maximum value, S TF (t r ,f a ) represents all the features of the real scene and deceptive interference after normalization.
[0075] Step S30: suppress the deceptive interference features, enhance the real scene features, and obtain real features.
[0076] Specifically, the echo signal after the range-direction matched filtering process (step 10) is de-skewed in azimuth and windowed to suppress the real scene features. The expression is as follows:
[0077] S J (t r ,f a )=F{ω(t a )s r (t r ,t a )s0(t a )};
[0078] Among them, ω(t a ) represents a rectangular window, de-skewing the reference signal s0(t a ) covers the entire time domain distribution of all targets. Since the jammer itself can only be illuminated by the radar within the synthetic aperture range, after windowing, the jamming target characteristics are not affected, while the real scene characteristics are weakened;
[0079] Then do the following processing to get S GJ (tr ,f a ), that is, suppressing the deceptive interference features and enhancing the real scene features to obtain the real features:
[0080]
[0081] Where α is a constant greater than 0, and the numerator in the formula is expressed as normalized and then raised to the power of α, S J (t r ,f a ) is the signal after suppressing the real scene. After subtraction, the interference target features can be greatly suppressed while retaining the real scene features. At this time, the real scene features are strong features and the deceptive interference features are weak features.
[0082] When the exponent α∈(0,1), the deceptive interference feature (weak feature) is relatively enhanced, while the real scene feature (strong feature) is relatively weakened; when α∈(1,+∞), the real scene feature (strong feature) is relatively enhanced, while the deceptive interference feature (weak feature) is relatively weakened; when it is 1, there is no change.
[0083] The normalized S r (t r ,f a ) into the denominator of the fraction, the real scene features will be enhanced to a certain extent.
[0084] Step S40: enhance the deceptive interference feature, suppress the real scene feature, and obtain the interference feature.
[0085] Specifically, a bandpass filter is added to the echo signal after the range-direction matched filtering process (step 10) in the Doppler domain, and then a de-skewing process is performed in the azimuth direction to suppress the interference target. The expression is as follows:
[0086] S T (t r ,f a )=F{F -1 {F[s(t r ,t a )]H f (f a )}s0(t a )};
[0087] Among them, F -1 {·} represents the inverse Fourier transform, H f (f a ) is a bandpass filter; after adding the bandpass filter in the frequency domain, the interference target is suppressed, while the real target is largely retained.
[0088] Next, do the following processing to get S BT(t r ,f a ), that is, suppressing the real scene features and enhancing the deceptive interference features to obtain the interference features:
[0089]
[0090] Where β is a constant greater than 0, and the numerator in the formula represents the normalization and then taking the β power, S T (t r ,f a ) is the signal after suppressing the interference target. After subtraction, the real scene features can be greatly suppressed, while the interference target features are retained. At this time, the deceptive interference features are strong features, and the real scene features are weak features.
[0091] When the exponent β∈(0,1), the real scene features (weak features) are relatively enhanced, while the deceptive interference features (strong features) are relatively weakened; when β∈(1,+∞), the deceptive interference features (strong features) are relatively enhanced, while the real scene features (weak features) are relatively weakened; when it is 1, there is no change.
[0092] The normalized S r (t r ,f a ) into the denominator of the fraction, the interference target features will be enhanced to a certain extent.
[0093] Step S50, multiply all the features obtained after normalization by the first coefficient and input them into the first feature channel, multiply the real features by the second coefficient and input them into the second feature channel, multiply the interference features by the third coefficient and input them into the third feature channel, and map the first feature channel, the second feature channel and the third feature channel to the R, G and B channels of the RGB color space respectively to generate an RGB image.
[0094] Specifically, S TF (t r ,f a ) multiplied by the first coefficient k1 to obtain (i.e. Figure 2 Processing method 1): S R (t r ,f a )=k1·S TF (t r ,f a ), S R (t r ,f a ) is input to the first feature channel (i.e. Figure 2 The characteristic channel 1 in S GJ (t r ,f a ) multiplied by the second coefficient k2 gives (i.e. Figure 2 Processing method 2): S G (t r ,f a )=k2·S GJ (t r ,f a ), S G (t r ,f a ) is input to the second feature channel (i.e. Figure 2 The characteristic channel 2 in S BT (t r ,f a ) multiplied by the third coefficient k3 to obtain (i.e. Figure 2 Processing method 3): S B (t r ,f a )=k3·S BT (t r ,f a ), S B (t r ,f a ) is input to the third feature channel (i.e. Figure 2 3); wherein, k1, k2 and k3 are weights of the corresponding feature channels, which are used to adjust the strength of the features, and are all constants of 0-255; then feature channels 1 to 3 (i.e., the first feature channel, the second feature channel and the third feature channel) are respectively mapped to the R, G, and B channels of the RGB color space to form an RGB image (i.e., feature image); further, CMYK mode, LAB mode, etc. can be used to generate feature images.
[0095] Step S60: input the obtained RGB image into a neural network for training, and use the weights obtained from the training to identify interference targets.
[0096] Specifically, the simulated RGB image is input into the neural network for training using a deep learning method, and the weights obtained from the training are used to identify interference targets.
[0097] The present invention processes the obtained echo signals differently and puts them into different feature channels respectively, which not only enhances the characteristics of the interference target, but also suppresses the real target information, making the characteristics of the interference target more obvious and bringing convenience to the later identification.
[0098] Furthermore, the effect of the present invention can be further illustrated by the following simulation experiments. For example, MATLAB software is used for simulation, and the parameters of the simulation data are as follows: the jammer position is set to the center of the scene, k1 is set to 0, k2 is set to 1, k3 is set to 1, α is set to 0.07, and β is set to 2.9.
[0099] like Figure 3 As shown, Figure 3 A schematic diagram showing the state before recognition is performed. The image of the interference target is added. The background part is the scene information, and the white part is the interference target. Figure 3 It can be seen that the interference target has obvious characteristics. Figure 4 As shown, Figure 4 The following is a schematic diagram after recognition. You can see that the interference target is framed and labeled "false 1.00", indicating that the probability of the target being a false target is 1.00, and the recognition result is relatively ideal. Figure 5 As shown, Figure 5 This means that even if a false target approaches the jammer in azimuth, it can still be identified with a probability of more than 0.5. Figure 6 As shown, Figure 6 The P-Ad graph is obtained by averaging multiple groups of data, where P is the probability (confidence) of the jammer being identified, and Ad is the percentage of the distance of the jammer from the jammer in the azimuth direction to the synthetic aperture length. Figure 6 It can be seen that after statistical averaging, good recognition effect can still be obtained within ±5% of the synthetic aperture.
[0100] Furthermore, if Figure 7 As shown, based on the above-mentioned single-channel SAR deception interference identification method based on feature map learning, the present invention also provides a single-channel SAR deception interference identification system based on feature map learning, wherein the single-channel SAR deception interference identification system based on feature map learning includes:
[0101] The signal processing module 51 is used to perform range-wise matched filtering and azimuth-wise de-skewing processing on the echo signals received by the receiver to obtain all the features of the real scene features and the deception interference features;
[0102] A normalization processing module 52, used for normalizing and adjusting the real scene feature and the deception interference feature;
[0103] A first feature processing module 53, used to suppress the deceptive interference feature, enhance the real scene feature, and obtain the real feature;
[0104] A second feature processing module 54 is used to enhance the deceptive interference feature and suppress the real scene feature to obtain an interference feature;
[0105] The feature image generation module 55 multiplies all the features obtained after normalization by the first coefficient and inputs them into the first feature channel, multiplies the real features by the second coefficient and inputs them into the second feature channel, multiplies the interference features by the third coefficient and inputs them into the third feature channel, and maps the first feature channel, the second feature channel and the third feature channel to the R, G and B channels of the RGB color space respectively to generate an RGB image;
[0106] The training recognition module 56 is used to input the obtained RGB image into the neural network for training, and use the weights obtained from the training to identify the interference target.
[0107] Furthermore, if Figure 8 As shown, based on the above-mentioned single-channel SAR deception interference identification method and system based on feature graph learning, the present invention also provides a terminal accordingly, and the terminal includes a processor 10, a memory 20 and a display 30. Figure 8 Only some components of the terminal are shown, but it should be understood that it is not required to implement all of the components shown, and more or fewer components may be implemented instead.
[0108] In some embodiments, the memory 20 may be an internal storage unit of the terminal, such as a hard disk or memory of the terminal. In other embodiments, the memory 20 may also be an external storage device of the terminal, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the terminal. Further, the memory 20 may also include both an internal storage unit of the terminal and an external storage device. The memory 20 is used to store application software and various types of data installed in the terminal, such as the program code of the installation terminal. The memory 20 may also be used to temporarily store data that has been output or is to be output. In some embodiments, a single-channel SAR deception interference recognition program 40 based on feature map learning is stored on the memory 20, and the single-channel SAR deception interference recognition program 40 based on feature map learning can be executed by the processor 10, thereby realizing the single-channel SAR deception interference recognition method based on feature map learning in the present application.
[0109] In some embodiments, the processor 10 may be a central processing unit (CPU), a microprocessor or other data processing chip, used to run the program code or process data stored in the memory 20, such as executing the single-channel SAR deception interference identification method based on feature map learning.
[0110] In some embodiments, the display 30 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode) touch device, etc. The display 30 is used to display information on the terminal and to display a visual user interface. The components 10-30 of the terminal communicate with each other via a system bus.
[0111] In one embodiment, when the processor 10 executes the single-channel SAR deception interference identification program 40 based on feature map learning in the memory 20, the steps of the single-channel SAR deception interference identification based on feature map learning are implemented.
[0112] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a single-channel SAR deception interference identification program based on feature map learning, and when the single-channel SAR deception interference identification program based on feature map learning is executed by a processor, the steps of the single-channel SAR deception interference identification method based on feature map learning as described above are implemented.
[0113] In summary, the present invention provides a single-channel SAR deception interference identification method, system, terminal and computer-readable storage medium based on feature map learning, the method comprising: performing range-wise matched filtering processing and azimuth de-skewing processing on the echo signals received by the receiver, respectively, to obtain all features of the real scene features and the deception interference features; normalizing and adjusting the real scene features and the deception interference features; suppressing the deception interference features, enhancing the real scene features, and obtaining real features; enhancing the deception interference features, suppressing the real scene features, and obtaining interference features; multiplying all the features obtained after normalization by a first coefficient and inputting them into a first feature channel, multiplying the real features by a second coefficient and inputting them into a second feature channel, multiplying the interference features by a third coefficient and inputting them into a third feature channel, and mapping the first feature channel, the second feature channel and the third feature channel to R, G and B channels of the RGB color space, respectively, to generate an RGB image; inputting the obtained RGB image into a neural network for training, and using the weights obtained by the training to identify interference targets. The present invention processes the echo signals differently and puts them into different feature channels respectively, which not only enhances the characteristics of the interference target, but also suppresses the real target information, making the characteristics of the interference target more obvious. Finally, the interference target is identified through a neural network to achieve the purpose of deception interference identification.
[0114] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or terminal including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or terminal. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or terminal including the element.
[0115] Of course, those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing related hardware (such as a processor, a controller, etc.) through a computer program, and the program can be stored in a computer-readable storage medium that can be read by a computer, and the program can include the processes of the above-mentioned method embodiments when executed. The computer-readable storage medium can be a memory, a disk, an optical disk, etc.
[0116] It should be understood that the application of the present invention is not limited to the above examples. For ordinary technicians in this field, improvements or changes can be made based on the above description. All these improvements and changes should fall within the scope of protection of the claims attached to the present invention.
Claims
1. A single-channel SAR deception jamming identification method based on feature graph learning, characterized in that: The single-channel SAR deception jamming identification method based on feature graph learning includes: The echo signals received by the receiver are processed by matched filtering in range and de-skewing in azimuth to obtain all the features of the real scene and the deceptive interference features; Normalizing and adjusting the real scene features and the deception interference features; Suppressing the deceptive interference feature, enhancing the real scene feature, and obtaining the real feature; enhancing the deceptive interference feature and suppressing the real scene feature to obtain an interference feature; All features obtained after normalization are multiplied by the first coefficient and input into the first feature channel, the true features are multiplied by the second coefficient and input into the second feature channel, the interference features are multiplied by the third coefficient and input into the third feature channel, and the first feature channel, the second feature channel and the third feature channel are mapped to the R, G and B channels of the RGB color space respectively to generate an RGB image; The obtained RGB image is input into the neural network for training, and the weights obtained from the training are used to identify the interference target.
2. The single-channel SAR deception jamming identification method based on feature graph learning according to claim 1 is characterized in that: The echo signals received by the receiver are subjected to range-wise matched filtering and azimuth-wise de-skewing processing respectively to obtain all the characteristics of the real scene and the deceptive interference, specifically including: The echo signal received by the receiver is processed by range matching filter, and then the inverse Fourier transform is performed to obtain the processed signal s r (t r ,t a ), where t r is the distance to time, t a is the azimuth time; to s r (t r ,t a ) performs azimuth de-skewing processing, and then performs azimuth Fourier transform to obtain the signal S r (t r ,f a ), where f a is the Doppler frequency, which is expressed as follows: S r (t r ,f a )=F{s r (t r ,t a )s0(t a )}; Among them, s0(t a ) is the de-skew reference signal, F(·) represents the Fourier transform, S r (t r ,f a ) represents all features including real scenes and deceptive interference.
3. The single-channel SAR deception jamming identification method based on feature graph learning according to claim 2 is characterized in that: The normalizing and adjusting the real scene feature and the deception interference feature specifically includes: To S r (t r ,f a ) after normalization, the expression is as follows: Among them, abs[·] means taking the absolute value, max{·} means taking the maximum value, S TF (t r ,f a ) represents all the features of the real scene and deceptive interference after normalization.
4. The single-channel SAR deception jamming identification method based on feature graph learning according to claim 3 is characterized in that: The suppressing of the deceptive interference feature, enhancing the real scene feature, and obtaining the real feature specifically includes: The echo signal after range-matched filtering is de-skewing in azimuth and windowed to suppress the real scene features. The expression is as follows: S J (t r ,f a )=F{ω(t a )s r (t r ,t a )s0(t a )}; Among them, ω(t a ) represents a rectangular window, de-skewing the reference signal s0(t a ) covers the entire temporal distribution of all targets; Suppress deceptive interference features and enhance real scene features to obtain real features: Where α is a constant greater than 0, and the numerator in the formula is expressed as normalized and then raised to the power of α, S J (t r ,f a ) is the signal after suppressing the real scene, in which case the real scene feature is a strong feature and the deception interference feature is a weak feature; When the exponent α∈(0,1), the deceptive interference features are relatively enhanced, and the real scene features are relatively weakened; when α∈(1,+∞), the real scene features are relatively enhanced, and the deceptive interference features are relatively weakened; The normalized S r (t r ,f a ) is put into the denominator of the fraction so that the real scene features are enhanced.
5. The single-channel SAR deception jamming identification method based on feature graph learning according to claim 4 is characterized in that: The step of enhancing the deceptive interference feature and suppressing the real scene feature to obtain the interference feature specifically includes: After the range-matched filtering, the echo signal is subjected to a bandpass filter in the Doppler domain and then subjected to azimuth de-skewing to suppress the interference target. The expression is as follows: S T (t r ,f a )=F{F -1 {F[s(t r ,t a )]H f (f a )}s0(t a )}; Among them, F -1 {·} represents the inverse Fourier transform, H f (f a ) is a bandpass filter; Suppressing the real scene features and enhancing the deceptive interference features to obtain the interference features: Where β is a constant greater than 0, and the numerator in the formula represents the normalization and then taking the β power, S T (t r ,f a ) is the signal after suppressing the interference target, in which case the deception interference feature is a strong feature and the real scene feature is a weak feature; When the exponent β∈(0,1), the real scene features are relatively enhanced, and the deception interference features are relatively weakened; when β∈(1,+∞), the deception interference features are relatively enhanced, and the real scene features are relatively weakened; The normalized S r (t r ,f a ) is put into the denominator of the fraction so that the interference target features are enhanced.
6. The single-channel SAR deception jamming identification method based on feature graph learning according to claim 5 is characterized in that: The method of multiplying all the normalized features by the first coefficient and inputting them into the first feature channel, multiplying the true features by the second coefficient and inputting them into the second feature channel, and multiplying the interference features by the third coefficient and inputting them into the third feature channel specifically includes: S TF (t r ,f a ) multiplied by the first coefficient k1 to obtain: S R (t r ,f a )=k1·S TF (t r ,f a ), S R (t r ,f a ) is input to the first feature channel; S GJ (t r ,f a ) multiplied by the second coefficient k2 to obtain: S G (t r ,f a )=k2·S GJ (t r ,f a ), S G (t r ,f a ) is input to the second feature channel; S BT (t r ,f a ) multiplied by the third coefficient k3 to obtain: S B (t r ,f a )=k3·S BT (t r ,f a ), S B (t r ,f a ) is input to the third feature channel; Among them, k1, k2 and k3 are the weights of the corresponding feature channels, which are used to adjust the strength of the features and are all constants ranging from 0 to 255.
7. The single-channel SAR deception jamming identification method based on feature graph learning according to claim 1 is characterized in that: The obtained RGB image is input into the neural network for training, specifically: The simulated RGB image is input into the neural network for training using deep learning.
8. A single-channel SAR deception jamming recognition system based on feature graph learning, characterized in that: The single-channel SAR deception jamming identification system based on feature map learning includes: The signal processing module is used to perform range-direction matched filtering and azimuth-direction de-skewing on the echo signals received by the receiver to obtain all the features of the real scene and the deceptive interference features; A normalization processing module, configured to perform normalization adjustment on the real scene features and the deception interference features; A first feature processing module is used to suppress the deceptive interference feature, enhance the real scene feature, and obtain the real feature; A second feature processing module is used to enhance the deceptive interference feature and suppress the real scene feature to obtain an interference feature; The feature image generation module multiplies all the normalized features by a first coefficient and inputs them into the first feature channel, multiplies the true features by a second coefficient and inputs them into the second feature channel, multiplies the interference features by a third coefficient and inputs them into the third feature channel, and maps the first feature channel, the second feature channel, and the third feature channel to the R, G, and B channels of the RGB color space, respectively, to generate an RGB image; The training recognition module is used to input the obtained RGB image into the neural network for training, and use the trained weights to identify the interference target.
9. A terminal, characterized in that: The terminal includes: a memory, a processor, and a single-channel SAR deception interference identification program based on feature map learning stored in the memory and executable on the processor. When the single-channel SAR deception interference identification program based on feature map learning is executed by the processor, the steps of the single-channel SAR deception interference identification method based on feature map learning are implemented according to any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a single-channel SAR deception interference identification program based on feature map learning. When the single-channel SAR deception interference identification program based on feature map learning is executed by a processor, the steps of the single-channel SAR deception interference identification method based on feature map learning according to any one of claims 1 to 7 are implemented.
