Interference suppression method and device based on dynamic threshold and edge notch

Abstract

The invention discloses an interference suppression method and device based on a dynamic threshold value and an edge notch. The method comprises the following steps: firstly, acquiring a radar echo signal for detecting a target object; performing time-frequency analysis on the radar echo signal to obtain a corresponding time-frequency graph; then, on the basis of the gray level of each pixel in the time-frequency graph, determining a gray level threshold value meeting a target function; then, based on a gray level threshold value and the gray level of each pixel in the time-frequency graph, determining a modulation matrix; and finally, according to the time-frequency diagram and the modulation matrix, separating from the radar echo signal to obtain a target signal. According to the method, the radar echo signal is converted into the time-frequency diagram, the gray level threshold value is dynamically determined based on the time-frequency diagram and the target function, the interference signal and the target signal in the radar echo signal are separated based on the gray level threshold value, an existing fixed grading strategy is avoided, and the classification accuracy is improved. Therefore, the gray level threshold value can be automatically adjusted according to the characteristics of the real-time radar echo signal, and the interference signal can be identified and suppressed more accurately.

Description

Technical Field

[0001] The present invention belongs to the technical field of radar signal processing, and in particular relates to an interference suppression method and device based on dynamic threshold and edge notch. Background Art

[0002] In modern electronic warfare environments, radar systems face increasingly complex interference threats, including typical active jamming methods such as slicing jamming, false target jamming, swept frequency jamming, and comb spectrum jamming. These jamming methods can significantly impact the radar system's subsequent target detection, positioning, and identification processes, and can even cause false alarms or misjudgments. In practical applications, when the signal-to-noise ratio of the echo signal received by the radar is low, in order to retain more valid echo signals of the real target, the jammer will sort the echo signals when forwarding them. The processed echo signals are then forwarded to the radar system, creating a complex scenario of "strong jammer + weak target."

[0003] Existing anti-interference technology usually uses a grayscale grading method to deal with complex scenarios where strong interference signals and weak target signals coexist. This method relies on manually pre-set gradation numbers to suppress signals in a graded manner. Therefore, it is easy to cause excessive suppression or poor suppression effect due to inaccurate gradation number settings. Summary of the Invention

[0004] In order to solve the above problems existing in the prior art, the present invention provides an interference suppression method and device based on a dynamic threshold and edge notch.

[0005] The technical problem to be solved by the present invention is achieved through the following technical solutions:

[0006] In a first aspect, the present invention provides an interference suppression method based on a dynamic threshold and an edge notch, comprising:

[0007] Acquiring a radar echo signal for detecting a target object, where the radar echo signal includes a target signal and an interference signal;

[0008] Perform time-frequency analysis on the radar echo signal to obtain the corresponding time-frequency diagram;

[0009] Based on the grayscale of each pixel in the time-frequency graph, a grayscale threshold that satisfies the objective function is determined. The grayscale threshold is used to separate the target pixels corresponding to the target signal contained in the time-frequency graph and the interference pixels corresponding to the interference signal;

[0010] Determine a modulation matrix based on a gray level threshold and the gray level of each pixel in the time-frequency graph;

[0011] According to the time-frequency diagram and modulation matrix, the target signal is separated from the radar echo signal.

[0012] In a second aspect, the present invention provides an interference suppression device based on a dynamic threshold and an edge notch, which is applied to a graphics processing unit (GPU), comprising:

[0013] A signal generation module is used to generate a simulated radar echo signal, wherein the simulated radar echo signal includes a simulated target signal and a simulated interference signal;

[0014] The interference suppression module is used to perform time-frequency analysis on the simulated radar echo signal to obtain the corresponding simulated time-frequency diagram; based on the grayscale of each pixel in the simulated time-frequency diagram, a simulated grayscale threshold that meets the objective function is determined, and the simulated grayscale threshold is used to separate the simulated target pixels corresponding to the simulated target signal contained in the simulated time-frequency diagram and the simulated interference pixels corresponding to the simulated interference signal; based on the simulated grayscale threshold and the grayscale of each pixel in the simulated time-frequency diagram, a simulated modulation matrix is ​​determined; according to the simulated time-frequency diagram and the simulated modulation matrix, the simulated target signal is separated from the simulated radar echo signal.

[0015] The present invention provides an interference suppression method and device based on dynamic thresholds and edge notches. The method first acquires a radar echo signal used to detect a target object; then, time-frequency analysis is performed on the radar echo signal to obtain a corresponding time-frequency graph; then, based on the grayscale level of each pixel in the time-frequency graph, a grayscale threshold that satisfies an objective function is determined; thereafter, a modulation matrix is ​​determined based on the grayscale threshold and the grayscale level of each pixel in the time-frequency graph; finally, the target signal is separated from the radar echo signal based on the time-frequency graph and the modulation matrix. The present invention converts the radar echo signal into a time-frequency graph, dynamically determines the grayscale threshold based on the time-frequency graph and the objective function, and separates the interference signal and the target signal in the radar echo signal based on the grayscale threshold. This avoids existing fixed classification strategies, thereby automatically adjusting the grayscale threshold based on the characteristics of the real-time radar echo signal, and achieving more accurate identification and suppression of interference signals.

[0016] The present invention will be further described in detail below with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 1 is a flow chart of an interference suppression method based on a dynamic threshold and an edge notch provided by an embodiment of the present invention;

[0018] Figure 2A and 2B 1 is a schematic diagram of a process for generating a time-frequency diagram in an interference suppression method based on a dynamic threshold and an edge notch provided in an embodiment of the present invention;

[0019] Figure 3 1 is a schematic structural diagram of an interference suppression device based on a dynamic threshold and an edge notch provided by an embodiment of the present invention;

[0020] Figure 4 Schematic diagram of a process for generating a simulated target signal by an interference suppression device based on a dynamic threshold and an edge notch provided by an embodiment of the present invention;

[0021] Figure 5 This is a schematic diagram of a process for generating a simulated comb spectrum interference signal by an interference suppression device based on a dynamic threshold and an edge notch provided by an embodiment of the present invention;

[0022] Figure 6 This is a schematic diagram of a process for generating a simulated false target interference signal by an interference suppression device based on a dynamic threshold and an edge notch provided by an embodiment of the present invention;

[0023] Figure 7 This is a schematic diagram of a process for generating a simulated slice interference signal by an interference suppression device based on a dynamic threshold and an edge notch provided by an embodiment of the present invention;

[0024] Figure 8 This is a schematic diagram of a process for generating a simulated frequency-sweep interference signal by an interference suppression device based on a dynamic threshold and an edge notch, provided by an embodiment of the present invention;

[0025] Figure 9 3 is a schematic diagram comparing the results of an interference suppression method based on dynamic threshold and edge notch provided by an embodiment of the present invention and an existing method. DETAILED DESCRIPTION

[0026] The present invention will be further described in detail below with reference to specific examples, but the embodiments of the present invention are not limited thereto.

[0027] The embodiment of the present invention provides an interference suppression method based on dynamic threshold and edge notch, see Figure 1 , the method comprises the following steps:

[0028] S10: Acquire a radar echo signal for detecting a target object.

[0029] Among them, the radar echo signal contains target signal and interference signal.

[0030] Specifically, the radar echo signal may be a time domain signal, and the interference signal may include a slice interference signal, a false target interference signal, a swept frequency interference signal, a comb spectrum interference signal, and the like.

[0031] S20. Perform time-frequency analysis on the radar echo signal to obtain a corresponding time-frequency graph.

[0032] Exemplarily, a time-frequency transformation is performed on the radar echo signal to obtain a time-frequency domain signal, and then the time-frequency domain signal is subjected to modulus logarithm processing to obtain a corresponding time-frequency graph.

[0033] Among them, the time-frequency domain signal STFT|y(t) (k,t) can be expressed as:

[0034]

[0035] Where y(t) represents the input radar echo signal, L represents the window length, l represents the window length index, w(l) represents the window function, t represents the fast time index, and k represents the frequency index.

[0036] The time-frequency diagram can be expressed as:

[0037] S = log(|STFT| y(t) (k,t)|+1);

[0038] Where S represents the gray level of each pixel in the time-frequency graph.

[0039] In this embodiment, since traditional time domain filtering makes it difficult to distinguish between interference signals and target signals, the target signal and the interference signal may be superimposed in the same time period, and may also overlap partially in the frequency domain. Therefore, by performing time-frequency analysis on the radar echo signal, a corresponding time-frequency diagram is obtained. On the time-frequency diagram, the radar echo signal has a clear time-frequency trajectory, and the interference signal often appears as an abnormal energy area. Therefore, the pixel area corresponding to the interference signal can be accurately identified and removed through image processing and other methods.

[0040] Optionally, performing time-frequency analysis on the radar echo signal to obtain a corresponding time-frequency graph may specifically include:

[0041] S201: Divide the radar echo signal into multiple signal segments.

[0042] For example, referring to Figure 2A and Figure 2B The radar echo signal can be divided into two signal segments in the CPU, and the two signal segments can be processed in parallel using multiple streams (stream 1 and stream 2 in the figure) using the CUDA cores in the GPU. The GPU first allocates memory and enables CUDA stream page-locked memory allocation. The following is the multi-stream parallel processing process:

[0043] S202 : Perform sliding windowing processing on each signal segment in parallel to obtain a windowed signal corresponding to each signal segment.

[0044] For example, referring to Figure 2A and Figure 2B Each stream copies the signal segment and the preset window function from the CPU, caches the window function to the shared memory, and assigns a thread to each data point in the signal segment for subsequent data processing. Multiple threads perform parallel sliding windowing processing on the data points based on the window function in the shared memory to obtain the windowed signal corresponding to each signal segment.

[0045] S203 : Perform short-time Fourier transform on the windowed signal corresponding to each signal segment in parallel to obtain a time-frequency domain signal corresponding to each signal segment.

[0046] For example, referring to Figure 2A and Figure 2B , multiple threads use the cuFFT library function in parallel to perform short-time Fourier transform (FFT) processing on the windowed signal corresponding to each data point to obtain the time-frequency domain signal corresponding to each signal segment.

[0047] S204: Generate a time-frequency diagram based on the time-frequency domain signals corresponding to all signal segments.

[0048] Exemplarily, the time-frequency domain signals corresponding to each signal segment may be spliced ​​to obtain a complete time-frequency domain signal, and then the time-frequency domain signal may be subjected to modulus logarithm processing to obtain a corresponding time-frequency graph.

[0049] This embodiment forms a pipelined concurrent execution structure by dividing sliding windowing, FFT transformation and data transmission into multiple subtasks, and assigning them to multiple independent CUDA Streams (CUDA Stream is an independent task queue on the GPU for managing and scheduling a series of GPU operations). Each Stream independently processes its corresponding signal segment and completes the window function weighting and FFT operation. The windowing stage further adopts a thread-level parallel design, in which each thread is responsible for the processing of one data point, and through the shared memory cache window function, only a small amount of global memory reading is required to implement the windowing operation of the entire data point, effectively reducing memory access delay and redundant access overhead. In addition, the FFT transformation stage also adopts parallel time-frequency domain transformation. Finally, combined with page-locked memory and the asynchronous data copy mechanism between the CPU and GPU, overlapping scheduling of calculation and data transmission can be achieved, which greatly improves the GPU resource utilization and data throughput performance.

[0050] S30 , determining a gray level threshold that satisfies an objective function based on the gray level of each pixel in the time-frequency graph.

[0051] In step S30, optionally, based on the gray level of each pixel in the time-frequency graph, determining a gray level threshold that satisfies the objective function may specifically include:

[0052] S301 : Based on the grayscale of each pixel in the time-frequency graph, determine the ratio of the number of pixels corresponding to each grayscale to the total number of pixels.

[0053] For example, a grayscale histogram of the time-frequency graph can be generated based on the grayscale of each pixel in the time-frequency graph, and the number of pixels appearing at each grayscale in the time-frequency graph can be obtained by counting, and then the proportion of the number of pixels corresponding to each grayscale to the total number of pixels can be calculated.

[0054] Optionally, the ratio p of the number of pixels corresponding to each gray level to the total number of pixels i It can be expressed as:

[0055]

[0056] Among them, i represents the gray level, n i Represents the number of pixels corresponding to each gray level, and N represents the total number of pixels in the time-frequency graph.

[0057] S302. Determine a first weight corresponding to a first grayscale range to which the target pixel belongs and a second weight corresponding to a second grayscale range to which the interference pixel belongs based on a ratio of the number of pixels corresponding to each grayscale level to the total number of pixels in the time-frequency diagram and a grayscale threshold parameter to be solved.

[0058] The first grayscale range is smaller than the second grayscale range.

[0059] Exemplarily, a grayscale threshold parameter H can be set, and the pixels in the time-frequency graph can be divided into two parts based on the grayscale threshold parameter H, one part is the first grayscale range [0, H] to which the target pixel belongs, and the other part is the second grayscale range [H+1, 255] to which the interference pixel belongs, and the first weight and second weight corresponding to the two grayscale ranges are calculated respectively.

[0060] Optionally, the first weight w0 and the second weight w1 can be expressed as:

[0061]

[0062] Wherein, H represents the grayscale threshold parameter, the first grayscale range is [0, H], and the second grayscale range is [H+1, 255].

[0063] S303 : Determine a first average grayscale value corresponding to the first grayscale range and a second average grayscale value corresponding to the second grayscale range based on the ratio of the number of pixels corresponding to each grayscale level to the total number of pixels, the first weight, and the second weight.

[0064] Optionally, the first average grayscale value μ0 and the second average grayscale value μ1 can be respectively expressed as:

[0065]

[0066] S304 : Setting an objective function based on the first weight, the second weight, the first average grayscale value, and the second average grayscale value, and solving the objective function to obtain a grayscale threshold.

[0067] The objective function is used to characterize the accuracy of separating target pixels and interference pixels based on the grayscale threshold.

[0068] Alternatively, the objective function can be expressed as:

[0069] η(H)=(μ0-μ1)·w0·w1·(μ0-μ1);

[0070] H * =argmax H∈[0,255] {η(H)};

[0071] Among them, H * Represents the gray level threshold obtained by solution, argmax H∈[0,255] {η(H)} represents the objective function, and η(H) represents the intermediate parameter.

[0072] Specifically, when solving, an iterative method is used to traverse H∈[0,255] to obtain an optimal threshold H (i.e. H * ), which can maximize the value of η(H), indicating that the gray level threshold H * It can achieve the best separation effect of target pixels and interference pixels.

[0073] The grayscale threshold is used to separate the target pixels corresponding to the target signal and the interference pixels corresponding to the interference signal contained in the time-frequency diagram.

[0074] S40 , determining a modulation matrix based on the grayscale threshold and the grayscale of each pixel in the time-frequency graph.

[0075] Exemplarily, the modulation matrix E can be expressed as:

[0076]

[0077] Where S represents the gray level of each pixel in the time-frequency graph.

[0078] S50. Separate the target signal from the radar echo signal according to the time-frequency diagram and the modulation matrix.

[0079] For example, the separated interference signal STFT| can be obtained by multiplying the modulation matrix with the time-frequency domain signal contained in the time-frequency diagram. y(t) (k,t) jam It can be expressed as:

[0080] STFT| y(t) (k,t) jam =STFT| y(t) (k,t)·E;

[0081] At the same time, the time-frequency domain signal STFT| after removing the interference signal can be obtained. y(t) (k,t) echo , which can be expressed as:

[0082] STFT|y(t) (k,t) echo =STFT| y(t) (k,t)·(1-E);

[0083] Finally, the target signal can be obtained by directly performing time domain transformation on the time-frequency domain signal.

[0084] In step S50, optionally, separating the target signal from the radar echo signal according to the time-frequency diagram and the modulation matrix may specifically include:

[0085] S501 : Separate an initial target pixel region and an initial interference pixel region from the time-frequency map according to the time-frequency map and the modulation matrix.

[0086] Exemplarily, the interference signal STFT| can be obtained according to the time-frequency diagram and the modulation matrix. y(t) (k,t) jam After the time-frequency domain signals, the time-frequency domain signals STFT| y(t) (k,t) echo The secondary interference signal left at the edge of the middle part is further filtered out, then based on the interference signal STFT| y(t) (k,t) jam and time-frequency domain signal STFT| y(t) (k,t) echo , the initial target pixel region and the initial interference pixel region are separated from the time-frequency graph.

[0087] S502 : Perform smoothing convolution processing and binarization processing on the initial interference pixel region to obtain an expanded interference pixel region.

[0088] For example, in STFT| y(t) (k,t) jam The initial interference pixel area S corresponding to the signal jam Perform smooth convolution on each pixel using the target Gaussian kernel:

[0089]

[0090] Where (a, b) represents the initial interference pixel area S jam The coordinates of each pixel on the .

[0091] Among them, the target Gaussian kernel uses a 5*5 two-dimensional Gaussian kernel with a standard deviation (σ1) of 2, which can be expressed as:

[0092]

[0093] Where (x, y) represents the row and column offset of each element in the target Gaussian kernel relative to the center position, and K(x, y) represents the weight value at the corresponding position (x, y) of the 5*5 Gaussian kernel.

[0094] After that, it is the regional image S' after convolution processing jam_gauss Set the threshold for binarization and get the extended interference pixel area S jam_gauss .

[0095] S503: Determine edge interference pixels in the initial target pixel area.

[0096] Exemplarily, the target time-frequency domain signal STFT| y(t) (k,t) echo Corresponding initial target pixel area Calculate the mean μ and standard deviation σ2 of its pixels:

[0097]

[0098] Where C and D represent the initial target pixel area The number of rows and columns, c and d represent the corresponding row index and column index respectively.

[0099] Afterwards, according to the judgment formula:

[0100]

[0101] The pixel points in the initial target pixel area that meet the above judgment formula are edge interference pixels.

[0102] S504 : Based on the edge interference pixels in the extended interference pixel area and the initial target pixel area, perform edge interference filtering on the initial target pixel area to obtain a target signal corresponding to the target pixel area.

[0103] In step S504, optionally, based on the edge interference pixels in the extended interference pixel region and the initial target pixel region, edge interference filtering is performed on the initial target pixel region to obtain a target signal corresponding to the target pixel region, which may specifically include:

[0104] S5041 , filtering out target edge interference pixels in the initial target pixel region to obtain a target time-frequency domain signal corresponding to the target pixel region; the target edge interference pixels are edge interference pixels located in the extended interference pixel region.

[0105] Exemplarily, an edge interference mask is constructed based on the edge interference pixels in the extended interference pixel area and the initial target pixel area:

[0106]

[0107] Afterwards, the initial target pixel area is subjected to edge interference filtering according to the edge interference mask, and the target time-frequency domain signal STFT| corresponding to the target pixel area is obtained. y(t) (k,t) final , which can be expressed as:

[0108] STFT| y(t) (k,t) final =STFT| y(t) (k,t) echo ·(1-B).

[0109] S5042. Perform time domain transformation on the target time-frequency domain signal to obtain a target signal.

[0110] Exemplarily, the time-frequency domain signal STFT| y(t) (k,t) final The target signal can be obtained by time domain transformation, which can be expressed as:

[0111]

[0112] Where, φ t is a discrete time domain signal.

[0113] In this embodiment, a more precise and efficient interference suppression effect is achieved by further suppressing the secondary high-level interference signals remaining in the edge region of the initial target pixel region.

[0114] The present invention provides an interference suppression method based on dynamic thresholds and edge notches. The method first obtains a radar echo signal used to detect a target object; then performs time-frequency analysis on the radar echo signal to obtain a corresponding time-frequency graph; then, based on the grayscale level of each pixel in the time-frequency graph, determines a grayscale threshold that satisfies an objective function; then, based on the grayscale threshold and the grayscale level of each pixel in the time-frequency graph, determines a modulation matrix; finally, based on the time-frequency graph and the modulation matrix, separates the target signal from the radar echo signal. The present invention converts the radar echo signal into a time-frequency graph, dynamically determines the grayscale threshold based on the time-frequency graph and the objective function, and separates the interference signal and the target signal in the radar echo signal based on the grayscale threshold. This avoids the existing fixed classification strategy, and can automatically adjust the grayscale threshold according to the characteristics of the real-time radar echo signal, thereby more accurately identifying and suppressing the interference signal.

[0115] Corresponding to the above-mentioned interference suppression method based on dynamic threshold and edge notch, an embodiment of the present invention further provides an interference suppression device based on dynamic threshold and edge notch, which is applied to a graphics processing unit (GPU); Figure 3 As shown, the device may include:

[0116] The signal generating module 301 is used to generate a simulated radar echo signal, wherein the simulated radar echo signal includes a simulated target signal and a simulated interference signal;

[0117] The interference suppression module 302 is used to perform time-frequency analysis on the simulated radar echo signal to obtain a corresponding simulated time-frequency diagram; based on the grayscale of each pixel in the simulated time-frequency diagram, determine a simulated grayscale threshold that meets the objective function, and the simulated grayscale threshold is used to separate the simulated target pixels corresponding to the simulated target signal and the simulated interference pixels corresponding to the simulated interference signal contained in the simulated time-frequency diagram; based on the simulated grayscale threshold and the grayscale of each pixel in the simulated time-frequency diagram, determine a simulated modulation matrix; and according to the simulated time-frequency diagram and the simulated modulation matrix, separate the simulated target signal from the simulated radar echo signal.

[0118] Optionally, the signal generating module 301 of this embodiment may include a simulated target signal unit, a simulated interference signal unit and a signal fusion unit.

[0119] The simulation target signal unit is used to generate multiple simulation signal data points according to the first preset parameters using multi-threaded parallelism to obtain a simulation target signal.

[0120] For example, multiple point targets may be set, and the echo expressions of K point targets at the fast time t in the nth pulse are:

[0121]

[0122] τ(n)=2·R t (n) / c

[0123]

[0124] Where λ is the wavelength, R t (n) is the spatial distance between the point target and the current radar, γ is the modulation frequency, which represents the frequency change rate of the linear frequency modulation signal, w is a rectangular time window, τ(n) is the round-trip time difference of the echo, c is the speed of light, s(t,n) is the echo expression of a single point target at the fast time t in the nth pulse, and x(t,n) is the total echo signal of K point targets at the fast time t in the nth pulse.

[0125] The first preset parameters may include the motion trajectory data of the current radar, the time series (fast time axis), the three-dimensional coordinates of the point target, and the pulse width of the radar pulse.

[0126] Fast time refers to the sampling time axis of the radar on the echo signal within one pulse (one transmission cycle), also known as the range direction, which corresponds to the row direction of the data matrix of the two-dimensional radar echo signal. Slow time refers to the serial time between pulses when the radar continuously transmits multiple pulses, also known as the azimuth direction, which corresponds to the column direction of the data matrix of the two-dimensional radar echo signal.

[0127] Reference Figure 4 The figure shows the entire process of generating simulated target signals for point targets on the GPU using the simulated target signal unit, based on the echo expressions of K point targets at fast time t in the nth pulse. The simulated target signal unit uses a two-dimensional thread block design, with each thread responsible for calculating a single simulated signal data point.

[0128] Since the generation of simulated target signals is a doubly nested calculation process: large-scale calculations are required in both the azimuth and range dimensions, and the data between them are relatively independent, a GPU is used here to accelerate the generation of simulated target signals. As shown in the figure, a column of simulated target signals corresponding to a certain pulse is then selected, and Gaussian white noise is mixed into the column of simulated target signals. The signal-to-noise ratio can be set to -5dB.

[0129] The simulated interference signal unit is used to generate multiple simulated interference signals according to the second preset parameters. The first preset parameters are different from the second preset parameters. The multiple simulated interference signals include at least simulated slice interference signals, simulated false target interference signals, simulated swept frequency interference signals and simulated comb spectrum interference signals.

[0130] For example, the second preset parameters may include the number of range points Nr, the number of azimuth points Na, the sampling rate Fs, and the carrier frequency Fc. Based on the generated simulated target signal of the point target, the number of range points Nr and the number of azimuth points Na, i.e., the rows and columns of the data matrix of the simulated target signal, can be obtained respectively.

[0131] The signal fusion unit is used to fuse the simulated target signal and multiple simulated interference signals to obtain a simulated radar echo signal.

[0132] Exemplarily, the simulated target signal and multiple simulated interference signals are added in the time domain to obtain a simulated radar echo signal.

[0133] Optionally, the process of the simulated interference signal unit generating a simulated comb spectrum interference signal may include: generating multiple time vectors based on a second preset parameter; generating multiple sub-interference signals using multi-threaded parallelism based on the multiple time vectors and preset sub-signal parameters; the preset sub-signal parameters include the number, frequency and order of sub-signals; using multi-threaded parallelism to superimpose multiple sub-interference signals to obtain a one-dimensional simulated comb spectrum interference signal; using multi-threaded parallelism to Doppler modulate multiple data points in the one-dimensional simulated comb spectrum interference signal to obtain a simulated comb spectrum interference signal.

[0134] Exemplarily, the expression for simulating the comb spectrum interference signal may be:

[0135]

[0136] Where A m is the amplitude of the sub-interference signal, f m is the frequency of the sub-interference signal, M is the number of sub-interference signals, t is the time vector in a single pulse (i.e., fast time), f d is the Doppler frequency, t s is the slow time axis, which represents the time vector between pulse repetition periods, C is the speed of light, J1(t) is the one-dimensional simulated comb spectrum interference signal under a single pulse, J1 n (t) is a two-dimensional simulated comb spectrum interference signal under multiple pulses, which is used to simulate the comb spectrum interference signal in a real radar environment.

[0137] Reference Figure 5 The figure shows the entire process of generating a simulated comb spectrum interference signal on the GPU using the simulated interference signal unit, based on the expression of the simulated comb spectrum interference signal. During the generation and superposition phase of the interference signal, a one-dimensional thread block is designed, with each thread responsible for calculating and superimposing a single data point of the one-dimensional simulated comb spectrum interference signal. During the Doppler modulation phase, a two-dimensional thread block design is used, with each thread responsible for calculating a single data point of the two-dimensional simulated comb spectrum interference signal.

[0138] Optionally, the process of generating the simulated false target interference signal by the simulated interference signal unit includes:

[0139] Based on the second preset parameters, multiple time vectors, frequency vectors and interference signal windows are generated in parallel using multi-threading; based on the multiple time vectors, frequency vectors and interference signal windows, a linear frequency modulation signal is generated; based on the preset false target parameters and unit pulse signal, a false target pulse signal is generated; the false target parameters include the number of false targets, the time interval and the intensity factor of the false target pulses; the linear frequency modulation signal and the false target pulse signal are phase modulated by multi-threading in parallel to obtain a one-dimensional simulated false target interference signal; the multiple data points in the one-dimensional simulated false target interference signal are Doppler modulated by multi-threading in parallel to obtain a simulated false target interference signal.

[0140] Exemplarily, the expression of the simulated false target interference signal can be:

[0141]

[0142] Where S(t) represents a pre-constructed single interference pulse waveform (a linear frequency modulation signal modulated by an interference signal window function), A RMF is the intensity factor of each simulated false target interference signal, Q is the number of false targets, T is the time interval of each false target pulse. t represents the time vector in a single pulse, f d is the Doppler shift, t s is the slow time axis, representing the time vector between pulse repetition periods, represents convolution, δ1 is the unit pulse signal, Q pulses are placed at T as the time interval, and convolution calculation is performed with S(t), C is the speed of light, J2(t) is the one-dimensional simulated false target interference signal under a single pulse, J2 n (t) is a two-dimensional simulated false target interference signal under multiple pulses, which is used to simulate the false target interference signal in a real radar environment.

[0143] Reference Figure 6 The figure shows the entire process of generating a simulated false target interference signal on the GPU, based on the expression of the simulated false target interference signal. During the generation process, a one-dimensional thread block is designed to generate the time vector, frequency vector, and interference signal window, with each thread calculating one data point. It is also responsible for range phase modulation, with each thread calculating the phase superposition of one frequency data point in the one-dimensional simulated false target interference signal. During the Doppler modulation stage, a two-dimensional thread block design is used, with each thread calculating one data point in the two-dimensional simulated false target interference signal.

[0144] Optionally, the process of generating the simulated slicing interference signal by the simulated interference signal unit includes:

[0145] Based on the second preset parameters, multiple first time vectors, first frequency vectors and first interference signal windows are generated in parallel using multi-threading; based on the multiple first time vectors, first frequency vectors and first interference signal windows, linear frequency modulation signals and frequency domain sub-pulse shifts are generated; based on the linear frequency modulation signals and frequency domain sub-pulse shifts, a one-dimensional simulated slice interference signal is synthesized; and multiple data points in the one-dimensional simulated slice interference signal are Doppler modulated in parallel using multi-threading to obtain a simulated slice interference signal.

[0146] For example, the expression of the simulated slice interference signal may be:

[0147]

[0148] Where, S(t-mT w ) indicates that the simulated slice interference signal is delayed accordingly in time and superimposed in the time domain, N1 indicates the number of pulses, n indicates the index of the pulse, M1 is used to control the time interval between each pulse, m1 indicates the index of the time interval between pulses, T w is the width of the rectangular time window, t represents the time vector in a single pulse, f d is the Doppler frequency, t s is the slow time axis, representing the time vector between pulse repetition periods. Represents a rectangular window function with a width of T and a center of 0. 1 only when t is 0, and 0 otherwise. C is the speed of light. J3(t) is the one-dimensional simulated slice interference signal under a single pulse. n (t) is a two-dimensional simulated slice interference signal under multiple pulses, which is used to simulate the slice interference signal in a real radar environment.

[0149] Reference Figure 7 The figure shows the entire process of generating the simulated slice interference signal on the GPU based on the expression of the simulated slice interference signal. During the generation process, a one-dimensional thread block is designed to generate the first time vector, the first frequency vector, and the first interference signal window, with each thread calculating one data point; and for frequency phase modulation, each thread calculates one data point of the one-dimensional simulated slice interference signal. During the Doppler modulation stage, a two-dimensional thread block design is adopted, with each thread calculating one data point of the two-dimensional simulated slice interference signal.

[0150] Optionally, the process of generating the simulated frequency sweep interference signal by the simulated interference signal unit includes:

[0151] Based on the second preset parameter, a sawtooth wave is generated in parallel using multi-threading; based on the sawtooth wave, the frequency integral area is calculated; based on the frequency integral area, a one-dimensional simulated swept-frequency interference signal is generated; and Doppler modulation is performed on multiple data points in the one-dimensional simulated swept-frequency interference signal using multi-threading in parallel to obtain a simulated swept-frequency interference signal.

[0152] For example, the expression of the simulated swept frequency interference signal may be:

[0153]

[0154] Where f0 is the initial center frequency of the simulated frequency sweep interference signal, Δf s is the sweep bandwidth, i.e. the frequency variation range, m fe is the effective modulation index, T s is the frequency sweep period, t represents the time vector in a single pulse, f d is the Doppler frequency, δ2 represents the idealized frequency focus point, δ2f represents the concentrated energy distribution on the frequency, and f j Expressed as the frequency of the interference signal, Δf j The total bandwidth of the interference signal is the radar frequency resolution f r 2-5 times of t s is the slow time axis, which represents the time vector between pulse repetition periods, A is the interference signal amplitude, C is the speed of light, J4(t) is the one-dimensional simulated swept frequency interference signal under a single pulse, J4 n (t) is a two-dimensional simulated frequency-sweep jammer signal under multiple pulses, which is used to simulate the frequency-sweep jammer signal in a real radar environment.

[0155] Reference Figure 8 , shows the entire process of generating a simulated swept-frequency interference signal on the GPU based on the expression of the simulated swept-frequency interference signal. During the generation process, a one-dimensional thread block is designed. In the sawtooth wave generation stage, it is responsible for generating the sawtooth frequency vector, with each thread calculating one data point; it is also responsible for calculating the frequency integral area; and generating the one-dimensional simulated swept-frequency interference signal, with each thread calculating one data point of the one-dimensional simulated swept-frequency interference signal. In the Doppler modulation stage, a two-dimensional thread block design is adopted, with each thread calculating one data point of the two-dimensional simulated swept-frequency interference signal.

[0156] After generating the simulated slice interference signal, simulated false target interference signal, simulated swept frequency interference signal and simulated comb spectrum interference signal, a column of interference data corresponding to the selected pulse in the simulated target signal is extracted respectively, and Gaussian white noise is mixed with each column of interference signal respectively, and the signal-to-noise ratio is set to 20dB.

[0157] In this embodiment, the parallel processing capability of the GPU is fully utilized to achieve rapid generation and output of large-scale simulated interference signals, thereby improving simulation efficiency.

[0158] It should be noted that, as for the interference suppression module in the device of this embodiment, since it is basically similar to the method embodiment of the first aspect mentioned above, it will not be described in detail here. For relevant details, please refer to the partial description of the method embodiment.

[0159] The interference suppression device based on dynamic thresholds and edge notching provided by the present invention performs time-frequency analysis on a simulated radar echo signal to obtain a corresponding simulated time-frequency graph. Then, based on the grayscale of each pixel in the simulated time-frequency graph, a simulated grayscale threshold that satisfies the objective function is determined. Next, a simulated modulation matrix is ​​determined based on the simulated grayscale threshold and the grayscale of each pixel in the simulated time-frequency graph. Finally, based on the simulated time-frequency graph and the simulated modulation matrix, a simulated target signal is separated from the simulated radar echo signal. This invention avoids existing fixed classification strategies, thereby enabling more accurate identification and suppression of interference signals.

[0160] The interference suppression method based on dynamic threshold and edge notch and the simulation device provided by the present invention are further described below through simulation experiments.

[0161] The present invention adopts a dynamic threshold method to replace the traditional fixed classification strategy, which can automatically adjust the threshold according to real-time signal characteristics and environmental conditions, thereby more accurately identifying and suppressing different types of interference signals.

[0162] In addition, for the secondary interference signals left over from the edge effect, the present invention adopts the interference edge notch method for fine processing, which effectively avoids the problem of over-filtering and improves the ability to suppress the secondary interference signals.

[0163] Figure 9 The comparison of the interference suppression effect between the traditional fixed threshold method and the method of the present invention is demonstrated.

[0164] Table 1 Comparison of interference suppression indexes between existing methods and the method of the present invention

[0165]

[0166] During the verification phase, the method developed by the present invention constructed a target signal and interference signal simulation device based on GPU acceleration. The simulated target signal size was 2048×3200, and the simulated interference signal size was 2048×6400, meeting the requirements of large-scale data processing. During the anti-interference processing, the range data corresponding to a single pulse was selected as input, focusing on analyzing the local interference effect. The experimental platform used an NVIDIA RTX 4050 GPU, which has excellent parallel computing capabilities and can effectively support high-speed simulation and processing of large-scale data.

[0167] Table 2 shows the average time consumption comparison of different methods in the process of generating simulated target signals, generating multiple simulated interference signals, and anti-interference processing, which further verifies the advantage of this method in computational efficiency.

[0168] Table 2 Comparison of simulation time consumption by different methods

[0169]

[0170] As can be seen from the above table, the method of the present invention significantly improves the generation efficiency of simulated target signals and simulated interference signals in the simulation verification stage. Thanks to the GPU parallel acceleration mechanism, it realizes the rapid generation of large-scale simulation data and significantly shortens the simulation time, fully verifying the actual effect and application value of the present invention in optimizing simulation performance.

[0171] To verify the performance improvements of the proposed GPU-accelerated time-frequency analysis (STFT) method in actual radar signal processing, a comparative experiment was designed to analyze the operational efficiency of a traditional single-stream serial GPU method and the proposed multi-stream parallel STFT implementation. By comparing computational times for the same input data size, the proposed method's combined advantages in task scheduling optimization, data transmission overlap, and computational parallelism were evaluated, further validating the proposed method's acceleration effect and potential for application in the time-frequency conversion process.

[0172] Table 3 shows the average time consumption comparison of different implementations when performing STFT transformation, further demonstrating the significant advantages of the present invention in terms of computational efficiency.

[0173] Table 3 STFT conversion efficiency comparison

[0174]

[0175] As can be seen from the results in the table, compared with the traditional GPU single-stream serial method, the present invention shows better execution efficiency in STFT calculation, which fully verifies the significant advantages of the multi-stream parallel strategy in improving processing speed and GPU resource utilization.

[0176] It should be noted that the terms "first," "second," and the like are used to distinguish similar objects and are not necessarily used to describe a particular order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate, so that the embodiments of the present invention described herein can be implemented in sequences other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Instead, they are merely examples of devices and methods consistent with some aspects of the present invention.

[0177] In the description of this specification, the reference terms "one embodiment," "some embodiments," "example," "specific example," or "some examples" mean that the specific features or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features or characteristics described can be combined in any suitable manner in any one or more embodiments or examples. In addition, those skilled in the art can combine and combine different embodiments or examples described in this specification.

[0178] Although the present invention is described herein in conjunction with various embodiments, in the process of implementing the claimed invention, those skilled in the art can understand and implement other variations of the disclosed embodiments by viewing the drawings and the disclosed content. In the description of the present invention, the word "comprising" does not exclude other components or steps, "one" or "a" does not exclude multiple situations, and "multiple" means two or more, unless otherwise clearly and specifically defined. In addition, certain measures are recorded in different embodiments, but this does not mean that these measures cannot be combined to produce good results.

[0179] The above is a further detailed description of the present invention in conjunction with specific preferred embodiments, and the specific implementation of the present invention should not be considered to be limited to these descriptions. For those skilled in the art to which the present invention belongs, several simple deductions or substitutions can be made without departing from the concept of the present invention, and all of these should be considered to fall within the scope of protection of the present invention.

Claims

1. An interference suppression method based on dynamic threshold and edge notch, characterized in that: include: Acquiring a radar echo signal for detecting a target object, wherein the radar echo signal includes a target signal and an interference signal; Performing time-frequency analysis on the radar echo signal to obtain a corresponding time-frequency graph; Determining a grayscale threshold that satisfies an objective function based on the grayscale level of each pixel in the time-frequency graph, wherein the grayscale threshold is used to separate target pixels corresponding to target signals and interference pixels corresponding to interference signals contained in the time-frequency graph; determining a modulation matrix based on the grayscale threshold and the grayscale of each pixel in the time-frequency graph; The target signal is separated from the radar echo signal according to the time-frequency diagram and the modulation matrix.

2. The interference suppression method based on dynamic threshold and edge notch according to claim 1, characterized in that: The step of determining a gray level threshold that satisfies an objective function based on the gray level of each pixel in the time-frequency graph includes: Based on the grayscale of each pixel in the time-frequency graph, determining the ratio of the number of pixels corresponding to each grayscale to the total number of pixels; Determining, based on a ratio of the number of pixels corresponding to each gray level to the total number of pixels in the time-frequency graph and a gray level threshold parameter to be solved, a first weight corresponding to a first gray level range to which the target pixel belongs and a second weight corresponding to a second gray level range to which the interference pixel belongs; the first gray level range is smaller than the second gray level range; Determining a first average grayscale value corresponding to a first grayscale range and a second average grayscale value corresponding to a second grayscale range based on a ratio of the number of pixels corresponding to each grayscale level to the total number of pixels, the first weight, and the second weight; An objective function is set based on the first weight, the second weight, the first average grayscale value and the second average grayscale value, and the objective function is solved to obtain a grayscale threshold; the objective function is used to characterize the accuracy of separating the target pixel and the interference pixel based on the grayscale threshold.

3. The interference suppression method based on dynamic threshold and edge notch according to claim 2, characterized in that: The ratio p of the number of pixels corresponding to each gray level to the total number of pixels i Expressed as: Among them, i represents the gray level, n i represents the number of pixels corresponding to each gray level, and N represents the total number of pixels in the time-frequency graph; The first weight and the second weight are respectively expressed as: Wherein, w0 represents the first weight; w1 represents the second weight; H represents the grayscale threshold parameter, the first grayscale range is [0, H], and the second grayscale range is [H+1, 255]; The first average grayscale value and the second average grayscale value are respectively expressed as: Here, μ0 represents the first average grayscale value, and μ1 represents the second average grayscale value. The objective function is expressed as: η(H)=(μ0-μ1)·w0·w1·(μ0-μ1); H * =argmax H∈[0,255] {η(H)}; Among them, H * Represents the gray level threshold obtained by solution, argmax H∈[0,255] {η(H)} represents the objective function, and η(H) represents an intermediate parameter.

4. The interference suppression method based on dynamic threshold and edge notch according to claim 3, characterized in that: Separating the target signal from the radar echo signal according to the time-frequency diagram and the modulation matrix includes: Separating an initial target pixel region and an initial interference pixel region from the time-frequency map according to the time-frequency map and the modulation matrix; Performing smoothing convolution processing and binarization processing on the initial interference pixel area to obtain an extended interference pixel area; Determining edge interference pixels in the initial target pixel area; Based on the edge interference pixels in the extended interference pixel area and the initial target pixel area, edge interference filtering is performed on the initial target pixel area to obtain a target signal corresponding to the target pixel area.

5. The interference suppression method based on dynamic threshold and edge notch according to claim 4, characterized in that: The step of filtering edge interference from the initial target pixel region based on the extended interference pixel region and the edge interference pixels in the initial target pixel region to obtain a target signal corresponding to the target pixel region includes: Filtering target edge interference pixels in the initial target pixel area to obtain a target time-frequency domain signal corresponding to the target pixel area; the target edge interference pixels are edge interference pixels located in the extended interference pixel area; Performing time domain transformation on the target time-frequency domain signal to obtain a target signal.

6. The interference suppression method based on dynamic threshold and edge notch according to claim 1, characterized in that: The performing time-frequency analysis on the radar echo signal to obtain a corresponding time-frequency graph includes: dividing the radar echo signal into a plurality of signal segments; Perform sliding windowing processing on each signal segment in parallel to obtain a windowed signal corresponding to each signal segment; Performing short-time Fourier transform on the windowed signal corresponding to each signal segment in parallel to obtain a time-frequency domain signal corresponding to each signal segment; A time-frequency diagram is generated based on the time-frequency domain signals corresponding to all signal segments.

7. An interference suppression device based on dynamic threshold and edge notch, characterized in that: Applicable to graphics processors (GPUs), including: A signal generating module is used to generate a simulated radar echo signal, wherein the simulated radar echo signal includes a simulated target signal and a simulated interference signal; The interference suppression module is used to perform time-frequency analysis on the simulated radar echo signal to obtain a corresponding simulated time-frequency diagram; based on the grayscale of each pixel in the simulated time-frequency diagram, determine a simulated grayscale threshold that satisfies the objective function, and the simulated grayscale threshold is used to separate the simulated target pixels corresponding to the simulated target signal and the simulated interference pixels corresponding to the simulated interference signal contained in the simulated time-frequency diagram; based on the simulated grayscale threshold and the grayscale of each pixel in the simulated time-frequency diagram, determine a simulated modulation matrix; according to the simulated time-frequency diagram and the simulated modulation matrix, separate the simulated target signal from the simulated radar echo signal.

8. The interference suppression device based on dynamic threshold and edge notch according to claim 7, characterized in that: The signal generation module includes a simulated target signal unit, a simulated interference signal unit and a signal fusion unit; The simulation target signal unit is used to generate multiple simulation signal data points according to the first preset parameters using multi-threaded parallelism to obtain the simulation target signal; The simulated interference signal unit is used to generate multiple simulated interference signals according to second preset parameters, where the first preset parameters are different from the second preset parameters, and the multiple simulated interference signals include at least a simulated slice interference signal, a simulated false target interference signal, a simulated swept frequency interference signal, and a simulated comb spectrum interference signal; The signal fusion unit is used to fuse the simulated target signal and the multiple simulated interference signals to obtain the simulated radar echo signal.

9. The interference suppression device based on dynamic threshold and edge notch according to claim 8, characterized in that: The process of the simulated interference signal unit generating the simulated comb spectrum interference signal includes: generating a plurality of time vectors based on the second preset parameter; Based on multiple time vectors and preset sub-signal parameters, multiple sub-interference signals are generated in parallel using multi-threading; the preset sub-signal parameters include the number, frequency and order of sub-signals; Multiple sub-interference signals are superimposed and processed in parallel using multi-threading to obtain a one-dimensional simulated comb spectrum interference signal; Doppler modulation is performed on multiple data points in the one-dimensional simulated comb spectrum interference signal in parallel using multiple threads to obtain the simulated comb spectrum interference signal.

10. The interference suppression device based on dynamic threshold and edge notch according to claim 8, characterized in that: The process of the simulated interference signal unit generating the simulated false target interference signal includes: Based on the second preset parameters, multiple time vectors, frequency vectors and interference signal windows are generated in parallel using multi-threading; Generate a linear frequency modulation signal based on multiple time vectors, frequency vectors and interference signal windows; Generate a false target pulse signal based on pre-set false target parameters and unit pulse signals; the false target parameters include the number of false targets, the time interval and intensity factor of the false target pulses; Phase modulation is performed on the linear frequency modulation signal and the false target pulse signal in parallel using multiple threads to obtain a one-dimensional simulated false target interference signal; Doppler modulation is performed on multiple data points in the one-dimensional simulated false target interference signal in parallel using multiple threads to obtain the simulated false target interference signal.

Images