Intermittent sampling and forwarding interference identification and suppression method and device based on instance segmentation

Abstract

The invention discloses an instance segmentation-based intermittent sampling forwarding interference identification and suppression method and device, and the method comprises the steps: inputting a time-frequency graph of an echo signal into a trained mask-based interference identification model to obtain prediction mask information of each signal, the prediction mask information comprises a mask position and a mask value, and the mask position and the mask value of each signal are obtained; mask values of different types of signals are different, and labels of sample interference signals and sample target signals in a sample mask time-frequency graph used for training the interference recognition model are made one by one; mapping the prediction mask information of the interference signal to a time-frequency domain to obtain time-frequency data of the interference signal; and suppressing the interference signal according to the time frequency data of the interference signal. According to the invention, interference signals can be identified one by one, and the interference identification suppression effect is enhanced.

Description

Technical Field

[0001] The present invention belongs to the technical field of signal processing, and in particular relates to a method and device for identifying and suppressing intermittent sampling and forwarding interference based on instance segmentation. Background Art

[0002] With the development of digital radio frequency memory, coherent false target jamming technology has been realized. Among them, the original intention of the ingenious intermittent sampling and forwarding jamming technology was to use the time-sharing of transmission and reception to solve the problem of antenna space isolation in engineering applications. Since intermittent sampling and forwarding jamming has the advantages of being coherent with the signal, fast response speed, diverse jamming patterns and rich modulation patterns, it is widely used in radar countermeasure technology research. In the research of radar anti-interference technology, radar interference identification and suppression technology in a single intermittent sampling and forwarding jamming scenario has made certain progress, but research in complex multi-intermittent sampling and forwarding jamming scenarios is still insufficient. In the increasingly complex jamming environment, interference identification and suppression technology in multiple jamming scenarios is in urgent need of development.

[0003] At present, the interference suppression technology for intermittent sampling and forwarding interference is mainly based on the time domain discontinuous characteristics of the interference. The radar receiving signal can be subjected to short-time Fourier transform, fractional order transform, short-time fractional order transform, etc. to obtain more obvious interference and target signal characteristics. Figure 1 , the distribution of radar echo signals in the time domain and frequency domain. It can be seen that in the time domain, the target and the interference, and the interference and the interference are mixed and cannot be clearly distinguished. Figure 2 After the time domain data is converted to the time-frequency domain, due to the time-sharing characteristics of intermittent sampling and forwarding interference, it appears as a discontinuous slice distribution in the time-frequency domain, which is significantly different from the continuous inclined target signal. Therefore, the interference can be suppressed according to its characteristics in the time-frequency domain.

[0004] Most of the current interference suppression methods based on the time-frequency domain identify interference signals based on the number of peaks of the echo signal in the time-frequency domain, the 3dB width of the peak, and the number of range gates with peaks. This method can only identify all interference signals as a whole, and cannot specifically identify which interference signal the time-frequency data belongs to; at the same time, the interference identification accuracy is low when the signal-to-noise ratio is lower than -5dB.

[0005] Therefore, the interference identification and suppression effects of current interference identification and suppression methods are relatively poor. Summary of the invention

[0006] The embodiments of the present invention provide an instance segmentation-based intermittent sampling forwarding interference identification and suppression method and device, which can solve the problem that the interference identification and suppression effects of the current interference identification and suppression methods are poor.

[0007] In a first aspect, an embodiment of the present invention provides a method for identifying and suppressing intermittent sampling and forwarding interference based on instance segmentation, the method comprising:

[0008] The time-frequency diagram of the echo signal is input into the trained mask-based interference recognition model to obtain the predicted mask information of each signal, wherein the predicted mask information includes the mask position and the mask value. The mask values ​​of different types of signals are different. The labels of the sample interference signal and the sample target signal in the sample mask time-frequency diagram used in the training interference recognition model are made one by one.

[0009] Mapping the predicted mask information of the interference signal to the time-frequency domain to obtain the time-frequency data of the interference signal;

[0010] The interference signal is suppressed according to the time-frequency data of the interference signal.

[0011] In a second aspect, an embodiment of the present invention provides an intermittent sampling forwarding interference identification and suppression device based on instance segmentation, comprising:

[0012] Interference identification module, the interference identification module includes a trained mask-based interference identification model, the interference identification model is used to input the time-frequency diagram of the echo signal into the trained mask-based interference identification model to obtain the predicted mask information of each signal, wherein the predicted mask information includes the mask position and the mask value, and the mask values ​​of different types of signals are different. The labels of the sample interference signal and the sample target signal in the sample mask time-frequency diagram used for training the interference identification model are prepared one by one;

[0013] An interference information processing module, the interference information processing module is used to map the predicted mask information of the interference signal to the time-frequency domain to obtain the time-frequency data of the interference signal;

[0014] The interference suppression module is used to suppress the interference signal according to the time-frequency data of the interference signal.

[0015] In a third aspect, an embodiment of the present invention provides an electronic device, comprising a processor and a memory, wherein the memory is used to store computer programs; the processor can be used to execute the computer program (instructions) stored in the memory to implement the method of the first aspect above.

[0016] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, in which a computer program is stored. When the computer program is executed, the method of the first aspect described above can be implemented.

[0017] Compared with the prior art, the embodiments of the present invention have the following beneficial effects: according to the method provided by the present invention, more accurate features of interference signals can be extracted through the interference recognition model, thereby improving the recognition accuracy of interference signals; by using samples whose information of each interference signal is known in training the interference recognition model, the interference recognition model can output prediction mask information for each interference signal to indicate time-frequency information when used, thereby realizing one-by-one recognition of interference signals and enhancing the recognition effect; thereby enhancing the suppression effect of interference signals. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 A schematic diagram of signal distribution of a radar echo signal in the time domain provided by an embodiment of the present invention;

[0019] Figure 2 A schematic diagram of signal distribution of a radar echo signal in the time-frequency domain provided by an embodiment of the present invention;

[0020] Figure 3 A flowchart of a method for training a mask-based interference recognition model provided by an embodiment of the present invention;

[0021] Figure 4 A sample time-frequency diagram and a sample mask time-frequency diagram provided by an embodiment of the present invention;

[0022] Figure 5 A schematic diagram of a training scenario of an interference recognition model provided by an embodiment of the present invention;

[0023] Figure 6 A flowchart for implementing an intermittent sampling forwarding interference identification and suppression method based on instance segmentation provided by an embodiment of the present invention;

[0024] Figure 7 A schematic diagram of a time-frequency diagram of an echo signal, a signal classification result, and predicted mask information provided by an embodiment of the present invention;

[0025] Figure 8 A schematic diagram of the structure of an intermittent sampling and forwarding interference identification and suppression device based on instance segmentation provided by an embodiment of the present invention;

[0026] Fig. 9 A schematic diagram of a pulse pressure result provided by an embodiment of the present invention;

[0027] Fig.10 A schematic diagram of another time-frequency diagram of an echo signal, a signal classification result, and predicted mask information provided by an embodiment of the present invention;

[0028] Fig.11 A schematic diagram of another pulse pressure result provided by an embodiment of the present invention;

[0029] Fig.12 A schematic diagram of interference suppression accuracy provided by an embodiment of the present invention;

[0030] Fig.13 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0031] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present invention. However, it should be clear to those skilled in the art that the present invention may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present invention.

[0032] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0033] It should also be understood that the term "and / or" used in the present description and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

[0034] As used in the present specification and the appended claims, the term "if" may be interpreted as "when" or "upon" or "in response to determining" or "in response to detecting", depending on the context. Similarly, the phrase "if it is determined" or "if [described condition or event] is detected" may be interpreted as meaning "upon determination" or "in response to determining" or "upon detection of [described condition or event]" or "in response to detecting [described condition or event]", depending on the context.

[0035] In addition, in the description of the present specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0036] References to "one embodiment" or "some embodiments" etc. described in the present specification mean that one or more embodiments of the present invention include specific features, structures or characteristics described in conjunction with the embodiment. Therefore, the statements "in one embodiment", "in some embodiments", "in some other embodiments", "in some other embodiments", etc. that appear in different places in this specification do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0037] The current interference identification and suppression method can perform interference identification and suppression through the following steps S201-S204.

[0038] S201, performing pulse compression processing and short-time Fourier transform processing on the echo signal to obtain time-frequency domain data of the echo signal.

[0039] S202, analyzing the number of branches M in each range gate based on the time-frequency domain data of the echo signal p , the 3dB width of the peak ΔF and the number of range gates N where the peak exists p .

[0040] S203: Identify interference signals according to the above parameters.

[0041] Specifically, if the 3dB width of the peak is approximately equal to the pulse width of the transmitted signal, it is determined to be the target signal. If the 3dB width of the peak is less than the pulse width of the transmitted signal and the number of peaks in each range gate is equal, the judgment is made based on the number of range gates with peaks. p =1, it is intermittent sampling direct forwarding interference, if N p >1, it is intermittent sampling repeated forwarding interference. If the 3dB width of the peak is less than the pulse width of the transmitted signal and the number of peaks in each range gate is not equal, it is intermittent sampling cyclic forwarding interference.

[0042] S204, determining the time dimension of the time-frequency domain where the interference signal and the target signal are located, and suppressing the interference based on a time-frequency domain filter.

[0043] Exemplarily, the time-frequency domain filter may satisfy the following formula:

[0044] F=[α1α2…α N ]

[0045] F is the time-frequency domain filter, α i is the column vector of the time-frequency domain filter in the i-th time dimension, and N is the number of sampling points in the time-frequency domain.

[0046] in:

[0047]

[0048] I is the unit column vector, r is the range gate where the target signal is located, p min is α i The smallest amplitude, t i is the column vector of the i-th time dimension.

[0049] Exemplarily, the time-frequency domain filter is multiplied by the time-frequency distribution of the pulse pressure result to achieve filtering. In this process, the signal of the time dimension of the target remains unchanged; while the signal output of the time dimension of the interference is equal to the minimum value of the column; thereby achieving interference suppression.

[0050] This interference identification method has low interference identification accuracy in a single intermittent sampling and forwarding interference scenario when the signal-to-noise ratio is lower than -5dB. In multiple intermittent sampling and forwarding interference scenarios, it is also difficult to identify each interference.

[0051] In view of this, the present invention provides an intermittent sampling forwarding interference identification and suppression method based on instance segmentation. Through the interference identification model, more precise features of the interference signal can be extracted to improve the recognition accuracy of the interference signal; by using samples whose information of each interference signal is known in the training interference identification model, the interference identification model can output the predicted mask information of each interference signal for indicating the time-frequency information when used, thereby realizing the identification of the interference signals one by one and enhancing the identification effect; thereby enhancing the suppression effect of the interference signal.

[0052] The present invention is further described in detail below with reference to specific embodiments, but the embodiments of the present invention are not limited thereto.

[0053] The wireless blockchain network sharding method provided in the embodiment of the present invention can be applied to electronic devices such as mobile terminals, personal laptops, supercomputers, etc. The embodiment of the present invention does not impose any restrictions on the specific type of electronic devices.

[0054] Figure 3 The flowchart shown is a method for training a mask-based interference recognition model according to an embodiment of the present invention. As an example but not a limitation, the training method may include steps S301-S306, which may be applied to the above electronic device. Each step is described below.

[0055] S301, extracting high-dimensional semantic information of the sample time-frequency graph and fusing the high-dimensional semantic information of the sample time-frequency graph with the low-dimensional semantic information of the sample time-frequency graph to obtain semantic features of the sample time-frequency graph.

[0056] Exemplarily, the sample time-frequency diagram may include time-frequency data of at least one sample interference signal.

[0057] For example, see Figure 4 The sample time-frequency diagram in (a) includes three intermittently sampled interference signals.

[0058] Exemplarily, different labels are created for each signal in the sample echo signal, for example, the target is marked as 1, the noise is marked as 0, and the interference is marked as 2, and then different signals are marked with labels in the sample time-frequency diagram of the sample echo signal to obtain a true sample mask time-frequency diagram.

[0059] Generally, the number of noise signals and target signals is 1, and the number of interference signals can be a positive integer n. j , then the sample time-frequency diagram includes n all =2+n j A signal.

[0060] For example, see Figure 4 The sample mask time-frequency diagram in (b) is a label mask made according to the label of each signal. Different label values ​​have different label mask color values, and slices of different colors represent different signals. The noise is blue-purple and merged into the background. The three interference signals are light blue, dark yellow and light yellow respectively, and the target is dark blue.

[0061] In a possible implementation, the sample time-frequency diagram may be drawn based on two-dimensional time-frequency domain data of the echo signal, where the two-dimensional time-frequency domain data is obtained by performing short-time Fourier transform on the sample echo signal.

[0062] Exemplarily, the two-dimensional time-frequency domain data may satisfy the following formula:

[0063]

[0064] Among them, STFT(n,f) is the two-dimensional time-frequency domain data, t0 is the sliding window interval, n is the number of sliding window times, f is the frequency, S r (t) is the sample echo signal, w(t-nt0) is the window function, and j is the imaginary unit.

[0065] in:

[0066] S r (t) = S e (t)+S j (t)+n0(t)

[0067]

[0068] S e (t) is the sample target signal in the sample echo signal, Sj (t) is the interference part, which is composed of at least one sample interference signal, n0(t) is the noise in the sample echo signal, t1 is the initial start time of the sliding window, t2 is the initial end time of the sliding window, and Δt=t2-t1 is the window length of the window function.

[0069] In one example, the sample transmission signal for generating the sample echo signal may be a linear frequency modulation signal.

[0070] Exemplarily, the sample transmission signal may satisfy the following formula:

[0071]

[0072] Where S(t) is the sample emission signal, Represents a rectangular window function, which represents the independent variable t in The function value within the range is 1, and the function value in the rest of the range is 0; k = B / T P is the modulation frequency, B is the bandwidth of the sample transmission signal, T p is the pulse width of the sample transmission signal.

[0073] In one possible implementation, see Figure 5 The interference recognition model can be a Mask Region-based Convolutional Neural Network (Mask R-CNN), which can include a backbone feature extraction network (Backbone), a region generation network (Region Proposal Network, RPN), a proposal layer (proposals), a pooling layer (Region of Interest Alignment, ROIAlign) and three prediction branches.

[0074] In one example, see Figure 5 , the backbone feature extraction network can include a residual network (ResNet-50) and a feature pyramid (FPN).

[0075] Exemplarily, the sample time-frequency graph can be input into the interference recognition model, and the residual network can perform encoder dimensionality reduction on the m-1th level semantic information to extract high-dimensional semantic information to obtain the mth level semantic information, and finally obtain the M-1th level semantic information. Then, the feature pyramid can start from the 3rd level semantic information, upsample the mth level semantic information, and fuse it with the m-1th level semantic information to obtain the m-1th level semantic feature, upsample the Mth level semantic information to obtain the Mth level semantic feature, perform maximum pooling on the Mth level semantic feature to obtain the M+1th level semantic feature, and obtain the Mth level semantic feature in total.

[0076] Exemplarily, m is a positive integer greater than or equal to 2 and less than or equal to M. For example, M may be equal to 5.

[0077] Exemplarily, the semantic information of level 2 to level M may be feature maps of different dimensions, for example Figure 5 The first level semantic information can be a sample time-frequency graph.

[0078] Exemplarily, the high dimension and low dimension of semantic information are relative. In the m-1th level semantic information and the mth level semantic information, the m-1th level semantic information is low-dimensional semantic information, and the mth level semantic information is high-dimensional semantic information.

[0079] For example, the semantic feature can be a feature layer, see Figure 5 P2 to P6 in.

[0080] S302, predicting based on the semantic features of the sample time-frequency graph to obtain changes in multiple prediction boxes and the confidence level of the signal contained in each prediction box.

[0081] For example, see Figure 5 , the semantic features of the sample time-frequency graph can be input into the RPN, and one branch passes through the convolutional layer (e.g. Figure 5 The orange filled part in the figure) and the normalized exponential function softmax are used to obtain the confidence that each prediction box at each pixel on each feature layer contains a signal. The other branch uses the convolution layer and the bounding box regression algorithm (bbox reg) to obtain the change of each prediction box at each pixel on the feature layer.

[0082] S303, screening and correcting the predicted box according to the confidence level and the change of the predicted box to obtain a suggested box.

[0083] For example, the prediction box can be input into the suggestion layer. The suggestion layer can first select the prediction boxes with the top 6000 confidence probabilities, and then correct the positions of the prediction boxes according to the changes in these prediction boxes, and discard the prediction boxes whose corrected boxes exceed the size of the sample time-frequency graph. Finally, the softmax function is used to obtain the prediction box information with the top 2000 probabilities, and the intersection and union ratio of these 2000 prediction box information and the corresponding true box information is calculated. If it is greater than 0.5, it is set as a positive sample, otherwise it is set as a negative sample. Then, 200 samples are selected as suggestion boxes according to the ratio of positive to negative samples of 1:2.

[0084] S304, pooling the semantic features of the sample time-frequency graph and the sample time-frequency graph in the suggestion box to obtain semantic features of a preset size and a local sample time-frequency graph of a preset size.

[0085] Exemplarily, the semantic features and the sample time-frequency diagram in the suggestion box may be pooled through a pooling layer to obtain semantic features of a preset size and a local sample time-frequency diagram of a preset size.

[0086] S305 , obtaining prediction mask information of each sample signal based on the semantic features of a preset size and the local sample frequency spectrum of a predicted size.

[0087] Exemplarily, the predicted mask information may include a mask position and a mask value, and different types of signals have different mask values. The type of the signal may be indicated by the mask value.

[0088] For example, see Figure 5 , the semantic features of the preset size and the local sample time-frequency map of the predicted size can be input into the three prediction branches respectively, the classification information of each sample signal is predicted by the fully connected layer and the softmax function in the classification branch, the position of each signal, that is, the position of the bounding box of the signal (that is, the position of the mask) is predicted by the fully connected layer and the bounding box regression algorithm in the regression branch, and the mask value of each signal is obtained by the deconvolution layer in the mask branch.

[0089] S306, determining a total loss according to the predicted mask information and the real mask information of the sample echo signal, and training an interference recognition model according to the total loss to obtain a trained interference recognition model.

[0090] Exemplarily, the predicted mask information of the sample echo signal and the real mask information in the sample mask frequency graph can be compared to obtain the total loss of the model, and then the interference recognition model is trained according to the total loss based on the gradient descent algorithm to obtain a trained interference recognition model. Since the interference recognition model recognizes interference signals based on instance segmentation and calculates the total loss based on the sample mask frequency graph in which each interference signal is respectively annotated, this can improve the recognition accuracy of the interference recognition model for interference signals and make it possible to recognize interference signals one by one.

[0091] Exemplarily, after obtaining the trained interference recognition model, the sample time-frequency diagram in the test set can also be input into the trained interference recognition model for recognition to obtain predicted mask information. The predicted mask information of the sample time-frequency diagram in the test set and the actual mask information in the sample mask time-frequency diagram are used to determine whether the model is successfully trained. If the two are the same, the training is successful; if they are different, the training fails.

[0092] Optionally, the ratio of the sample time-frequency graphs in the training set, the validation set, and the test set may be 4:1:5.

[0093] Figure 6The flowchart shown is an implementation flow of an example segmentation-based intermittent sampling forwarding interference identification and suppression method provided by an embodiment of the present invention. The interference identification and suppression method can be applied to the above electronic device, which can include steps S601-S603, and each step is described below.

[0094] S601, inputting the time-frequency diagram of the echo signal into the trained mask-based interference recognition model to obtain the predicted mask information of each signal.

[0095] Exemplarily, the interference recognition model can be trained according to the above training method. During use, the operation process of each module of the interference recognition model is similar to that during training, except that the sample time-frequency diagram is input during training, and the time-frequency diagram of the echo signal to be detected is input during use. The specific operation process of the interference recognition model can be found in the description of steps S301-S305 in the above training method, which will not be repeated here.

[0096] Exemplarily, the interference identification model can be based on the time-frequency diagram of the echo signal (see Figure 7 (a) in the figure), and obtain the classification result of each signal (see Figure 7 (b) in the figure), and according to the classification results, different types of signals are assigned different mask values ​​on the time-frequency diagram to obtain predicted mask information (see Figure 7 (c) in the figure.

[0097] For example, see Figure 7 In (b), the purple frame is the interference signal, and the red frame is the target signal. Different mask values ​​are assigned according to the category of each signal, and the mask values ​​are reflected by different colors to obtain Figure 7 (c) in the figure. The light blue, dark yellow, and light yellow areas are the areas where the three interference signals are located, and the dark blue line is the area where the target signal is located.

[0098] S602, mapping the prediction mask information of the interference signal to the time-frequency domain to obtain the time-frequency data of the interference signal.

[0099] Exemplarily, the mask position of the signal is the position of the mask on the time-frequency diagram, and the time-frequency domain information of the signal can be obtained by mapping the mask to the time-frequency domain.

[0100] S603: Suppress the interference signal according to the time-frequency data of the interference signal.

[0101] In one example, a corresponding filter may be produced according to the time-frequency data of the interference signal, and the interference may be filtered out by the filter.

[0102] In an example, the position where the mask of the interference signal is located may be set to 0 on the time-frequency diagram of the echo signal, so as to suppress the interference and obtain the time-frequency diagram of the echo signal after interference suppression.

[0103] According to the interference identification and suppression method provided by the present invention, more precise features of interference signals can be extracted through the interference identification model, thereby improving the recognition accuracy of interference signals; by using samples whose information of each interference signal is known in training the interference identification model, the interference identification model can output prediction mask information of each interference signal for indicating time-frequency information when used, thereby realizing one-by-one identification of interference signals and enhancing the recognition effect; thereby enhancing the suppression effect of interference signals.

[0104] Figure 8 The structure diagram of an intermittent sampling and forwarding interference identification and suppression device based on instance segmentation provided by an embodiment of the present invention is shown. As an example but not limitation, the device 800 may include an interference identification module 810, an interference information processing module 820 and an interference suppression module 830.

[0105] The interference identification module 810 includes a trained mask-based interference identification model, which is used to input the time-frequency diagram of the echo signal into the trained mask-based interference identification model to obtain the predicted mask information of each signal, wherein the predicted mask information includes the mask position and the mask value, and the mask values ​​of different signals are different. The labels of the sample interference signal and the sample target signal in the sample mask time-frequency diagram used by the training interference identification model are made one by one;

[0106] The interference information processing module 820 is used to map the prediction mask information of the interference signal to the time-frequency domain to obtain the time-frequency data of the interference signal;

[0107] The interference suppression module 830 is used to suppress the interference signal according to the time-frequency data of the interference signal.

[0108] In order to better illustrate the beneficial effects of the interference identification and suppression method provided by the present invention, the following simulation experiment was carried out.

[0109] Fig. 9 Shown is a schematic diagram of a pulse pressure result provided by an embodiment of the present invention.

[0110] Exemplarily, the echo signal in this experiment is an echo signal with three intermittent sampling interference signals under the condition of 30 dB.

[0111] See also Fig. 9 , Fig. 9 (a) is the pulse pressure result of the echo signal. Fig. 9 (b) in the figure is the pulse pressure result of the echo signal after interference suppression.

[0112] Based on the formula: JIR=p_jam-a_jam, where p_jam is the interference power of the echo signal, and a_jam is the interference power of the echo signal after interference suppression; the interference suppression ratio JIR is calculated according to the pulse pressure results before and after interference suppression, and the JIR is 31.3363dB.

[0113] Fig.11 Shown is a schematic diagram of another pulse pressure result provided by an embodiment of the present invention.

[0114] For example, this experiment was conducted under the condition of -10dB signal-to-noise ratio and three interferences. The time-frequency diagram of the echo signal input in this experiment is as follows: Fig.10 As shown in (a) in the figure, the pulse pressure result of the echo signal is as follows Fig.11 As shown in (a) of the figure, the classification results and predicted mask information predicted by the interference recognition model are respectively as follows: Fig.10 The pulse pressure result of the echo signal after interference suppression is obtained by performing interference suppression according to the recognition result of the model as shown in (b) and (c). Fig.11 As shown in (b) in .

[0115] After calculation, it can be obtained that the interference suppression ratio of this experiment reaches 27.1318%.

[0116] Fig.12 Shown is a schematic diagram of interference suppression accuracy provided by an embodiment of the present invention.

[0117] See also Fig.12 ,It can be seen that when the interference-to-signal ratio is set to 20dB and the signal-to-noise ratio is set to -10dB~10dB (with 5dB as a step), the interference identification accuracy reaches more than 94%.

[0118] According to the interference identification and suppression method provided by the present invention, more precise features of interference signals can be extracted through the interference identification model, thereby improving the recognition accuracy of interference signals; by using samples whose information of each interference signal is known in training the interference identification model, the interference identification model can output prediction mask information of each interference signal for indicating time-frequency information when used, thereby realizing one-by-one identification of interference signals and enhancing the recognition effect; thereby enhancing the suppression effect of interference signals.

[0119] Fig.13 FIG. 1 is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. Fig.13 The electronic device 1300 shown may include: at least one processor 1310 ( Fig.13Only one processor is shown in the figure), a memory 1320, and a computer program 1330 stored in the memory 1320 and executable on the at least one processor 1310, wherein the processor 1310 implements the steps of any of the above-mentioned method embodiments when executing the computer program 1330.

[0120] The electronic device 1300 may be a processing device such as a robot that can implement the above method. The embodiment of the present invention does not impose any limitation on the specific type of the electronic device.

[0121] Those skilled in the art will understand that Fig.13 The electronic device 1300 is merely an example and does not constitute a limitation on the electronic device. The electronic device 1300 may include more or fewer components than shown in the figure, or may combine certain components, or may include different components. For example, the electronic device 1300 may also include an input and output interface.

[0122] The processor 1310 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASTC), field programmable gate arrays (FPGA), or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0123] The memory 1320 may be an internal storage unit in some embodiments, such as a hard disk or a memory. The memory 1320 may also be an external storage device in other embodiments, such as a plug-in hard disk, a smart memory card (SmartMemory Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. Further, the memory 1320 may also include both an internal storage unit and an external storage device. The memory 1320 is used to store an operating system, an application, a boot loader (Boot Loader), data, and other programs, such as the program code of the computer program, etc. The memory 1320 may also be used to temporarily store data that has been output or is to be output.

[0124] It should be understood that the order of execution of the steps in the above embodiment does not necessarily mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present invention.

[0125] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated into a processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0126] An embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in the above-mentioned method embodiments can be implemented.

[0127] An embodiment of the present invention provides a computer program product. When the computer program product runs on an electronic device, the electronic device can implement the steps in the above-mentioned method embodiments when executing the computer program product.

[0128] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, which can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, the steps of the above-mentioned various method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium can at least include: any entity or device that can carry the computer program code to the camera / terminal device, recording medium, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, RandomAccess Memory), electric carrier signal, telecommunication signal and software distribution medium. For example, a USB flash drive, a mobile hard disk, a magnetic disk or an optical disk. In some jurisdictions, according to legislation and patent practice, computer-readable media cannot be electric carrier signals and telecommunication signals.

[0129] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0130] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

Claims

1. A method for identifying and suppressing intermittent sampling forwarding interference based on instance segmentation, characterized in that: include: Inputting the time-frequency diagram of the echo signal into the trained mask-based interference recognition model to obtain the predicted mask information of each signal, wherein the predicted mask information includes the mask position and the mask value, and the mask values ​​of different signals are different. The labels of the sample interference signal and the sample target signal in the sample mask time-frequency diagram used to train the interference recognition model are made one by one; Mapping the predicted mask information of the interference signal to the time-frequency domain to obtain the time-frequency data of the interference signal; The interference signal is suppressed according to the time-frequency data of the interference signal.

2. The method according to claim 1, characterized in that The interference identification model is used to: Extracting high-dimensional semantic information of the time-frequency graph and fusing the high-dimensional semantic information with the low-dimensional semantic information of the time-frequency graph to obtain semantic features of the time-frequency graph; Predicting based on the semantic features of the time-frequency graph to obtain changes in multiple prediction boxes and the confidence of the signal contained in each prediction box; The prediction box is screened and modified according to the confidence and the change of the prediction box to obtain a suggestion box; Pooling the semantic features of the time-frequency graph and the time-frequency graph in the suggestion box to obtain semantic features of a preset size and a local time-frequency graph of a preset size; Prediction mask information of each signal is obtained based on the semantic features of the preset size and the local time-frequency diagram of the preset size.

3. The method according to claim 2, characterized in that The step of extracting high-dimensional semantic information of the time-frequency graph and fusing the high-dimensional semantic information with the low-dimensional semantic information of the time-frequency graph to obtain semantic features of the time-frequency graph includes: Performing encoder dimensionality reduction on the m-1th level semantic information to obtain the m-th level semantic information, wherein m is a positive integer greater than or equal to 2 and less than or equal to M, and between the m-1th level semantic information and the m-1th level semantic information, the m-1th level semantic information is low-dimensional semantic information, and the m-th level semantic information is high-dimensional semantic information; Starting from the third level semantic information, upsampling the mth level semantic information and fusing it with the m-1th level semantic information to obtain the m-1th level semantic feature; Upsampling the M-th level semantic information to obtain the M-th level semantic feature; The Mth level semantic features are pooled to obtain the M+1th level semantic features.

4. The method according to claim 1, characterized in that: The sample masked time-frequency diagram is obtained by adding the label mask of each signal to the sample time-frequency diagram, and the sample time-frequency diagram is drawn based on the two-dimensional time-frequency domain data of the sample echo signal, and the two-dimensional time-frequency domain data is obtained by performing short-time Fourier transform on the sample echo signal.

5. The method according to claim 4, characterized in that The two-dimensional time-frequency domain data satisfies the following formula: Wherein, STFT(n,f) is the two-dimensional time-frequency domain data, t0 is the sliding window interval, n is the number of sliding window times, f is the frequency, S r (t) is the sample echo signal, w(t-nt0) is the window function, and j is the imaginary unit.

6. The method according to claim 4, characterized in that The sample transmission signal for generating the sample echo signal is a linear frequency modulation signal.

7. The method according to claim 1, characterized in that The suppressing the interference signal according to the time-frequency data of the interference signal includes: The time-frequency data of the interference signal is set to zero in the time-frequency data of the echo signal to suppress the interference signal.

8. An intermittent sampling forwarding interference identification and suppression device based on instance segmentation, characterized in that: include: An interference identification module, wherein the interference identification module includes a trained mask-based interference identification model, wherein the interference identification model is used to input the time-frequency diagram of the echo signal into the trained mask-based interference identification model to obtain predicted mask information of each signal, wherein the predicted mask information includes a mask position and a mask value, and the mask values ​​of different signals are different. The labels of the sample interference signal and the sample target signal in the sample time-frequency diagram used to train the interference identification model are prepared one by one; An interference information processing module, the interference information processing module is used to map the prediction mask information of the interference signal to the time-frequency domain to obtain the time-frequency data of the interference signal; An interference suppression module, wherein the interference suppression module is used to suppress the interference signal according to the time-frequency data of the interference signal.

9. An electronic device comprising a memory, a processor and a computer program stored in the memory, characterized in that: When the processor executes the computer program, the method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by the electronic device, the method according to any one of claims 1 to 7 is implemented.

Images