Weak target detection method based on inversion filtering

Abstract

The present invention proposes a weak target detection method based on inversion filtering, which is mainly used to solve the problem of detecting weak targets obscured by strong targets. The implementation steps are: performing pulse compression processing without adding a window on radar echo data; performing constant false alarm detection on the echo data after pulse compression to obtain a strong target set; using radar transmission parameter information to construct a subdivision time matrix after subdivision sampling, and using the subdivision time matrix and strong target information to construct an inversion data matrix for each strong target; constructing a weight vector of the inversion filter for each distance unit; using the weight vector of the inversion filter to filter the data after pulse compression; performing constant false alarm detection on the filtered data to obtain the distance unit where the weak target is located. The present invention can accurately detect weak targets obscured by strong targets and has strong robustness.

Description

Technical Field

[0001] The present invention belongs to the field of radar technology, and further relates to a weak target detection method based on inversion filtering in the field of radar target detection technology. The present invention is applied to effectively extract weak targets that are obscured by strong targets and difficult to detect from pulse pressure echo data. Background Art

[0002] With the development of modern stealth aircraft and UAV technology, such targets are often difficult to detect weak targets in actual radar processing due to their small radar cross section (RCS) and low detection signal-to-noise ratio. Therefore, in order to effectively detect weak targets, it is necessary to improve the radar detection signal-to-noise ratio and suppress the masking of weak targets by strong targets. The traditional CFAR (Constant False Alarm Rate, CFAR) constant false alarm detection method can effectively suppress the influence of strong target range sidelobes on weak targets by processing echo data using windowed pulse compression coefficients. However, windowing processing has the problem of target signal-to-noise ratio loss. The sidelobes of pulse compression output without windowing are easily aliased with the peak of weak targets, causing strong targets to mask weak targets, making it impossible to detect weak targets.

[0003] Nanjing University of Aeronautics and Astronautics proposed a radar weak target detection method based on pre-detection estimation in its patent document "Radar target detection method based on pre-detection estimation" (patent application number: 202010736841.4, application publication number: CN 111965615 A). The specific steps of this method are as follows: first, the prior information of the target of interest is obtained by unwindowed pulse compression, and then the distance parameters of the target are pre-estimated using the prior information. Subsequently, a dimensionality reduction observation model of the received signal is established through the obtained pre-estimated information; reconstruction is performed through a sparse recovery algorithm; and the reconstruction result is used to detect the target through a generalized likelihood ratio detector. This method can effectively improve the performance of weak target detection. However, the method still has the disadvantage that the sparse recovery algorithm needs to estimate the sparsity. When the number of targets is unknown, the estimation of the sparsity will have a certain deviation, so the algorithm will produce random pseudo-peaks. In the actual engineering application process, it will affect the performance of weak target detection, thereby detecting false targets.

[0004] The Air Force Engineering University of the Chinese People's Liberation Army has disclosed a radar weak target detection method using a clustering extrapolation method in its patent document "Slow and weak target detection method based on clustering extrapolation" (patent application number: 202110298948.X, application publication number: CN 113093180 A). The specific steps of this method are as follows: first, the echo is processed by MTI and MTD, and then the noise and clutter are removed using the multi-frame MTD peak detection results, and then the target detection is achieved using the extrapolation clustering method. This method can effectively detect slow targets under low signal-to-noise ratio conditions. However, the method still has the disadvantage that since this method only considers the impact of noise and clutter on weak targets, for weak targets near strong targets, this method does not suppress the impact of strong targets, so there is missed detection in the detection stage, which causes the radar system to be unable to detect the target. Summary of the invention

[0005] The purpose of the present invention is to propose a weak target detection method based on inversion filtering in view of the deficiencies of the above-mentioned prior art, so as to solve the problems of missed detection in constant false alarm detection of the existing clustering extrapolation method, and the problems of pseudo peaks and insufficient robustness of the sparse recovery algorithm.

[0006] The technical solution for realizing the purpose of the present invention is that the present invention extracts strong target information from pulse pressure echo data, uses the obtained strong target information and subdivided time matrix to construct inversion data of strong targets, uses inversion data to construct inversion filter, and performs filtering on the processed data, effectively suppresses the sidelobe influence of strong targets, improves the robustness of weak target detection, thereby avoiding the problem of missed detection in constant false alarm detection of existing clustering extrapolation methods. Due to the filtering processing performed by the present invention on radar echo data, the pseudo peak problem existing in sparse recovery algorithms is avoided. Finally, the filtered data is subjected to constant false alarm detection, so that weak targets can be detected.

[0007] The implementation steps of the present invention are as follows:

[0008] Step 1: Pulse compression of radar echo data:

[0009] The radar received echo data is pulse compressed by using the pulse compression coefficient without adding a window to obtain the radar echo data after pulse compression;

[0010] Step 2: Perform constant false alarm detection on the echo data after radar pulse compression:

[0011] The unit average constant false alarm detection method is adopted. The amplitude of a distance unit higher than its detection threshold is regarded as a strong target. The distance unit where each strong target is located is recorded, and the recorded distance units are combined into a strong target set.

[0012] Step 3, construct the inversion data matrix of each strong target:

[0013] Step 3.1, calculate the value of each element in each column of the segmented time matrix, and form all columns into a segmented time matrix;

[0014] Step 3.2, calculating the inversion value of each element in the subdivided time matrix, and using the inversion value of the element to update the corresponding element value in the subdivided time matrix;

[0015] Step 3.3, select a strong target from the detection target set, construct an M×N inversion data matrix, each element of the matrix is ​​initially 0, take the position of the selected strong target in the echo data after pulse compression as the starting point, put each column of the updated subdivision time matrix into the matrix, and obtain the inversion data matrix of the selected strong target, where the value of M is equal to the length of the echo data, and the value of N is equal to the number of columns of the subdivision time matrix;

[0016] Step 3.4, determine whether all strong targets in the strong target set have been selected, if so, execute step 4 after obtaining the inversion data matrix of all strong targets, otherwise, execute step 3.1;

[0017] Step 4: Use the constructed inversion filter weight vector to filter the radar echo data:

[0018] Step 4.1, randomly select a strong target that has not been selected;

[0019] Step 4.2, randomly select an unselected distance unit from the l distance units on both sides of the distance unit where the strong target is located, where the value of l is determined by the radar system parameters;

[0020] Step 4.3, taking the position of the selected range unit in the post-pulse compression echo data as the starting point and the length of the discrete sequence emitted by the radar in one pulse time as the interception length, intercept each column in the inversion data matrix corresponding to the selected strong target, and use all elements in the interception length as each column of the inversion matrix;

[0021] Step 4.4, calculate the covariance matrix of the inversion matrix;

[0022] Step 4.5, calculating the weight vector of the inversion filter corresponding to the selected distance unit;

[0023] Step 4.6, taking the position of the pre-pulse pressure echo data corresponding to the selected distance unit as the starting position, intercepting a section of data equal to the length of the inversion filter weight vector in the pre-pulse pressure echo data as the data to be filtered;

[0024] Step 4.7, multiplying the weight vector of the inversion filter corresponding to the selected distance unit with the data to be filtered to obtain the filtered data of the selected distance unit of the radar echo data;

[0025] Step 4.8, determine whether the l distance units on both sides of the currently selected strong target have been selected, if so, execute step 4.9, otherwise, execute step 4.2;

[0026] Step 4.9, determine whether all strong targets in the radar pulse compression front echo data have been selected, if so, execute step 5, otherwise, execute step 4.1;

[0027] Step 5: Perform constant false alarm detection on the filtered data:

[0028] The same method as the unit average constant false alarm detection in step 2 is used to detect the filtered data of each distance unit to obtain the detection threshold of the distance unit. The amplitude of the distance unit greater than or equal to its detection threshold is regarded as a weak target, and the distance units where all weak targets are located are recorded.

[0029] Compared with the prior art, the present invention has the following advantages:

[0030] First, when there is a strong target covering a weak target in the radar echo data, the present invention constructs an inversion data matrix by using the distance information of the strong target. The inversion filter constructed by the inversion data matrix suppresses the influence of the strong target side lobe, overcomes the shortcoming of missed detection caused by the clustering extrapolation method in the prior art, and enables the present invention to effectively detect the weak target obscured by the strong target, thereby improving the accuracy of weak target detection.

[0031] Second, because the present invention filters the radar echo data when detecting a weak target obscured by a strong target, it overcomes the defect of the prior art that a sparse recovery algorithm produces pseudo peaks, so that the present invention can accurately detect weak targets and reduce the probability of misjudgment of weak target detection.

[0032] Third, because the present invention constructs a subdivision time matrix to subdivide the time interval of the echo data when detecting weak targets, it can better correspond to the situation in which the position of the strong target and the distance unit do not correspond accurately in actual engineering, and overcome the defect of the sparse recovery algorithm used in the prior art that the robustness is not high, so that the present invention can accurately detect weak targets and improve the robustness of weak target detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 It is a flow chart of the implementation of the present invention;

[0034] FIG. 2 is a simulation diagram of the present invention. DETAILED DESCRIPTION

[0035] The present invention is further described in detail below in conjunction with the accompanying drawings and embodiments.

[0036] Reference Figure 1 The implementation steps of the present invention are further described in detail in the following embodiments.

[0037] Step 1: Perform pulse compression processing on the radar echo signal.

[0038] The pulse compression coefficient without windowing is used to perform pulse compression processing on the radar received echo data to obtain the echo data after radar pulse compression and the signal-to-noise ratio gain of pulse compression. The detailed steps are as follows:

[0039] Step 1.1, construct the unwindowed pulse compression coefficient of the radar transmit signal according to the radar transmit signal:

[0040]

[0041] Where h represents the unwindowed pulse compression coefficient of the radar transmission signal, represents the i-th element of s, i = 1, 2, ..., I, I represents the total length of the discrete sequence emitted by the radar in one pulse time, Indicates the rounding operation, s=[s1,s2,...,s I ] represents a discrete sequence of radar pulse time transmission, S(t) represents the radar transmission waveform function, Δt=1 / f s represents the radar sampling time interval, f s represents the sampling rate of the radar, T p represents the pulse width of the radar, (·) * represents the conjugation operation.

[0042] Step 1.2, by convolving the radar echo data with the unwindowed pulse compression coefficient, the radar echo data after pulse compression is obtained:

[0043]

[0044] Among them, y represents the radar echo data after pulse compression, represents the convolution operation, and r represents the echo data received by the radar.

[0045] Step 2: adopt the unit average constant false alarm detection algorithm and use the echo data after radar pulse compression to perform constant false alarm detection.

[0046] Step 2.1, arbitrarily select an unselected distance unit in the radar pulse compression echo data, take the distance units with a length of U on both sides of the selected distance unit as protection units, take the reference window length as 2F, and take the reference unit as the distance unit with a length of FU outside the protection unit;

[0047] Step 2.2, calculate the average value of the noise of all reference cells in the selected distance cell, and use it as the noise estimation value of the selected distance cell.

[0048] Step 2.3, calculate the threshold factor of the constant false alarm detection of the selected distance unit according to the following formula:

[0049]

[0050] Among them, α represents the threshold factor of the constant false alarm detection of the selected distance unit, In represents the logarithmic operation with the natural constant e as the base, and p f Represents the false alarm probability of the selected distance unit under Gaussian white noise.

[0051] Step 2.4, multiplying the threshold factor of the constant false alarm detection of the selected distance unit by the noise estimation value to obtain the detection threshold of the selected distance unit.

[0052] Step 2.5, determine whether the absolute value of the amplitude under the selected distance unit is greater than or equal to the detection threshold. If so, determine that there is a target under the selected distance unit and record the position of the selected distance unit and then execute step 2.6. Otherwise, it is considered that there is no target under the selected distance unit and execute step 2.1.

[0053] Step 2.6, determine whether all distance units in the radar pulse compression echo data have been selected, if so, execute step, otherwise, execute step 2.1.

[0054] Step 3: construct the inversion data matrix for each strong target.

[0055] Step 3.1, calculate the value of each element in each column of the segmented time matrix according to the following formula, and form all columns into a segmented time matrix:

[0056]

[0057] Among them, D k Represents the value of each element in the kth column of the subdivision time matrix, k = 1, 2, ..., K, K represents the total number of radar sampling interval subdivisions, and the total number of subdivision time matrix rows is equal to the total length of the discrete sequence emitted by the radar in one pulse time.

[0058] In the embodiment of the present invention, the radar pulse width T p =40μs, the sampling rate of radar receiving echo data is f s=1MHz, K=5, Δt=1 / f s =1μs, the unit of each element of the subdivision time matrix is ​​microseconds, and the subdivision time matrix is ​​as follows:

[0059]

[0060] In actual engineering projects, the actual position of strong targets in radar echo data is not strictly aligned with the sampling position. The starting time is shifted forward by half a sampling unit, the radar sampling interval is subdivided, and a subdivided time matrix is ​​constructed to ensure that the inversion data is consistent with the actual strong target echo data, thereby improving the robustness of weak target detection.

[0061] Step 3.2, calculate the inversion value of each element in the subdivided time matrix according to the following formula, and use the inversion value of the element to update the corresponding element value in the subdivided time matrix. The calculation formula of the inversion value is as follows:

[0062] O mn =S(D mn )

[0063] Among them, O mn represents the inversion value of the mth row and nth column in the updated subdivision time matrix, D mn Represents the element value in the mth row and nth column of the segmented time matrix.

[0064] Step 3.3, select a strong target from the detection target set, construct an M×N inversion data matrix, each element of the matrix is ​​initially 0, and take the position of the selected strong target in the echo data after pulse compression as the starting point, put each column of the updated subdivision time matrix into the matrix, and obtain the inversion data matrix of the selected strong target, where M is the echo data length and N is the number of columns of the subdivision time matrix.

[0065] In an embodiment of the present invention, a strong target in a strong target set is randomly selected, d represents the position of the selected strong target in the echo data after pulse compression, and in the first column of the inversion data matrix, the data after the dth row are sequentially placed in each element of the first column of the updated subdivision time matrix, and so on, in the kth column of the inversion data matrix, the data after the dth row are sequentially placed in each element of the kth column of the updated subdivision time matrix.

[0066] Step 3.4, determine whether all strong targets in the strong target set have been selected. If so, the inversion data matrix of all strong targets is obtained, and execute step 4; otherwise, execute step 3.1.

[0067] Step 4: Filter the radar echo data using the weight vector of the constructed inversion filter.

[0068] Step 4.1, randomly select a strong target that has not been selected;

[0069] Step 4.2, randomly select an unselected distance unit from l distance units on both sides of the distance unit where the strong target is located, wherein the value of l is determined by the radar system parameters, and in the embodiment of the present invention, l=20.

[0070] Step 4.3, taking the position of the selected distance unit in the post-pulse compression echo data as the starting point and the length of the discrete sequence emitted by the radar in one pulse time as the truncation length, truncate each column in the inversion data matrix corresponding to the selected strong target, and use all elements in the truncation length as each column of the inversion matrix.

[0071] In the embodiment of the present invention, the distance unit where the selected strong target is located is 20, the 30th distance unit in the radar pulse pressure echo data is selected, and the length of the discrete sequence emitted by the radar in one pulse time is 40. In the inversion data matrix corresponding to the selected strong target, the data from the 30th row to the 69th row of the first column is intercepted, and all elements in the intercepted length are used as the first column of the inversion matrix, and so on, for the kth column of the matrix, the data from the 30th row to the 69th row of the inversion data matrix is ​​intercepted, and all elements in the intercepted length are used as the kth column of the inversion matrix.

[0072] Step 4.4, solve the covariance matrix of the inversion matrix according to the following formula:

[0073] R=WW H / K+δE I

[0074] Where R represents the covariance matrix of the inversion matrix, W represents the inversion matrix, (·) H represents the conjugate transpose operation, δ represents the added Gaussian white noise power, E I Represents the identity matrix of dimension I×I.

[0075] The value of δ is much smaller than the radar transmission signal power, which ensures that the covariance matrix of the inversion matrix is ​​non-singular.

[0076] Step 4.5, calculate the inversion filter weight vector corresponding to the selected distance unit according to the following formula:

[0077] G=R -1 h T

[0078] Where G represents the inversion filter weight vector, (·) -1 represents the matrix inversion operation, (·) T Represents a transpose operation.

[0079] Step 4.6, taking the position of the pre-pulse pressure echo data corresponding to the selected distance unit as the starting position, intercepting a section of data equal to the length of the inversion filter weight vector in the pre-pulse pressure echo data as the data to be filtered.

[0080] Step 4.7, multiply the weight vector of the inversion filter corresponding to the selected distance unit by the data to be filtered to obtain the filtered data of the selected distance unit of the radar echo data.

[0081] Step 4.8, determine whether the l distance units on both sides of the currently selected strong target have been selected. If so, execute step 4.9, otherwise, execute step 4.2.

[0082] Step 4.9, determine whether all strong targets in the radar pulse compression front echo data have been selected. If so, execute step 5, otherwise, execute step 4.1.

[0083] Step 5: Perform constant false alarm detection on the filtered data.

[0084] Using the same method as the unit average constant false alarm detection method in step 2, perform constant false alarm detection on each distance unit in the filtered data, take the distance units in the data that are greater than or equal to their detection threshold as weak targets, and record the distance unit where each weak target is located.

[0085] The effect of the present invention is further described below in conjunction with simulation experiments:

[0086] 1. Conditions of simulation experiment.

[0087] The hardware platform of the simulation experiment of the present invention is: the processor is Intel i7 5930k CPU, the main frequency is 3.5GHz, and the memory is 16GB.

[0088] The experimental environment for simulation of the present invention is: Windows 7 operating system and MATLAB R2016b.

[0089] The simulation experiment data of the present invention is: the time width T of the LFM signal emitted by the radar p =140×10 -6 s, bandwidth B = 2 × 10 6 Hz, sampling frequency f s =4×10 6 Hz, the radar echo data contains strong target signals, weak target signals and Gaussian white noise signals. The detection signal-to-noise ratio of the strong target is SNR1=53dB, the detection signal-to-noise ratio of the weak target is SNR2=13dB, the strong target is located at the distance unit 20, and the weak target is located at the distance unit 30.

[0090] 2. Simulation content and results analysis:

[0091] The simulation experiment of the present invention is to process the radar echo data in the simulation experiment data of the present invention respectively by using the existing pulse compression processing method without windowing and the method of the present invention under the conditions of the above simulation experiment, and the result is shown in Figure 2. The abscissa of Figure 2 represents the range unit, the ordinate represents the amplitude value, and the curve represents the one-dimensional range image of the target distribution on the range dimension.

[0092] FIG. 2( a ) is a diagram showing the result of processing radar echo data in simulation experiment data by the pulse compression processing method without windowing in the prior art, and FIG. 2( b ) is a diagram showing the result of processing radar echo data in simulation experiment data by the method of the present invention.

[0093] It can be seen from FIG2(a) that after the radar echo data in the simulation experiment data of the present invention is processed by the pulse compression processing method without windowing, the processing result only has a peak at the position of the strong target, but no obvious peak can be seen at the position of the weak target. If the processing result is detected by the unit average constant false alarm detection method, only the strong target can be detected, and the weak target near the strong target cannot be detected.

[0094] It can be seen from FIG2( b) that after the radar echo data in the simulation experiment data of the present invention is processed by the method of the present invention, the peak value at the strong target is suppressed, and an obvious peak value appears at the distance unit where the weak target is located. If the processing result is detected by the unit average constant false alarm detection method, the weak target can be detected.

[0095] The above simulation results show that after the echo data is processed by the existing unwindowed pulse pressure processing method, the unit average constant false alarm detection method is difficult to detect weak targets near strong targets. However, after the inversion filtering-based method adopted by the present invention processes the received echo data, weak targets near strong targets can be effectively detected in the unit average constant false alarm detection stage.

Claims

1. A weak target detection method based on inversion filtering, characterized in that: Construct an inversion data matrix for each strong target, and use the weight vector of the constructed inversion filter to filter the radar echo data; the specific steps of this method are as follows: Step 1: Pulse compression of radar echo data: The radar received echo data is pulse compressed by using the pulse compression coefficient without adding a window to obtain the radar echo data after pulse compression; Step 2: Perform constant false alarm detection on the echo data after radar pulse compression: The unit average constant false alarm detection method is adopted. The amplitude of a distance unit higher than its detection threshold is regarded as a strong target. The distance unit where each strong target is located is recorded, and the recorded distance units are combined into a strong target set. Step 3, construct the inversion data matrix of each strong target: Step 3.1, calculate the value of each element in each column of the segmented time matrix, and form all columns into a segmented time matrix; Step 3.2, calculating the inversion value of each element in the subdivided time matrix, and using the inversion value of the element to update the corresponding element value in the subdivided time matrix; Step 3.3, select a strong target from the detection target set, construct an M×N inversion data matrix, each element of the matrix is ​​initially 0, take the position of the selected strong target in the echo data after pulse compression as the starting point, put each column of the updated subdivision time matrix into the matrix, and obtain the inversion data matrix of the selected strong target, where the value of M is equal to the length of the echo data, and the value of N is equal to the number of columns of the subdivision time matrix; Step 3.4, determine whether all strong targets in the strong target set have been selected, if so, execute step 4 after obtaining the inversion data matrix of all strong targets, otherwise, execute step 3.1; Step 4: Use the constructed inversion filter weight vector to filter the radar echo data: Step 4.1, randomly select a strong target that has not been selected; Step 4.2, randomly select an unselected distance unit from the l distance units on both sides of the distance unit where the strong target is located, where the value of l is determined by the radar system parameters; Step 4.3, taking the position of the selected range unit in the post-pulse compression echo data as the starting point and the length of the discrete sequence emitted by the radar in one pulse time as the interception length, intercept each column in the inversion data matrix corresponding to the selected strong target, and use all elements in the interception length as each column of the inversion matrix; Step 4.4, calculate the covariance matrix of the inversion matrix; Step 4.5, calculating the weight vector of the inversion filter corresponding to the selected distance unit; Step 4.6, taking the position of the pre-pulse pressure echo data corresponding to the selected distance unit as the starting position, intercepting a section of data equal to the length of the inversion filter weight vector in the pre-pulse pressure echo data as the data to be filtered; Step 4.7, multiplying the weight vector of the inversion filter corresponding to the selected distance unit with the data to be filtered to obtain the filtered data of the selected distance unit of the radar echo data; Step 4.8, determine whether the l distance units on both sides of the currently selected strong target have been selected, if so, execute step 4.9, otherwise, execute step 4.2; Step 4.9, determine whether all strong targets in the radar pulse compression front echo data have been selected, if so, execute step 5, otherwise, execute step 4.1; Step 5: Perform constant false alarm detection on the filtered data: The same method as the unit average constant false alarm detection in step 2 is used to detect the filtered data of each distance unit to obtain the detection threshold of the distance unit. The amplitude of the distance unit greater than or equal to its detection threshold is regarded as a weak target, and the distance units where all weak targets are located are recorded.

2. The method for weak target detection based on inversion filtering according to claim 1, characterized in that: The steps of the pulse compression process described in step 1 are as follows: The first step is to construct the unwindowed pulse compression coefficient of the radar transmit signal according to the radar transmit signal as follows: Where h represents the unwindowed pulse compression coefficient of the radar transmission signal, represents the i-th element of s, i = 1, 2, ..., I, I represents the total length of the discrete sequence emitted by the radar in one pulse time, Indicates the rounding operation, T p represents the pulse width of the radar, f s represents the sampling rate of the radar, s=[s1,s2,...,s I ] represents a discrete sequence of radar pulse time transmission, S(t) represents the radar transmission waveform function, Δt=1 / f s represents the sampling time interval of the radar, (·) * represents the conjugation operation; In the second step, the radar received echo data is convolved with the unwindowed pulse compression coefficient to obtain the radar echo data after pulse compression.

3. The weak target detection method based on inversion filtering according to claim 1 is characterized in that: The specific steps of the unit average constant false alarm detection method described in step 2 are as follows: In the first step, an unselected distance unit in the radar pulse compression echo data is randomly selected, and the distance units with a length of U on both sides of the selected distance unit are used as protection units, the reference window length is 2F, and the reference unit is taken as the distance unit with a length of FU outside the protection unit; The second step is to calculate the average value of the noise of all reference cells in the selected distance cell, and use the average value as the noise estimation value of the selected distance cell; The third step is to calculate the threshold factor of the constant false alarm detection of the selected distance unit according to the following formula: Among them, α represents the threshold factor of the constant false alarm detection of the selected distance unit, In represents the logarithmic operation with the natural constant e as the base, and p f It represents the false alarm probability of the selected distance unit under Gaussian white noise; The fourth step is to multiply the threshold factor of the constant false alarm detection of the selected distance unit by the noise estimation value to obtain the detection threshold of the selected distance unit; The fifth step is to determine whether the absolute value of the amplitude under the selected distance unit is greater than or equal to the detection threshold. If so, it is determined that there is a target under the selected distance unit and the sixth step is executed. Otherwise, it is considered that there is no target under the selected distance unit and the first step is executed. Step 6, record the position of the selected distance unit; In the seventh step, the detection results of all distance units are obtained using the same method as in the first to sixth steps.

4. The method for weak target detection based on inversion filtering according to claim 1, characterized in that: The value of each element in each column of the subdivided time matrix described in step 3.1 is obtained by the following formula: Among them, D k Represents the value of each element in the kth column of the subdivision time matrix, k = 1, 2, ..., K, K represents the total number of radar sampling interval subdivisions, and the total number of subdivision time matrix rows is equal to the total length of the discrete sequence emitted by the radar in one pulse time.

5. The method for weak target detection based on inversion filtering according to claim 1, characterized in that: The inverted value of each element in the subdivided time matrix described in step 3.2 is obtained by the following formula: O mn =S(D mn ) Among them, O mn represents the inversion value of the mth row and nth column in the updated subdivision time matrix, D mn Represents the element value in the mth row and nth column of the segmented time matrix.

6. The method for weak target detection based on inversion filtering according to claim 1, characterized in that: The covariance matrix of the inversion matrix described in step 4.4 is obtained by the following formula: R=WW H / K+δE I Where R represents the covariance matrix of the inversion matrix, W represents the inversion matrix, (·) H is the conjugate transpose operation, δ represents the added Gaussian white noise power, E I Represents the identity matrix of dimension I×I.

7. The method for weak target detection based on inversion filtering according to claim 1, characterized in that: The inversion filter weight vector described in step 4.5 is obtained by the following formula: G=R -1 h T Where G represents the inversion filter weight vector, (·) -1 represents the inverse operation, (·) T Represents a transpose operation.

Images