Clutter map method for dual-time-dimension processing

Abstract

The invention discloses a clutter map method for dual-time-dimension processing, belongs to the field of radar data processing, and is used for solving the problems of high measurement false alarm rate and low-speed target self-shielding during low-altitude target detection of a traditional clutter map. The clutter map method mainly comprises four parts of clutter map data setting, clutter map fast time dimension updating, clutter map slow time dimension updating and clutter map filtering, wherein the clutter map data setting comprises the steps of clutter map parameter setting, measurement data input, clutter unit confirmation and clutter map initialization; the clutter map fast time dimension updating provides accurate fast time dimension clutter estimation through the steps of fast time dimension parameter updating and fast time dimension parameter resetting; the slow time dimension updating of the clutter map refers to the iteration updating of a clutter map storage value according to the fast time dimension clutter estimation; the clutter map filtering means that a result of multiplying a clutter map storage value and a threshold factor is used as a clutter map detection threshold, and the measurement type is judged according to the detection threshold.

Description

Technical Field

[0001] The present invention belongs to the field of radar data processing, and particularly relates to a clutter map method with dual time dimensions for processing. Background Art

[0002] Clutter map is a type of time-series constant false alarm rate (CFAR) processing. It estimates the clutter intensity by utilizing the stability of clutter in the time domain, and obtains the clutter estimate value by exponentially weighting the multi-frame scan echo powers of the detection unit. The classic Nitzberg clutter map processing mainly includes two steps: clutter map update and detection decision. It iteratively updates the stored value of the clutter map using the latest scan value, and its iterative model is a first-order autoregressive model. The specific iterative formula is,

[0003] p n = ωq n +(1 - ω)p n-1

[0004] In the formula, q n represents the power value output by the detection unit of the nth frame scan, p n represents the stored value of the clutter map during the detection decision of the (n + 1)th frame scan, and ω represents the forgetting factor.

[0005] The criterion for determining the measurement data of the (n + 1)th frame scan as a target is,

[0006] q n+1 ≥ Tp n

[0007] In the formula, T represents the clutter map threshold factor.

[0008] The main problems of traditional clutter map methods in low-altitude surveillance radars are high measurement false alarm rates and slow target self-concealment problems. The main clutter sources of low-altitude surveillance radars are ground clutter and meteorological clutter. Such clutter will increase the measurement false alarm rate, thereby affecting the target observation of the terminal. Because the measurement data received by the radar data processing system is sparse in space, traditional clutter maps usually choose to expand the clutter unit range in the spatial dimension to reduce the system measurement false alarm rate. However, this method is prone to self-concealment problems for slow targets. The self-concealment problem refers to the situation where the detection threshold of a clutter unit is increased because the target stays in the same clutter unit multiple times, and finally the target is treated as clutter. Therefore, the clutter map method of the present invention needs to meet two requirements: reducing the measurement false alarm rate and improving the slow target self-concealment problem. Summary of the Invention

[0009] The object of the present invention is to provide a clutter map method with dual time dimension processing. The present invention draws on the slow time dimension idea in clutter map delay detection and the multi-frame detection idea of the multi-frame sliding window dual-threshold clutter map algorithm in the field of radar signal processing, and uses the clutter map method with dual time dimension processing in the field of radar data processing, ultimately reducing the measurement false alarm rate while improving the target self-shadowing problem.

[0010] The solution to achieve the object of the present invention is: a clutter map method with dual time dimension processing, which utilizes the stability of clutter in the time domain, uses the mean power of multiple frames of clutter cells as clutter estimation, and improves the stability of clutter estimation; at the same time, the method weakens the influence of the target on the clutter map detection threshold through mean processing in the fast time dimension, and improves the slow target self-shadowing problem. The method includes:

[0011] Step 1: Clutter map parameter setting, setting fixed parameters: forgetting factor ω, threshold factor T, initial scan frame number M0, clutter cell size N, fast time dimension update condition M, mean selection threshold NUM, and setting variable parameters: fast time dimension scan frame number m, fast time dimension target number num[i][j][k], fast time dimension target energy accumulation value energy[i][j][k], fast time dimension clutter estimation ave[i][j][k], and clutter map storage value estimate[i][j][k], where i represents the distance number subscript, j represents the wave position number subscript, and k represents the pulse type subscript;

[0012] Step 2: Input measurement data, parse the measurement data information received by radar data processing according to the communication protocol, use the measurement data as the input of the clutter map method, and execute Step 3;

[0013] Step 3: Clutter cell confirmation, confirm the clutter cell number of the measurement data in Step 2, divide the radar detection range into a three-dimensional clutter map according to wave position, distance, and pulse type, use N distance cells, 1 wave position cell, and 1 pulse type as 1 clutter map cell, and use the clutter cell confirmed in this step as the processing unit in the following. The clutter cell contains the measurement data in Step 2, and execute Step 4;

[0014] Step 4: Clutter map initialization, after the clutter map starts working, use the power mean of continuous M0 frames of measurement in the clutter cell in Step 3 as the initial clutter map storage value. If the initialization is completed, skip this step and execute Step 5, otherwise execute Step 2 after this step is completed;

[0015] Step 5: Traverse the measurement data, traverse the measurement data in the clutter cell in Step 3. When the measurement data is not completely traversed, execute Step 6, otherwise execute Step 8;

[0016] Step 6: Clutter map filtering. The product of estimate[i][j][k] and f is used as the clutter map detection threshold. Compare each measurement power with the detection threshold to confirm whether the measurement is clutter. If it does not exceed the detection threshold, it is identified as clutter and its clutter flag is set to 1. If the measurement exceeds the detection threshold, it is determined as a target, and go to Step 7;

[0017] Step 7: Fast time dimension parameter update. Update num[i][j][k] and energy[i][j][k] according to the measurement data, and increment m by 1 after processing each frame of data. Then go to Step 5;

[0018] Step 8: Clutter map update determination. Whenever a frame of data is processed, check whether the clutter map update condition is met, that is, whether m is equal to M. When the update condition is satisfied, go to Step 9; otherwise, go to Step 2;

[0019] Step 9: Fast time dimension clutter map update. When the number of measurements num[i][j][k] of a clutter cell within M frames is less than NUM, the fast time dimension clutter estimate is

[0020] ave[i][j][k] = energy[i][j][k] / M

[0021] On the contrary, when num[i][j][k] is greater than or equal to NUM, the fast time dimension clutter estimate is

[0022] ave[i][j][k] = energy[i][j][k] / num[i][j][k]

[0023] Then go to Step 10;

[0024] Step 10: Clutter map stored value iteration. Use the first-order autoregressive model to iterate the clutter map stored value. Select the forgetting factor ω′ corresponding to the clutter cell according to ave[i][j][k] / estimate[i][j][k], and use ω′ and ave[i][j][k] to update estimate[i][j][k]. The update method is

[0025] estimate[i][j][k] = ω′×ave[i][j][k] + (1 - ω′)×estimate[i][j][k]

[0026] Then go to Step 11;

[0027] Step 11: Fast time dimension parameter reset. Clear m, num, ave, and energy, and go to Step 2.

[0028] The advantages of the dual-time-dimensional clutter map processing method provided by the present invention compared with the traditional clutter map are as follows: 1) The clutter intensity can be effectively estimated through the mean processing in the fast time dimension, enhancing the stability of the detection threshold; 2) The clutter map update rate is reduced by cross-frame updating in the slow time dimension, and its update rate is 1 / M of the traditional method, thus improving the self-shadowing problem of slow targets; 3) Two fast-time-dimensional clutter estimation methods can prevent the measurement with unstable time dimension from affecting the clutter map detection threshold.

[0029] The disadvantage of the dual-time-dimensional clutter map processing method provided by the present invention compared with the traditional clutter map is that the dual-time-dimensional clutter map has a slow update rate and poor ability to adapt to environmental changes. Description of the Drawings

[0030] Figure 1 It is a processing framework diagram of the "clutter map method with dual-time-dimensional processing" of the present invention.

[0031] Figure 2 It is a detection principle diagram of the "clutter map method with dual-time-dimensional processing" of the present invention.

[0032] Figure 3 It is an implementation flowchart of the "clutter map method with dual-time-dimensional processing" of the present invention.

[0033] Figure 4 It is a graph showing the change of the detection threshold curve in the embodiment of the present invention and other clutter map methods.

[0034] Figure 5 is Figure 4 a partial enlarged view of Detailed Embodiment

[0035] To facilitate the understanding of the present invention, the following will combine the attached Figures 1 to 3 to describe the present invention more comprehensively. A clutter map method with dual-time-dimensional processing of the present invention has a processing framework as Figure 1 shown, a detection process as Figure 2 shown, and an implementation process as Figure 3 shown. The method includes:

[0036] Step 1: Clutter map parameter setting. Set fixed parameters: forgetting factor ω = 0.125, threshold factor T = 2, initial scan frame number M0 = 25, clutter cell size N = 15, fast-time dimension update condition M = 10, mean selection threshold NUM = 2. Set variable parameters: fast-time dimension scan frame number m, fast-time dimension target number num[i][j][k], fast-time dimension target energy accumulation value energy[i][j][k], fast-time dimension clutter estimate ave[i][j][k], and clutter map storage value estimate[i][j][k], where i represents the range number subscript and 0 ≤ i ≤ 137, j represents the wave position number subscript and 0 ≤ j ≤ 107, k represents the pulse type subscript and 0 ≤ k ≤ 1;

[0037] Step 2: Input measurement data. Analyze the measurement data information received by radar data processing according to the communication protocol. The measurement data is used as the input of the clutter map method, and go to Step 3;

[0038] Step 3: Clutter cell confirmation. Confirm the clutter cell number of the measurement data in Step 2. Divide the radar detection range into a three-dimensional clutter map according to wave position, range, and pulse type. Take 15 range cells, 1 wave position cell, and 1 pulse type as 1 clutter map cell. Subsequently, use the clutter cell confirmed in this step as the processing unit. The clutter cell contains the measurement data in Step 2, and go to Step 4;

[0039] Step 4: Clutter map initialization. After the clutter map starts working, use the power mean value of 25 consecutive frames of measurements in the clutter cell in Step 3 as the initial clutter map storage value. If the initialization is completed, skip this step and go to Step 5; otherwise, go to Step 2 after completing this step;

[0040] Step 5: Traverse the measurement data. Traverse the measurement data in the clutter cell in Step 3. When the measurement data has not been completely traversed, go to Step 6; otherwise, go to Step 8;

[0041] Step 6: Clutter map filtering. Use 2 times of estimate[i][j][k] as the clutter map detection threshold. Compare each measurement power with the detection threshold to confirm whether the measurement is clutter. Those not exceeding the detection threshold are identified as clutter and set their clutter flag to 1. The measurements exceeding the detection threshold are determined as targets, and go to Step 7;

[0042] Step 7: Fast-time dimension parameter update. Update num[i][j][k] and energy[i][j][k] according to the measurement data, and increment m by 1 after processing each frame of data, then go to Step 5;

[0043] Step 8: Clutter map update determination. Whenever the data processing for the wave position 107 is completed, check whether the clutter map update condition is met, that is, whether m is equal to 10. When the update condition is satisfied, execute Step 9; otherwise, execute Step 2.

[0044] Step 9: Fast-time dimension clutter estimation. When the number of measurements num[i][j][k] of the clutter cell within 10 frames is less than 2, the fast-time dimension clutter estimation is

[0045] ave[i][j][k] = energy[i][j][k] / 10

[0046] Conversely, when num[i][j][k] is greater than or equal to 2, the fast-time dimension clutter estimation is

[0047] ave[i][j][k] = energy[i][j][k] / num[i][j][k]

[0048] Then execute Step 10.

[0049] Step 10: Iteration of the clutter map stored value. Use the first-order autoregressive model to iterate the clutter map stored value. When 2 ≤ ave[i][j][k] / estimate[i][j][k] ≤ 4, let ω′ = 2ω = 0.25; otherwise, let ω′ = ω = 0.125. Then use ω′ and ave[i][j][k] to update estimate[i][j][k], and the update method is

[0050] estimate[i][j][k] = ω′ × ave[i][j][k] + (1 - ω′) × estimate[i][j][k]

[0051] Then execute Step 11.

[0052] Step 11: Reset of fast-time dimension parameters. Clear m, num, ave, and energy, and execute Step 2.

[0053] The present invention will be further described below in conjunction with embodiments.

[0054] There is currently a situation where a certain UAV target is continuously detected 4 times within a clutter cell. The distance number of the clutter cell where the UAV target is located is 12, the wave position number is 94, and the pulse type is 0. Within the above clutter cell, the UAV target appears in the scans from frame 373 to 376 and the target power change is less than 3 dB.

[0055] Here, the detection thresholds of the above clutter cells using the traditional clutter map, the clutter map with threshold delay processing, and the clutter map with dual-time dimension processing are compared respectively. Among them, the threshold delay processing is based on the traditional clutter map processing but updates the clutter map with a delay of 10 scan frames. The curves of the detection thresholds of the clutter cells under the three processing methods changing with the number of scan frames are as Figure 4 and Figure 5 shown.

[0056] In the embodiment, before the target appears, the detection threshold of the clutter map with dual-time dimension processing is more stable than the other two processing methods, that is, the dual-time dimension processing estimates the clutter more accurately; from Figure 5 it can be seen that the maximum detection threshold of the clutter map with dual-time dimension processing is less than half of that of the other two processing methods. The influence of the UAV target on the detection threshold of the clutter map under the dual-time dimension processing is small, that is, the problem of target self-shadowing is improved to some extent, and the highest threshold value of the dual-time dimension processing appears at the 382nd frame, having a certain threshold delay effect. In summary, the clutter map with dual-time dimension processing has the advantages of stable detection threshold and improved target self-shadowing.

Claims

1. A dual-time-dimensional clutter map processing method, comprising: Step 1: Clutter map parameter setting, setting fixed parameters: forgetting factor ω, threshold factor T, initial scan frame number M0, clutter cell size N, fast-time-dimensional update condition M, mean selection threshold NUM, and setting variable parameters: fast-time-dimensional scan frame number m, fast-time-dimensional target number num[i][j][k], fast-time-dimensional target energy accumulation value energy[i][j][k], fast-time-dimensional clutter estimation ave[i][j][k], and clutter map storage value estimate[i][j][k], where i represents the range number subscript, j represents the wave position number subscript, and k represents the pulse type subscript; Step 2: Input measurement data, parsing the measurement data information received by the radar data processing according to the communication protocol, using the measurement data as the input of the clutter map method, and performing Step 3; Step 3: Clutter cell confirmation, confirming the clutter cell number of the measurement data in Step 2, dividing the radar detection range into a three-dimensional clutter map according to wave position, range, and pulse type, taking N range cells, 1 wave position cell, and 1 pulse type as 1 clutter map cell, and subsequent processing units are all the clutter cells confirmed in this step. The clutter cell contains the measurement data in Step 2, and Step 4 is performed; Step 4: Clutter map initialization, after the clutter map starts working, taking the power mean of the continuous M0 frames of measurements in the clutter cell in Step 3 as the initial clutter map storage value. If the initialization is completed, this step is skipped and Step 5 is performed. Otherwise, after this step is completed, Step 2 is performed; Step 5: Traverse the measurement data, traverse the measurement data in the clutter cell in Step 3. When the measurement data has not been completely traversed, Step 6 is performed. Otherwise, Step 8 is performed; Step 6: Clutter map filtering, taking the product of estimate[i][j][k] and T as the clutter map detection threshold, comparing each measurement power with the detection threshold to confirm whether the measurement is clutter. The measurement that does not exceed the detection threshold is identified as clutter and its clutter flag is set to 1. The measurement that exceeds the detection threshold is determined as a target, and Step 7 is performed; Step 7: Fast-time-dimensional parameter update, updating num[i][j][k] and energy[i][j][k] according to the measurement data, and adding 1 to m after each frame of data is processed, and performing Step 5; Step 8: Clutter map update determination, whenever a frame of data is processed, check whether the clutter map update condition is reached, that is, whether m is equal to M. When the update condition is met, Step 9 is performed. Otherwise, Step 2 is performed; Step 9: Fast-time-dimensional clutter estimation, obtaining the fast-time-dimensional clutter estimation according to num[i][j][k] and energy[i][j][k] updated in Step 7, and performing Step 10; Step 10: Clutter map storage value iteration, using the first-order autoregressive model to iterate the clutter map storage value, and iterating according to the fast-time-dimensional clutter estimation and forgetting factor in Step 9, and performing Step 11; Step 11: Fast-time-dimensional parameter reset, clearing m, num, ave, and energy, and performing Step 2.

2. The dual-time-dimensional clutter map processing method according to claim 1, characterized in that In step 3, the measurement data is partitioned according to wave position, distance, and pulse type. The wave position is distinguished based on the center pointing angle of the scanning beam, and the airspace pointed to by the same wave position in each frame of scanning is consistent. The pulse type includes two types: wide pulse and narrow pulse. If the radar operates in a search-plus-track mode, the measurement data transmitted back by the tracking beam does not participate in the clutter map processing.

3. The dual-time-dimensional clutter map processing method according to claim 1, wherein In step 8, when the update condition is met, first obtain the fast-time dimension clutter estimate through step 9, and then perform the iteration of the clutter map storage value through step 10. When the update condition is not met, execute step 2. This processing method can play a role in delaying the change of the detection threshold.

4. The dual-time-dimensional clutter map processing method according to claim 1, wherein The fast-time dimension clutter estimate ave in step 9 has two calculation methods. When the number of measurements num[i][j][k] in the clutter cell within M frames is less than NUM, the fast-time dimension clutter estimate is ave[i][j][k] = energy[i][j][k] / M Conversely, when num[i][j][k] is greater than or equal to NUM, the fast-time dimension clutter estimate is ave[i][j][k] = energy[i][j][k] / num[i][j][k] 5. The dual-time-dimensional clutter map processing method according to claim 1, wherein, In step 10, when iterating the clutter map storage value, select the forgetting factor ω′ of the corresponding clutter cell according to ave[i][j][k] / estimate[i][j][k]. When 2 ≤ ave[i][j][k] / estimate[i][j][k] ≤ 4, let ω′ = 2ω, otherwise let ω′ = ω. Then use ω′ and ave[i][j][k] to update estimate[i][j][k]. The update method is estimate[i][j][k] = ω′ × ave[i][j][k] + (1 - ω′) × estimate[i][j][k]

Images