Ground clutter suppression method for ground reconnaissance radar

Abstract

The invention relates to the technical field of radar information, in particular to a ground clutter suppression method for a ground reconnaissance radar, which comprises the following steps of: firstly, establishing a clutter profile map by using echo data, judging information inside and outside a clutter according to the clutter profile map, and selecting different signal processing flows, namely, adopting a detection channel 1 outside the clutter, adopting a detection channel 2 and a detection channel 3 inside the clutter and adopting a detection channel 2 and a detection channel 3 outside the clutter; after the detection channel 1 passes through the MTD filter, clutter suppression processing is not carried out, and a detection result is output; the detection channel 2 is a coherent processing channel, firstly performs mean filtering processing, then performs coherent processing, performs clutter suppression processing through phasor mean filtering, and outputs a detection result; and the detection channel 3 is a zero Doppler processing channel and is used for detecting a low-speed target in a clutter area, compared with the prior art, the method can reduce Doppler sidelobes as much as possible on the premise of ensuring that the ground clutter height is suppressed, and has the minimum main lobe width and the minimum signal-to-noise ratio loss.

Description

Technical Field

[0001] This invention relates to the field of radar information technology, specifically a ground clutter suppression method for ground reconnaissance radar applicable to scenarios such as border security surveillance, protection of important facilities, and countermeasures against unmanned aerial vehicles, and effective detection of ground moving targets and low-altitude targets such as personnel and vehicles in specific areas. Background Technology

[0002] Modern high-tech warfare is a deep, three-dimensional war integrating air and ground capabilities. Intelligence reconnaissance is crucial for combat forces to achieve victory at all levels of combat. Many countries are striving to develop a series of reconnaissance radars, ranging from spaceborne and airborne to vehicle-mounted, fixed ground-based, and portable, forming a comprehensive battlefield reconnaissance and intelligence system through communication systems and GPS technology. Ground-based reconnaissance radar, as one of these systems, offers advantages over other detection sensors for acquiring battlefield intelligence, including portability, speed, real-time reconnaissance, large-area surveillance, long detection range, precise positioning, and all-weather operation. Ground-based reconnaissance radar is a general term for various reconnaissance radars used to detect various military targets in a specific scenario to obtain intelligence. It can detect low-speed moving targets and low-flying targets in complex ground clutter environments and has become an indispensable and important reconnaissance equipment on the modern battlefield.

[0003] Market demands for border security surveillance, critical facility protection, and drone countermeasures have significantly expanded the application areas of ground-based radar, leading to its increasing importance. In real-world environments, slow-moving targets (such as personnel, vehicles, and drones) pose a high threat, making their detection capability a crucial indicator of radar system performance. However, due to the influence of ground clutter, most radars face insufficient detection capabilities for targets with slow radial speeds. The main reasons include: First, radar echoes contain not only moving target information but also strong ground clutter, weather clutter, and slow-moving interference clutter. Compared to fast-moving targets, slow-moving targets exhibit more severe overlap with ground clutter spectra in the Doppler domain, making effective extraction from ground clutter difficult. Second, pedestrians, a typical example of slow-moving ground targets, are often small and maneuverable, with low reflection coefficients but large RCS fluctuations, significantly impacting effective detection. While offset phase center antenna (DPCA) and space-time adaptive processing (STAP) techniques are commonly used for slow-moving target detection on moving platforms, these are difficult to directly apply to ground-based radar. Currently, common methods primarily involve suppressing stationary clutter using MTI and MTD filters, followed by high-sensitivity constant false alarm rate (CFAR) detection methods such as ML-CFAR, OS-CFAR, and OSGO-CFAR. However, due to system performance limitations, the order of the MTI filter cannot be designed too high, and its wide transition band inevitably leads to a loss of target energy during broadband clutter suppression. Traditional signal processing frequency domain filtering suppresses both clutter and target signals, making it difficult to achieve a good signal-to-clutter ratio.

[0004] Based on the currently retrieved public literature, no information related to the use of phasor mean filtering for ground reconnaissance radar was found, nor was any technology or literature on suppressing ground clutter by combining phasor mean filtering with multidimensional spatial clutter maps found. Summary of the Invention

[0005] This invention addresses the shortcomings and deficiencies of existing technologies by proposing a method for ground clutter suppression that combines static clutter suppression with spatial multidimensional clutter maps for detecting slow-moving ground targets. The aim is to improve the radar's ability to detect ground targets and provide a ground clutter suppression method for ground reconnaissance radar.

[0006] This invention is achieved through the following technical solution: A ground clutter suppression method for ground reconnaissance radar is characterized by first establishing a clutter profile map using echo data, determining clutter-inside and clutter-outside information based on the clutter profile map, and selecting different signal processing flows. Specifically, detection channel 1 is used for clutter-outside clutter, while detection channels 2 and 3 are used for clutter-inside clutter. Detection channel 1, after passing through an MTD filter, does not undergo clutter suppression processing and outputs the detection result. Detection channel 2 is a coherent processing channel, which first performs mean filtering, then coherent processing, and finally clutter suppression processing via phasor mean filtering before outputting the detection result. Detection channel 3 is a zero-Doppler processing channel used to detect low-speed targets within the clutter region. The raw data is first processed by phasor mean filtering to obtain residual clutter, and a residual multidimensional clutter map is established using the residual clutter. Then, the multidimensional clutter map CFAR is used to detect targets.

[0007] This invention receives the intermediate frequency echo signal from the reconnaissance radar and sequentially performs A / D sampling processing, digital down-conversion processing, and pulse compression processing. Then, it performs phasor mean filtering processing, coherent accumulation processing, and sum-difference normalization processing, and finally completes the detection through CA-CFAR or multidimensional clutter map CFAR.

[0008] The specific steps for establishing the clutter profile map as described in this invention are as follows: First, estimate the clutter power of the clutter cells and perform a first-level comparison: compare the estimated clutter power value with the noise threshold. If the estimated clutter power value is greater than the noise threshold, increment the counter by 1. Perform a second-level comparison: the threshold is based on the K / L criterion, that is, the counter stores the number of times the first threshold has been crossed in the most recent L scans. If the counter is greater than K, it indicates that the second threshold has been crossed, which means that the clutter is inside the clutter; otherwise, it means that the clutter is outside the clutter.

[0009] The clutter profile map creation in this invention includes creating a multi-channel spatial static clutter map and a dynamic clutter map. The multi-channel clutter map adds Doppler information to the planar clutter map, dividing the clutter within a CPI into Doppler dimensions to obtain clutter maps with multiple Doppler channels. Detection is performed on multiple channels. The combination of planar and three-dimensional clutter maps achieves a more refined division of the clutter region, and the three-dimensional clutter map reduces storage and computational requirements, which is beneficial to improving the target detection performance.

[0010] In this invention, phasor mean filtering utilizes the constant distance from the stationary target to the radar antenna and the constant time delay on the radar received pulses. This constant distance cancels out the mean obtained by summing the phasors of the stationary target, achieving the purpose of attenuating the static clutter amplitude. Specifically, it includes: first, averaging all received pulses to obtain a reference received pulse; then, subtracting the reference received pulse from each received pulse to obtain the target echo signal. The expression for the reference received pulse is: (1), in, For fast time dimension or distance dimension sampling points, For time sampling points in slow or velocity dimensions, the formula for the phasor mean cancellation algorithm is: (2), Before coherent accumulation, mean filtering is first performed to store the accumulated range dimension data in SRAM. After all echoes accumulated by the current beam are stored, the velocity dimension data is read out frame by frame according to the velocity dimension order. Secondly, the average value of each frame of velocity dimension data is subtracted to remove zero-frequency clutter components. Then, it is weighted with a Taylor window to suppress spectral leakage. Finally, FFT operation is performed, where mean filtering first rearranges the data storage. The specific process is as follows: implemented on the peripheral SRAM, the SRAM is divided into two regions according to the address for ping-pong operation. The SRAM region allocation horizontally represents the range dimension and vertically represents the velocity dimension. Each frame of range dimension echo data is written into the SRAM row by row. The number of points written in each frame is 8192, and a total of 128 frames of echo data are written. Due to the pipelined processing inside the FPGA, the data of the previous frame is read out while the current wave position data is written. The data is read out column by column. Each frame of data has 128 points, and a total of 8192 frames of data are read out. Different frames represent different range gates. In the mean filtering process, the average of the data is calculated using an accumulation method. First, it is determined whether the input data is the first frame of pulse accumulation. If it is, it is directly cached in RAM. If not, when the data is valid, the data cached in the previous frame is read from RAM and added to the data of the current frame, and then written back to RAM. Here, RAM is the called IP core. The IP core is configured as a simple dual-port RAM, and read and write operations are performed simultaneously. The input data and the stored data are added cyclically until the last frame of data is reached, completing the accumulation of 128 frames of data. The result of the accumulation is divided by 128 to obtain the average value of the velocity dimension data on each distance gate. Direct division in the FPGA or calling the division IP core would consume DSP and LUT resources. Since 128 is 2 to the power of 7, the division operation is implemented by shifting the data 7 bits to the right.

[0011] The steps for establishing the clutter map in this invention include: First, the radar's effective range is divided into many clutter cells according to different requirements. The clutter cells are larger than the target resolution cells. Indicates the distance dimension of the clutter cell. Indicates the azimuth dimension, where △R is the radar's range resolution unit. Let PRT be the antenna scanning angle corresponding to one pulse repetition period. and That is, a clutter unit spans M range units and receives N pulses; To estimate the clutter intensity of a clutter cell, the antenna scans one revolution, obtaining M×N echo data points for each clutter cell, where N is the number of pulses and M is the number of range cells. To estimate the clutter intensity of that cell, the N pulses are first accumulated, which can be expressed as: (3), Where r represents the number of antenna scans, k represents the clutter cell number, m=1,2,...,M represents the range cell number, and n=1,2,...,N represents the pulse number. After pulse accumulation, echo data from M range cells are obtained. Using these data, the clutter intensity is estimated, expressed as: (4); Clutter map updating and detection: When the antenna scans the location of a clutter cell, it estimates the clutter intensity of that clutter cell. The data is written into the clutter image storage unit. Once the antenna scans once, a complete clutter image is formed. After accumulating data for several to dozens of antenna scan cycles, a stable clutter image is established.

[0012] In this invention, clutter map detection includes: Xr represents the clutter intensity estimate of a single clutter cell in a single frame. Using Xr as input to the update algorithm, the cumulative formula for clutter map update can be expressed as: (5), Where r represents the current radar sweep number, k represents the clutter cell number, and w is the forgetting factor. This represents the clutter intensity estimate after the clutter map is updated. Expanding the above equation, we get: (6), False alarm probability P fa The forgetting factor w is 1 / 8, and the relative threshold T of clutter and the false alarm probability can be expressed as: (7), When designing a detector, a false alarm probability P is given. fa Given the forgetting factor w, the relative threshold T can be calculated using the above formula. Compare the echo intensity with the absolute threshold. If the echo intensity is greater than the absolute threshold, then a target exists; otherwise, no target exists.

[0013] In this invention, Doppler information is incorporated into the multi-channel clutter map. Clutter within a single CPI is divided into Doppler channels, resulting in clutter maps with multiple Doppler channels. Detection is performed on these multiple channels to improve target detection performance. Since the remaining clutter map covers a maximum of three Doppler channels, only three channel clutter maps are established, while the remaining channels are processed normally. The workflow for multi-channel clutter map detection is as follows: Based on the radar scanning angle range α and the 3dB beamwidth β, using the formula... (8) Obtain the number of radar azimuths N, where [·] is the floor function, and then use formula (9): (9) Angle of each azimuth resolution unit ; Based on the 3dB beamwidth, the number of accumulation points within one PRT is calculated for moving target detection. Assuming k is the Doppler dimension number, the m-th ring, the k-th Doppler channel, and the azimuth angle are recorded. The data is It is the forgetting factor, according to formula (10): (10) Update the clutter plot to a stable state; Clutter diagram established based on the m-th scan and the scan data of the (m+1)th cycle Clutter detection is performed on the data from the (m+1)th cycle to obtain the detection results. and The size relationship between them determines whether a target exists. (11) Detection has a target, where h is the threshold factor.

[0014] Compared with existing technologies, this invention fully utilizes clutter characteristics and employs phasor-means filtering combined with multidimensional clutter map processing to improve radar's ability to detect ground targets. The phasor-means filter provides a certain level of clutter suppression while ensuring the passage of low-speed target signals, effectively improving the signal-to-clutter ratio (SNR) for low-speed targets. The residual clutter map is a clutter map established using the residual clutter signal after the original signal has been filtered by the phasor-means filter. Addressing the shortcomings of MTI or MTD (Moving Target Interception) technologies, this invention combines phasor-means filtering with spatial multidimensional clutter map technology to detect clutter conditions in real time, determine clutter intensity and characteristics, correct stored clutter, and minimize Doppler sidelobes while ensuring high suppression of ground clutter, achieving the minimum main lobe width and minimum SNR loss, thus realizing clutter suppression technology for low-speed target detection.

[0015] Compared with existing technologies, this invention has the following advantages: it improves the radar's ability to resist clutter and detect low-speed targets; it distinguishes between clutter-inside and clutter-outside targets using clutter profile maps, avoiding performance loss caused by suppressing clutter outside the clutter area; it employs a phasor mean cancellation algorithm, which greatly improves the signal-to-noise ratio of moving or slightly moving targets while suppressing the phase of stationary targets; it maintains a high signal-to-noise ratio by utilizing the advantages of the phasor mean cancellation algorithm, such as no reduction in target amplitude, relatively clean background noise, and relatively complete preservation of micro-Doppler information; the combination of planar and three-dimensional clutter maps enables more refined division of clutter regions, and the planar clutter map reduces storage and computational requirements; it uses dual-branch signal processing, with one branch processing data from clutter regions and the other processing data from non-clutter regions, thus avoiding detection performance loss caused by clutter suppression processing; it can achieve very ideal results for detecting ground personnel targets; it has strong engineering feasibility and has promotional application value. Attached Figure Description

[0016] Appendix Figure 1 This is a flowchart of the radar echo signal processing in this invention.

[0017] Appendix Figure 2 This is a block diagram illustrating the principle of the ground clutter suppression method of the present invention.

[0018] Appendix Figure 3 This is a block diagram for establishing the clutter profile diagram of the present invention.

[0019] Appendix Figure 4 This is a block diagram of the moving target detection module of the present invention.

[0020] Appendix Figure 5This is a schematic diagram of the phasor mean cancellation algorithm in this invention.

[0021] Appendix Figure 6 This is the logic diagram for averaging in this invention.

[0022] Appendix Figure 7 This is a schematic diagram of clutter diagram division in this invention, wherein... Figure 7 (a) is a radar power diagram, and (b) is an enlarged view of clutter elements.

[0023] Appendix Figure 8 This is a logic diagram for clutter map updating and detection in this invention. Appendix Figure 9 This is a block diagram of the spatial clutter map detection logic of the present invention. Detailed Implementation

[0024] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0025] This example provides a method for suppressing ground clutter in ground reconnaissance radar, specifically including: Step 1: Create a clutter profile: such as Figure 3 The clutter profile is constructed through two comparisons. First, the clutter power of the clutter cells is estimated. For uniform clutter, the mean method works best. If the influence of clutter edges is considered, a larger mean value is even better. The first comparison is between the estimated clutter power and the noise threshold. If the estimated clutter power is greater than the noise threshold, the counter is incremented by 1. The second threshold is compared between the counter value and threshold 2. For example, the K / L criterion can be used. The counter stores the number of times the first threshold has been crossed in the most recent L scans. If the counter is greater than K, it means that the second threshold has been crossed, indicating that the clutter is inside the clutter; otherwise, it means that the clutter is outside the clutter. Step 2: Employing the phasor mean cancellation algorithm, leveraging the constant distance from the stationary target to the radar antenna and the constant time delay on the radar received pulses, the mean value obtained by summing the phasors of the stationary target and the received pulse is subtracted to achieve amplitude cancellation. First, the reference received pulse is obtained by averaging all received pulses. Then, the target echo signal is obtained by subtracting the reference received pulse from each received pulse. The expression for the reference received pulse is: (1), where, For fast time dimension (distance dimension) sampling points, For time sampling points in the slow time dimension (velocity dimension), the formula for the phasor mean cancellation algorithm is: (2), like Figure 2As shown, before the coherent accumulation operation, mean filtering is first performed, and the accumulated multi-frame range dimension data is stored in SRAM. After all the echoes accumulated by the current beam are stored, the velocity dimension data is read out frame by frame according to the velocity dimension order. Secondly, the average value of each frame's velocity dimension data is subtracted to remove zero-frequency clutter components. Then, it is weighted with a Taylor window to suppress spectral leakage. Finally, FFT operation is performed. The implementation block diagram of the moving target detection module is shown below. Figure 4 As shown; Figure 5 The data storage rearrangement process is implemented on the peripheral SRAM. The SRAM is divided into two regions based on address for ping-pong operations. The SRAM region allocation and read / write order are as follows: Figure 4 As shown, the horizontal axis represents the distance dimension, and the vertical axis represents the velocity dimension. Gray squares indicate that the memory cell is full, and green squares indicate that the memory cell is empty or being written. Each frame of distance dimension echo data is written to the SRAM row by row. Each frame of data has 8192 points written, and a total of 128 frames of echo data are written. Due to the pipelined processing inside the FPGA, when writing the current wave position data, the data of the previous wave position is also read at the same time. The data is read by column. Each frame of data has 128 points, and a total of 8192 frames of data are read. Different frames represent different distance gates. Figure 6 To calculate the average value through accumulation, the system first determines whether the input data is the first frame of pulse accumulation. If it is, it is directly cached in RAM. If not, when the data is valid, the data from the previous frame cached in RAM is read, added to the current frame data, and then written back to RAM. Here, RAM is the called IP core, which is configured as a simple dual-port RAM. This type of memory has two independent read and write ports, allowing simultaneous read and write operations. The input data and the stored data are added cyclically until the last frame data is reached, thus completing the accumulation of 128 frames of data. The accumulated result is divided by 128 to obtain the average value of the velocity dimension data at each distance gate. Direct division in the FPGA or calling the division IP core would consume DSP and LUT resources. Since 128 is 2 to the power of 7, the division operation is implemented by shifting the data 7 bits to the right, which can save some resources. Step 3: Clutter Map Detection. This example uses a planar amplitude clutter map to illustrate the steps for creating and detecting clutter maps: First, the radar's effective range is divided into many clutter cells according to different requirements. To reduce storage capacity, in most cases, the clutter cells are larger than the target resolution cells, such as... Figure 7 (a) A schematic diagram of the azimuth-range partitioning of two-dimensional clutter cells is given. Figure 7 (a) Indicates the distance dimension of the clutter cell. Indicates orientation dimensions. Figure 7(b) is an enlarged view of the clutter elements, where ΔR represents the range resolution element of the radar. The antenna scanning angle corresponding to one pulse repetition period (PRT).

[0026] set up and That is, a clutter unit spans M range units and receives N pulses; To estimate the clutter intensity of a clutter cell, the antenna scans one revolution, obtaining M×N echo data points for each clutter cell, where N is the number of pulses and M is the number of range cells. To estimate the clutter intensity of this cell, the N pulses are first accumulated, which can be expressed as: (3), Where r represents the number of antenna scans, k represents the clutter cell number, m=1,2,...,M represents the range cell number, and n=1,2,...,N represents the pulse number. After pulse accumulation, echo data from M range cells are obtained. These data are used to estimate clutter intensity. The most commonly used method is cell averaging, which is most effective for spatially uniform clutter and can be expressed as: (4), Clutter map updating and detection: When the antenna scans the location of a clutter cell, the clutter intensity of that clutter cell is estimated using the method described above. The data is then written into the clutter image storage unit. After one antenna scan, a complete clutter image is formed. The clutter image formed at this time is not stable or reliable. It generally takes several to dozens of antenna scan cycles to accumulate data to establish a relatively stable clutter image.

[0027] like Figure 8 The general process of clutter map updating and detection is presented. To simplify the equipment and processing, this estimation (clutter map establishment and updating) typically employs a single-loop feedback accumulation method. Here, Xr represents the clutter intensity estimate of a single clutter cell in a single frame. Using Xr as input to the update algorithm, the accumulation formula for clutter map updating can be expressed as: (5), where r represents the current radar sweep number, k represents the clutter cell number, and w is the forgetting factor. This represents the clutter intensity estimate after the clutter map is updated. Expanding the above equation, we get: (6), False alarm probability P fa The forgetting factor w is 1 / 8, and the relative threshold T of clutter and the false alarm probability can be expressed as: (7), When designing a detector, a false alarm probability P is given.fa Given the forgetting factor w, the relative threshold T can be calculated using the above formula. Compare the echo intensity with the absolute threshold. If the echo intensity is greater than the absolute threshold, then a target exists; otherwise, no target exists. Figure 9 This includes multi-channel, multi-dimensional static clutter maps and dynamic clutter maps. The multi-channel clutter map mainly incorporates Doppler information, dividing the clutter within a CPI into Doppler dimensions to obtain clutter maps with multiple Doppler channels. Detection is performed on multiple channels, which helps improve target detection performance. This example has been verified in various ground scenarios. The remaining clutter map covers a maximum of 3 Doppler channels. Therefore, this solution only establishes clutter maps for 3 channels, and the remaining channels are processed as normal. The workflow for multi-channel clutter map detection is as follows: Based on the radar scanning angle range α and the 3dB beamwidth β, using the formula: (8) Obtain the number of radar azimuths N, where [·] is the floor function, and then use formula (10): (9) Angle of each azimuth resolution unit ; Based on the 3dB beamwidth, the number of accumulation points within one PRT is calculated for moving target detection. Assume k is the Doppler dimension number, and denote the m-th ring, the k-th Doppler channel, and the azimuth angle. The data is It is the forgetting factor, according to the formula (10) Update the clutter plot to a stable state.

[0028] like Figure 9 The clutter detection shown is based on the clutter map established by the m-th scan. and the scan data of the (m+1)th cycle Clutter detection is performed on the data from the (m+1)th cycle to obtain the detection results. Based on... and The size relationship between them determines whether a target exists. (11) Detection has a target, where h is the threshold factor.

[0029] This example addresses the market demand for ground reconnaissance radar in protecting critical border security surveillance facilities and countering drones. In reality, slow-moving targets (such as personnel, vehicles, and drones) are affected by ground clutter. Most radars face insufficient detection capability for targets with slow radial speeds. Due to system limitations, the MTI filter order cannot be designed too high, and its wide transition band inevitably leads to a loss of target energy during broadband ground clutter suppression. Based on traditional signal processing, this example proposes the idea of ​​using phasor mean filtering combined with a multi-dimensional spatial clutter map for slow-moving target detection. By applying zero-frequency clutter suppression and an improved clutter map to slow-moving ground target detection, this example enhances the radar's ability to detect ground targets.

[0030] Compared with existing technologies, this invention has the following advantages: it improves the radar's ability to resist clutter and detect low-speed targets; it distinguishes between clutter-inside and clutter-outside targets using clutter profile maps, avoiding performance loss caused by suppressing clutter outside the clutter area; it employs a phasor mean cancellation algorithm, which greatly improves the signal-to-noise ratio of moving or slightly moving targets while suppressing the phase of stationary targets; it maintains a high signal-to-noise ratio by utilizing the advantages of the phasor mean cancellation algorithm, such as no reduction in target amplitude, relatively clean background noise, and relatively complete preservation of micro-Doppler information; the combination of planar and three-dimensional clutter maps enables more refined division of clutter regions, and the planar clutter map reduces storage and computational requirements; it uses dual-branch signal processing, with one branch processing data from clutter regions and the other processing data from non-clutter regions, thus avoiding detection performance loss caused by clutter suppression processing; it can achieve very ideal results for detecting ground personnel targets; it has strong engineering feasibility and has promotional application value.

Claims

1. A ground surveillance radar ground clutter suppression method, characterized by, Firstly, the echo data is used to establish a clutter profile, and according to the clutter profile, the information inside and outside the clutter is judged, and different signal processing procedures are selected, wherein the outside clutter uses a detection channel 1, the inside clutter uses a detection channel 2 and a detection channel 3, the detection channel 1 is outputted after passing through an MTD filter without clutter suppression processing; the detection channel 2 is a coherent processing channel, which is firstly subjected to mean filtering processing, and then subjected to coherent processing, and the detection result is outputted after the clutter suppression processing through the phasor mean filtering; the detection channel 3 is a zero Doppler processing channel, which is used for detecting low-speed targets in the clutter area, and the original data is subjected to phasor mean filtering processing to obtain residual clutter, and a residual multi-dimensional clutter map is established by using the residual clutter, and then the multi-dimensional clutter map CFAR is used to detect targets.

2. The ground surveillance radar ground clutter suppression method of claim 1, wherein, The intermediate frequency echo signal of the detection radar is received, and then A / D sampling processing, digital down conversion processing and pulse compression processing are sequentially performed, and then phasor mean filtering processing, coherent accumulation processing and difference normalization processing are performed, and then CA-CFAR or multi-dimensional clutter map CFAR is used to complete detection.

3. The ground surveillance radar ground clutter suppression method of claim 2, wherein, The clutter profile is established by firstly estimating the clutter power of a clutter unit, performing a first level comparison of the clutter power estimation value and a noise threshold, if the clutter power estimation value is greater than the noise threshold, then the counter is increased by 1; and performing a second level comparison of the threshold according to K / L criterion, that is, the counter saves the number of times of passing the first threshold in the last L times of scanning, if the counter is greater than K, then the second threshold is passed, which indicates that it is inside the clutter, otherwise it is outside the clutter.

4. The ground surveillance radar ground clutter suppression method of claim 3, wherein, The clutter profile establishment includes establishment of a multi-channel spatial static clutter map and a dynamic clutter map, wherein the multi-channel clutter map adds Doppler velocity information on the plane clutter map, divides the clutter in a CPI in the Doppler velocity to obtain a clutter map of multiple Doppler velocity channels, and detects on the multi-channel, so that the plane and the three-dimensional clutter map are combined to realize more fine division of the clutter area, and the three-dimensional clutter map reduces the storage and operation amount, which is beneficial to improve the target detection performance.

5. The ground surveillance radar ground clutter suppression method of claim 4, wherein, The phasor mean filtering processing uses the characteristics that the distance from a stationary target to a radar antenna is constant and the time delay on the received pulse is constant, and the mean value obtained by accumulation after phasor summation of the stationary target is offset, so that the static clutter amplitude is attenuated, and the phasor mean filtering processing specifically includes: firstly, all received pulses are averaged to obtain a reference received pulse, and then each received pulse is used to subtract the reference received pulse to obtain a target echo signal, and the expression of the reference received pulse is: （1）， wherein, is the fast time or distance dimension sample point, is the slow time or velocity dimension time sample point, the formula of the phasor mean cancellation algorithm is: （2）， Before the coherent accumulation operation, firstly, mean filtering is carried out, and the accumulated multi-frame distance dimension data is stored in the SRAM, and after the echo of the current beam accumulation is completely stored, the velocity dimension data is read out in sequence according to the velocity dimension and frame; secondly, each frame of velocity dimension data is subtracted by the average value of the frame to remove the clutter component of zero frequency; then, it is weighted by Taylor window to suppress the spectrum leakage; finally, FFT operation is carried out, wherein the mean filtering firstly carries out data storage rearrangement, and the specific process is as follows: on the peripheral SRAM, the SRAM is divided into two areas according to the address to carry out ping-pong operation, the area distribution of the SRAM is transversely represented as distance dimension and longitudinally represented as velocity dimension, each frame of distance dimension echo data is written into the SRAM in sequence by row, the point number of each frame of data is 8192 points, and a total of 128 frames of echo data are written, due to the pipeline processing in the FPGA, when the current wave position data is written, the data of the previous wave position is also read out at the same time, the reading data is taken out by column, each frame of data has 128 points, and a total of 8192 frames of data are read out, and different frames represent different distance gates; In the mean filtering process, the data mean value is obtained by accumulation, first, it is judged whether the input data is the first frame of pulse accumulation data, if yes, it is directly cached into the RAM, if not, the last frame of cached data is read out from the RAM and added to the frame data when the data is valid, and then written into the RAM, wherein the RAM is an IP core, the IP core is configured as a simple dual-port RAM, and the reading and writing operations are carried out at the same time, the input data and the stored data are cyclically added until the last frame of data, the accumulation of 128 frames of data is completed, the result of the accumulation is divided by 128 to obtain the average value of the velocity dimension data of each distance gate, and in the FPGA, direct division or calling of the division IP core will consume DSP and LUT resources, and 128 is 2 raised to the power of 7, and the division operation is realized by right shifting the data by 7 bits.

6. The ground surveillance radar ground clutter suppression method of claim 4, wherein, The clutter map establishing step includes: first, according to different requirements, the radar range is divided into many clutter units, the clutter unit is greater than the target resolution unit, The distance dimension of the clutter unit is represented by ΔR, The azimuth dimension is represented by ΔA, and ΔR is the distance resolution unit of the radar, The antenna scanning angle corresponding to one pulse repetition period PRT is represented by θ, and it is assumed that And That is, one clutter unit spans M distance units and receives N pulses; The clutter strength of the estimation clutter unit is estimated, and for each clutter unit, M*N echo data can be obtained in one antenna scanning, N is the number of pulses, and M is the number of distance units. In order to estimate the clutter strength of the unit, first, N pulses are accumulated, which can be represented as: (3), where r denotes the number of antenna scans, k denotes the number of clutter cells, m = 1, 2,..., M denotes the number of range cells, and n = 1, 2,..., N denotes the number of pulses, after which pulse accumulation results in echo data over M range cells The clutter strength is estimated using these data and is given by (4) clutter map update and detection, when the antenna scans to the direction of the clutter cell, the clutter intensity of the clutter cell is estimated and written into the clutter map storage unit, after one antenna scanning cycle, a complete clutter map is formed, and a stable clutter map is established after accumulation of several to tens of antenna scanning cycles.

7. The ground surveillance radar ground clutter suppression method of claim 6, wherein, The clutter map detection includes: Xr represents the single clutter unit single frame clutter strength estimation value, Xr is taken as the input of the update algorithm, and the accumulation formula of the clutter map update can be represented as: （5）， where r denotes the current radar sweep number, k denotes the clutter cell index, w is a forgetting factor, denotes the updated clutter intensity estimate, expand the above equation to: （6）， False alarm probability P fa With the forgetting factor w = 1 / 8, the clutter relative threshold T threshold and the false alarm probability can be expressed as: （7）， When designing the detector, given a false alarm probability P fa and a forgetting factor w, the relative threshold T can be computed from the above equation, and the echo intensity is compared with the absolute threshold. If it is greater than the absolute threshold, there is a target, otherwise there is no target.

8. The ground surveillance radar ground clutter suppression method of claim 7, wherein, The multi-channel clutter map adds Doppler velocity information, divides the clutter in one CPI in the Doppler velocity to obtain the clutter maps of multiple Doppler velocity channels, and detects on the multi-channel to improve the detection performance of the target, wherein the residual clutter map covers at most 3 Doppler channels, so only three channel clutter maps are established, and the rest of the channels are processed according to the normal detection; the working process of the multi-channel clutter map detection is as follows: According to the radar scanning angle range α and 3dB beam width β, through the formula (8) to obtain the number of azimuths N of the radar, where [· ] is a down-rounding operation, and then through formula (9): (9) the angle of each azimuthally resolving element ; According to 3dB beam width, the accumulated point number in a PRT is calculated to carry out moving target detection. Assuming k is the number of Doppler frequency, the data of the mth circle, the kth Doppler channel and the azimuth angle is is a forgetting factor, according to formula (10): (10) update the clutter map to be stable; clutter map established according to the mth scan and the (m+1)th scan data , the (m+1)th scan data is subjected to clutter detection to obtain a detection result, and whether there is a target is determined according to the size relationship between and ​ (11), detecting a target, where h is a threshold factor.

Images