A Radar Weak Target Detection Method Based on Improved CFAR

Abstract

The present invention relates to a method for detecting weak radar targets based on improved CFAR, comprising the following steps: performing two-dimensional FFT processing on radar echoes to obtain a range-Doppler map; performing CFAR calculations on the Doppler dimension and the range dimension of the RD map respectively, and by optimizing the settings of the reference unit, the protection unit and the adjustment of the SNR threshold, ensuring that the target unit is detected in both dimensions and is determined as a target only when the power is higher than that of adjacent units; for special Doppler indices and edge range units in the zero Doppler dimension, calculating the power difference to further improve the detection ability of weak targets; the present invention can stably detect weak targets in the DC region and the large-angle sidelobe region of the radar, effectively solve the problems of complex front-end signal processing and parameter setting in the prior art, and significantly improve the detection rate and accuracy of weak targets in complex scenarios.

Description

Technical Field

[0001] The present invention relates to the technical field of radar, and particularly to a method for detecting weak radar targets based on improved CFAR. Background Art

[0002] The constant false alarm rate (CFAR) detection technology is a commonly used method in radar signal processing for detecting targets in complex noise and clutter environments. The CFAR technology dynamically adjusts the detection threshold to keep the false alarm rate constant, so as to effectively detect targets in different environments. Millimeter-wave radars are often used in scenarios such as autonomous driving, unmanned aerial vehicles, and security monitoring due to their high resolution and anti-interference capabilities. In these applications, the CFAR technology is used to extract target signals from complex background noise. Therefore, the CFAR technology is widely applied in fields such as millimeter-wave radars and airborne radars.

[0003] Although the CFAR detection technology has been widely applied in radar target detection, there are still deficiencies in weak target detection:

[0004] Complex front-end signal processing: Existing weak target detection mainly focuses on the front-end radio frequency signal processing part, and improves the weak target detection performance by enhancing the signal processing ability, such as increasing the signal-to-noise ratio (SNR) and the signal processing bandwidth. However, these methods are complex and limited by hardware costs and computing resources.

[0005] Limited ability to detect weak targets: CFAR detection relies on the prominent peaks of target signals, while the signal intensity of weak targets is low and is easily submerged by noise or clutter, resulting in missed detections. Especially at both ends of the radar ranging range (the nearest and the farthest), due to the influence of the DC component, it is more difficult to detect weak targets.

[0006] The parameter settings have a great impact on the detection results: The effect of CFAR detection highly depends on the settings of parameters such as reference units and protection units. Different application scenarios require different parameter configurations, and improper parameter settings will lead to a decline in detection performance. Especially in weak target detection, small changes in parameter settings may result in missed detections of targets or an increase in the false alarm rate.

[0007] Difficult detection in the large-angle sidelobe region: When the target is in the sidelobe region of the radar beam, the signal intensity is weak, and the signals in the sidelobe region are usually masked by noise or clutter. Conventional CFAR technologies are difficult to effectively detect target signals, resulting in missed detections of targets. Summary of the Invention

[0008] In view of this, the present invention improves the CFAR algorithm in the backend signal processing, provides a radar weak target detection method based on improved CFAR, and can stably detect weak targets in the DC region and large-angle sidelobe region of the radar, effectively solving the problems of complex front-end signal processing and parameter setting in the prior art.

[0009] To achieve the above object, a radar weak target detection method based on improved CFAR of the present invention includes the following steps:

[0010] S1. Perform two-dimensional fast Fourier transform (FFT) processing on the radar echo to obtain a range-Doppler map;

[0011] S2. Perform CFAR calculations on the Doppler dimension and range dimension of the range-Doppler map respectively;

[0012] S201. Starting from the Doppler dimension, each cell of the range-Doppler map is sequentially used as a target cell by row, and the power of the target cell is calculated;

[0013] S202. The a cells adjacent to the target cell on the left and right are used as guard cells, and the b cells adjacent to the guard cells on the left and right are used as reference cells;

[0014] S203. Calculate the average power values of the reference cells on both the left and right sides respectively, and the smaller average power value is used as the background power;

[0015] S204. Compare the power of the target cell with the sum of the background power and a preset signal-to-noise ratio (SNR). If the power of the target cell is greater than the sum of the background power and the preset signal-to-noise ratio, and greater than the power of the two adjacent cells on the left and right, it is determined that there is a target in the target cell and it is detected;

[0016] S205. Repeat steps S201 - S204 to perform CFAR calculation from the range dimension;

[0017] S206. If the target cell is detected in both the Doppler dimension and the range dimension and the power of the target cell is higher than the adjacent cells on the left and right, it is determined that there is a target in the target cell, otherwise, it is determined that there is no target in the target cell;

[0018] S3. Perform step S2 using a lower preset signal-to-noise ratio for Doppler indices 2, 3, N - 1, and N, where N represents the number of radar chirps;

[0019] S4. Perform zero Doppler dimension processing;

[0020] S5. Output the detected target cells.

[0021] Preferably, when the target cell with a Doppler index of 3 compares its power with the adjacent cells on the left and right, it does not compare with the cell with a Doppler index of 2, but only with the cell with a Doppler index of 4.

[0022] Preferably, when the target cell with a Doppler index of N - 1 compares its power with the adjacent cells on the left and right, it does not compare with the cell with a Doppler index of N, but only with the cell with a Doppler index of N - 2.

[0023] Preferably, the zero - Doppler dimension processing includes the following steps:

[0024] Calculate the power difference between the first range cell and the second range cell in the zero - Doppler dimension, denoted as , calculate the power difference between the second range cell and the third range cell in the zero - Doppler dimension, denoted as , calculate the power difference between the (M - 1)th range cell and the Mth range cell in the zero - Doppler dimension, denoted as , calculate the power difference between the (M - 2)th range cell and the (M - 1)th range cell in the zero - Doppler dimension, denoted as , where M represents the number of ADC sampling points;

[0025] Compare , , and with a preset threshold TD respectively. If only one power difference satisfies , then detect the range cell that is subtracted in the power difference;

[0026] If there are two or more power differences that satisfy , then detect the range cell with a higher power among the range cells, but the first range cell and the last range cell cannot be used as the detected cells.

[0027] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0028] The present invention utilizes the characteristic that the target energy will spread among adjacent range cells, improves the detection rate of weak targets at the nearest and farthest positions of the radar ranging range in a complex environment, and successfully detects weak targets in the sidelobe region of the radar beam by optimizing the CFAR comparator logic, reducing the problem of target missed detection caused by noise or clutter;

[0029] For special regions such as both ends and the sidelobe region of the radar ranging range, the present invention adopts targeted optimization processing methods, reduces the SNR and adjusts the power comparison rules, avoiding misjudgment caused by DC influence, so that weak targets can also be accurately detected in the sidelobe region. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 is the range-Doppler map obtained by performing two-dimensional FFT processing on the radar echo signal in step S1 of the present invention;

[0031] Figure 2 It is a schematic diagram of target detection by the CFAR algorithm of the present invention;

[0032] Figure 3 It is a comparison diagram of the one-dimensional range profiles of weak targets in the near range and an empty scene of the present invention. Detailed implementation manners

[0033] To further elaborate on the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following will, in conjunction with the accompanying drawings and preferred embodiments, detail the specific implementation manners, structures, features and their effects of the present invention as follows.

[0034] To ensure the stable detection and non-missing detection of weak targets in the DC region and large-angle sidelobe region of the radar, the present invention provides the following technical solution: a radar weak target detection method based on improved CFAR. In this embodiment, the maximum measurement distance is 100 meters, the number of ADC sampling points is 256, the number of chirps is 64, the number of reference cells is 8, the number of guard cells is 4, the signal-to-noise ratio SNR in the Doppler dimension is 10 dB, the signal-to-noise ratio SNR in the range dimension is 10 dB, and the threshold value TD of the unit power difference is 1500. The method includes the following steps:

[0035] S1. As Figure 1 shown, perform two-dimensional fast Fourier transform (FFT) processing on the radar echo to obtain a range-Doppler map (Range-Doppler map). Figure 1 Each grid of the RD map shown represents a cell;

[0036] S2. Perform CFAR calculations on the Doppler dimension and range dimension of the range-Doppler map respectively:

[0037] S201. Starting from the Doppler dimension, take each cell of the range-Doppler map row by row as the target cell, and calculate the power of the target cell;

[0038] S202. The 4 cells adjacent to the target cell on the left and right are used as guard cells, and the 8 cells adjacent to the left and right of the guard cells are used as reference cells;

[0039] S203. Calculate the average power of the reference cells on the left and right sides respectively, and take the smaller average power as the background power;

[0040] S204. Compare the power of the target cell with the sum of the background power and the preset signal-to-noise ratio SNR. If the power of the target cell is greater than the sum of the background power and the preset signal-to-noise ratio, and greater than the power of the two adjacent cells on the left and right, it is determined that there is a target in the target cell and it is detected;

[0041] S205. Repeat steps S201 - S204 to perform CFAR calculation in the range dimension;

[0042] S206. If the target cell is detected in both the Doppler dimension and the range dimension and the power of the target cell is higher than that of the adjacent cells on the left and right, it is determined that there is a target in the target cell; otherwise, it is determined that there is no target in the target cell;

[0043] For the cells at both ends, when selecting the left and right cells, they are selected in a loop-back manner. For example, if there is no cell on the left of the first cell, during the calculation, the four rightmost cells, that is, cells with indices 61 - 64, are used as guard cells, and the cells with indices 53 - 60 are used as reference cells;

[0044] S3. After retrieving the targets in the RD map in both the range dimension and the Doppler dimension, supplement the detected targets to avoid missing detections;

[0045] For Doppler indices 2, 3, N - 1, and N, perform step S2 with a lower preset signal - to - noise ratio. The preset signal - to - noise ratio SNR is reduced by 3 dB respectively, where N represents the number of radar chirps (in this embodiment, the maximum Doppler index is 64, that is, the number of radar chirps is 64);

[0046] When the cell in the third column is used as the target cell to compare the power with the adjacent cells on the left and right, it is not compared with the previous cell, that is, the cell with Doppler index 2, but only compared with the next cell, that is, the cell with Doppler index 4;

[0047] When the cell in the 63rd column is used as the target cell to compare the power with the adjacent cells on the left and right, it is not compared with the cell with Doppler index 64, but only compared with the cell with Doppler index 62;

[0048] S4. Zero - Doppler dimension processing, including the following steps:

[0049] Calculate the power difference between the first 3 and the last 3 range cells in the first column of the RD map, that is, the power difference between the first and the second range cells in the zero - Doppler dimension, denoted as , calculate the power difference between the second and the third range cells in the zero - Doppler dimension, denoted as , calculate the power difference between the 256th and the 255th range cells in the zero - Doppler dimension, denoted as , calculate the power difference between the 255th and the 254th range cells in the zero - Doppler dimension, denoted as ;

[0050] Let , , and Compare with the preset threshold value TD respectively. If there is only one power difference that satisfies , then detect the distance unit that is subtracted in the power difference. For example, if , then detect the second distance unit; if , then detect the 255th distance unit;

[0051] If there are two or more power differences that satisfy , then detect the distance unit with higher power among the distance units, but the first distance unit and the last distance unit cannot be used as the detected units. For example 、 both satisfy the conditions, then select and detect the distance unit with higher power among the two units for subtraction respectively;

[0052] As Figure 3 shown, the power schematic diagram of the first few distance units of the zero Doppler one-dimensional range image when there is a weak target at 0.4 meters nearby is given. It can be seen that when there is a target, the power of the second distance unit is significantly higher than that of the empty scene. However, because the power of the first distance unit is higher than that of the second distance unit due to the influence of direct current, the conventional CFAR algorithm fails to detect it, but it can be successfully detected through this step;

[0053] S5. Output all the target units detected in the above steps.

[0054] The above is only a preferred embodiment of the present invention, and it does not impose any form of limitation on the present invention. Although the present invention has been disclosed above with a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art can make some changes or modifications to it as equivalent embodiments within the scope of the technical solution of the present invention. However, as long as it does not depart from the content of the technical solution of the present invention, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention still fall within the scope of the technical solution of the present invention.

Claims

1. A radar weak target detection method based on improved CFAR, characterized in that: The following steps are involved: S1, perform two-dimensional fast Fourier transform (FFT) processing on the radar echo to obtain a range-Doppler map; S2, performing CFAR calculation on the Doppler dimension and the distance dimension of the range-Doppler image respectively; S201, starting from the Doppler dimension, taking each unit in the range-Doppler map as a target unit in turn by row, and calculating the power of the target unit; S202, a units adjacent to the target unit on the left and right are used as protection units, and b units adjacent to the protection unit on the left and right are used as reference units; S203, respectively calculating the power averages of the reference units on the left and right sides, wherein the smaller power average is used as the background power; S204, comparing the target unit power with the sum of the background power and the preset signal-to-noise ratio SNR. If the target unit power is greater than the sum of the background power and the preset signal-to-noise ratio, and greater than the power of the two adjacent units on the left and right, it is determined that there is a target in the target unit and detected; S205, repeat steps S201-S204, and perform CFAR calculation from the distance dimension; S206, if the target unit is detected in both the Doppler dimension and the range dimension and the power of the target unit is higher than that of the left and right adjacent units, it is determined that the target unit has a target, otherwise, it is determined that the target unit does not have a target; S3, performing step S2 for Doppler indexes 2, 3, N-1, and N using a lower preset signal-to-noise ratio, where N represents the number of radar chirps; S4, zero-Doppler processing, comprising the following steps: Calculate the power difference between the first and second distance units of zero Doppler dimension, denoted as d1 = |P1-P2|, calculate the power difference between the second and third distance units of zero Doppler dimension, denoted as d2 = |P2-P3|, calculate the power difference between the M-1th distance unit and the Mth distance unit of zero Doppler dimension, denoted as d3 = |P M -P M-1 |, calculate the power difference between the M-2th distance unit and the M-1th distance unit of zero Doppler spectrum, denoted as d4=|P M-1 -P M-2 |, where M represents the number of ADC sampling points; Compare d1, d2, d3, and d4 with a preset threshold value TD respectively. If there is only one power difference that satisfies d i < TD, where i = 1, 2, 3, 4, then detect the distance unit being subtracted in the power difference. If there are two or more power differences that satisfy d i <TD, where i = 1, 2, 3, 4, then detect the range cell with the higher power in the detection range cells, but the first range cell and the last range cell cannot be used as detection cells; S5. Output the detected target unit.

2. The radar weak target detection method based on improved CFAR according to claim 1 is characterized in that: When the target cell with the Doppler index of 3 performs power comparison with the adjacent cells on the left and right sides, it is not compared with the cell with the Doppler index of 2, but only compared with the cell with the Doppler index of 4.

3. The radar weak target detection method based on improved CFAR according to claim 2 is characterized in that: When the target cell with the Doppler index of N-1 performs power comparison with the adjacent cells on the left and right sides, it is not compared with the cell with the Doppler index of N, but only with the cell with the Doppler index of N-2.

Images