A Radar CFAR Target Detection Method Based on the Prefix Sum Algorithm

Abstract

The present invention discloses a radar CFAR target detection method based on the prefix sum algorithm. The method performs Fourier transform and square-law detection on the radar echo signal to obtain the range-Doppler matrix A, and performs constant false alarm rate (CFAR) processing on the range-Doppler matrix A to determine whether a target exists. When performing CFAR detection, a rectangular reference sliding window is used. First, the prefix sum algorithm is used to obtain the prefix sum matrix SUM of the range-Doppler matrix A, and then the average value of the data of the surrounding reference cells of the cell a to be measured is calculated through the prefix sum matrix. The average value is multiplied by a coefficient to obtain an estimated value of the background clutter, and the estimated value is compared with the cell to be detected to determine whether the cell to be detected is a target. The method of the present invention ensures the accuracy of detection while reducing the complexity of the algorithm, reducing the computational amount of the original CFAR detection, and improving the detection efficiency. i,j The method of the present invention ensures the accuracy of detection while reducing the complexity of the algorithm, reducing the computational amount of the original CFAR detection, and improving the detection efficiency.

Description

Technical Field

[0001] The present invention relates to the field of radar signal processing, and in particular to a radar CFAR target detection method based on a prefix sum algorithm. Background Art

[0002] With the increasing volume of road traffic, traffic monitoring and traffic safety have attracted more and more attention. Millimeter-wave radar has the advantages of small size, light weight, low cost, little influence by weather and environmental factors, high resolution, etc., and can greatly reduce the accident rate in bad weather and environment. It has broad application prospects in the fields of vehicle-mounted anti-collision radar, automatic navigation control, etc. There are clutter and various noise interferences during the monitoring process of millimeter-wave radar. Detecting the presence of targets in these complex background environments and maintaining a constant false alarm probability is an important part of radar target detection. Constant false alarm rate (CFAR) processing refers to the technology in which a radar system discriminates the signal and noise output by a receiver under the condition of keeping the false alarm probability constant to determine whether a target signal exists, including classical CFAR algorithms such as mean-based CFAR and statistically ordered CFAR. Among them, CA-CFAR (Cell Average-Constant False Alarm Rate) in mean-based CFAR is a commonly used algorithm with the least loss rate in engineering applications.

[0003] The echo signal is subjected to CFAR detection after Fourier transform and square-law detection to determine whether there is a target in the unit to be detected. Generally, a radar system considers range-Doppler two-dimensional information during decision-making. The CFAR detection principle is as Figure 1 shown, where the reference unit, protection unit, and unit to be detected are called the CFAR processing window. The unit to be detected is in the middle. The mean value of the reference unit data is used to estimate the background clutter. The protection unit is adjacent to the unit to be detected, and the data samples of the protection unit are not included during the averaging process. The reason for this processing is that when there is a target spanning multiple units, the power of the target may leak into adjacent units. The threshold is determined by multiplying the estimated value of the clutter by a coefficient, and this value is compared with the value of the unit to be detected. If it exceeds this threshold, it is judged that there is a target; otherwise, it is judged that there is no target. In traditional CFAR detection, the operation steps of the mean value of the reference unit are repetitive and have a high complexity. For each decision of the unit to be detected, it is necessary to traverse and calculate the mean value of the two-dimensional reference unit. Therefore, the time complexity is O(n 4 ), with a high complexity and a long operation time.

[0004] In order to improve the processing efficiency of the CFAR algorithm, the patent "A fast two-dimensional CFAR target detection system and detection method" (application number: CN201811267196.5) first performs one-dimensional CFAR detection on the distance dimension, extracts the points that pass the threshold, and then performs two-dimensional CFAR processing on these points. The patent "A two-dimensional scaled fast constant false alarm detection method" (application number: CN201811344417.4) divides the two-dimensional detection plane of the radar into equal blocks, first extracts the sub-blocks with targets, and then performs two-dimensional CFAR detection on them. The patent "An improved two-dimensional constant false alarm method" (application number: CN201210288754.2) finds the maximum value in the Doppler unit under each distance unit, saves the Doppler unit address corresponding to the maximum value, and then performs one-dimensional CFAR detection on the corresponding Doppler unit along the distance dimension. Although the above method reduces the processing time of the algorithm, it sacrifices performance and is prone to missed alarms and false alarms. Summary of the invention

[0005] Aiming at the high time complexity of the CA-CFAR target detection method used in existing radars, the present invention proposes a CFAR target detection method based on prefix sum algorithm, which can greatly save calculation time.

[0006] The technical solution adopted by the present invention is:

[0007] A radar CFAR target detection method based on prefix sum algorithm comprises the following steps:

[0008] S1, using radar to emit millimeter wave signals to detect moving targets;

[0009] S2, determine whether the target exists. The specific steps are as follows:

[0010] S21, performing two-dimensional Fourier transform and square-rate detection on the echo signal received by the radar to obtain a range-Doppler matrix A;

[0011] S22, select the target a to be detected ij , a ij Represents the value of the i-th row and j-th column of the range-Doppler matrix A, which determines the size of the surrounding protection unit and the reference unit; due to the particularity of the elements at the edge, it is necessary to perform edge expansion on the range-Doppler matrix A according to the circular symmetry of the spectrum so that the values ​​at the edge of the matrix also meet the conditions of the CFAR operation;

[0012] S23, calculating the prefix sum matrix SUM of the range-Doppler matrix A according to the prefix sum principle;

[0013] S24, calculate the target a to be detected by using the prefix sum matrix SUM obtained in step S23 i,j Surrounding reference unit R data ri,j sum ∑ i,j∈R r i,j and obtain their arithmetic mean where K is the number of reference cells K;

[0014] S25. Multiply the arithmetic mean Z calculated in step S24 by the threshold product factor α to calculate the estimated value of the detection threshold k which is the estimated value of the background clutter;

[0015] S26. Compare the target a to be detected i,j with the threshold estimated value T. If the value of the target a to be detected i,j is greater than the threshold estimated value T, it is determined that a target exists; if the value of the target a to be detected ij is less than the threshold estimated value T, it is determined that the target does not exist; i,j

[0016] S27. Increase the values of i and j one by one, and repeat steps S22 - S26. After completing the detection of all elements in the range - Doppler matrix A, end the target detection.

[0017] Furthermore, in step S23, the calculation formula for the prefix sum matrix SUM is:

[0018] sum i,j = sum i-1,j + sum i,j-1 - sum i-1,j-1 + a i,j

[0019] where sum i,j and a i,j are the values of the prefix sum matrix SUM and the range - Doppler matrix A at the i - th row and j - th column respectively.

[0020] Furthermore, in step S24, use the prefix sum matrix SUM to calculate the sum of the data of the reference cells around the target a to be detected i,j which is the difference between the matrix A(i - M:i + M, j - N:j + N) and the matrix A(i - m:i + m, j - n:j + n):

[0021] A(i - M:i + M, j - N:j + N) - A(i - m:i + m, j - n:j + n)

[0022] = sum i+M,j+N - sum i-M-1,j+N - sum i+M,j-N-1 + sum i-M-1,j-N-1 - (sum i+m,j+n - sum i-m-1,j+n - sum i+m,j-n-1 + sumi-m-1,j-n-1 )

[0023] Where M and N are the length and width of the reference unit, respectively; m and n are the length and width of the protection unit, respectively.

[0024] The present invention uses the prefix sum algorithm to quickly calculate the sum of the reference unit data in CFAR detection, and then calculate the estimated value of clutter interference. The CFAR detection method optimized by the prefix sum algorithm only needs to traverse the range-Doppler matrix once in advance to calculate its prefix sum matrix, and the time complexity is O(n 2 ), compared with the traditional radar CFAR method, there is no need to traverse the two-dimensional matrix once when comparing each element, the time complexity is reduced, the amount of calculation is sharply reduced, and the detection efficiency is greatly improved while ensuring the accuracy of two-dimensional detection. Therefore, compared with the current prior art, the present invention can not only perform constant false alarm rate detection of target detection well, but also reduce the calculation time complexity, reduce the calculation amount of existing CFAR detection, and greatly improve the efficiency of CFAR detection method. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 It is a schematic diagram of the principle of the CFAR detection method;

[0026] Figure 2 is a schematic diagram of the prefix sum algorithm;

[0027] Figure 3 It is a flow chart of the radar CFAR target detection method of the present invention;

[0028] Figure 4 : (a) the range-Doppler matrix in the embodiment and (b) the processing result of the method of the present invention. DETAILED DESCRIPTION

[0029] The present invention is described in detail below according to the accompanying drawings, which is a preferred embodiment of the present invention in various embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0030] The method of this embodiment uses a millimeter wave radar to collect the echo signal of a moving vehicle, performs a two-dimensional Fourier transform and a square-rate detection on the echo signal received by the radar to obtain a range-Doppler matrix A, and selects a to-be-detected target a ij , is the value of the i-th row and j-th column of the range-Doppler matrix A. Due to the particularity of the elements at the edge, it is necessary to perform edge expansion on the range-Doppler matrix A according to the circular symmetry of the spectrum so that the values ​​at the edge of the matrix also meet the conditions of the CFAR operation.

[0031] According to the prefix sum principle, calculate the prefix sum matrix SUM of the range-Doppler matrix A. The prefix sum algorithm is a preprocessing method that calculates the sum of the first n sub-arrays of an array. For a one-dimensional array a[i], its prefix sum array sum[i] represents the sum of all items before i, which can be written as:

[0032]

[0033] For a two-dimensional matrix, each element in the prefix sum matrix represents the sum of all elements in the matrix with the element at the corresponding position in the original matrix as the lower right corner. The prefix sum matrix can be calculated according to the following formula:

[0034]

[0035] In repeated addition operations, the prefix sum matrix can be used to greatly save calculation time. As Figure 2 shown, when finding the sum of the sub-matrix with the line connecting the two points (i,j) and (i+p,j+q) in the original matrix as the diagonal, the prefix sum matrix sum of the original matrix can be first calculated and obtained through the following formula:

[0036] sum i+p,j+q -sum i-1,j -sum i+p,j-1 +sum i-1,j-1

[0037] The mean operation of each reference unit in CFAR detection is a repeated operation. As Figure 1 shown, it can be obtained by calculating the difference between the sub-matrices A(i-M:i+M,j-N:j+N) and A(i-m:i+m,j-n:j+n) of the range-Doppler matrix:

[0038] A(i-M:i+M,j-N:j+N)-A(i-m:i+m,j-n:j+n)

[0039] =sum i+M,j+N -sum i-M-1,j+N -sum i+M,j-N-1 +sum i-M-1,j-N-1 -(sum i+m,j+n -sum i-m-1,j+n -sum i+m,j-n-1 +sum i-m-1,j-n-1 )

[0040] where M and N are the lengths of the reference unit and the protection unit respectively, and m and n are the widths of the reference unit and the protection unit respectively.

[0041] Calculate the sum ∑ of the data r of the surrounding reference units R of the target a to be detected through the prefix sum matrix SUM i,j around the target a i,j of the data r of the surrounding reference units Ri,j∈ R r i,j , and obtain their arithmetic mean where K is the number of reference units K. Multiply the arithmetic mean Z k by the threshold multiplication factor α using a multiplier to calculate the estimated value of the detection threshold which is the estimated value of the background clutter. Then, compare the target a to be detected i,j with the threshold estimated value T using a comparator. If the value of the target a to be detected i,j is greater than the threshold estimated value T, it is determined that a target exists. If the value of the target a to be detected ij is less than the threshold estimated value T, it is determined that no target exists. Finally, increment the i value and the j value one by one, and repeat the above steps. After detecting all the elements in the range-Doppler matrix A, the target detection ends.

[0042] Embodiment

[0043] In radar signal processing, the most time-consuming operations are 2D-FFT and CFAR detection. In this embodiment, M = N = 13, m = n = 5, K = 144, and α = 14.5. Table 1 shows the time consumed for FFT and CFAR processing of range-Doppler matrices with different dimensions. It can be seen that as the dimension of the range-Doppler matrix increases, the time consumed by these two steps increases. The time taken to process using the traditional CFAR method is about 10 times that of 2D-FFT. However, the time taken to process using the method of the present invention is about 1 / 60 of the time taken by the traditional CFAR method and about 1 / 6 of the time taken by 2D-FFT. Compared with the traditional CFAR method and 2D-FFT processing, the time consumed by the method of the present invention is greatly reduced.

[0044] Table 1. Time consumed for FFT and CFAR processing of range-Doppler matrices with different dimensions

[0045] Matrix dimension 2D-FFT time consumption / ms Traditional CFAR time consumption / ms CFAR time consumption of the present invention / ms 256×1024 28.5 238.2 4.9 256×512 13.6 118.9 3.7 256×256 6.7 62.8 1.9 64×64 0.35 3.7 0.16

[0046] Figure 4 For a 64×64 range-Doppler matrix, use the traditional CFAR detection method and the CFAR method of the present invention to perform CFAR detection on this matrix. The two detections are consistent, indicating that the operation results of the two CFAR methods are the same.

[0047] Those of ordinary skill in the art will understand that the above description is only a preferred embodiment of the invention and is not intended to limit the invention. Although the invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or perform equivalent replacements for some of the technical features. Any modifications, equivalent replacements, etc. made within the spirit and principle of the invention shall be included within the protection scope of the invention.

Claims

1. A radar CFAR target detection method based on the prefix sum algorithm, characterized in that The method comprises the following steps: S1, using a radar to transmit millimeter-wave signals to detect a moving target; S2, judging whether a target exists, and the specific steps are as follows: S21, performing two-dimensional Fourier transform and square-law detection on the echo signals received by the radar to obtain a range-Doppler matrix A; S22. Select the target a to be detected ij , a ij represents the value of the i-th row and j-th column of the range-Doppler matrix A. Determine the sizes of the surrounding guard cells and reference cells. Due to the special nature of the elements at the edges, it is necessary to perform edge extension on the range-Doppler matrix A according to the circular symmetry of the spectrum so that the values at the matrix edges also meet the conditions for CFAR operation; S23, calculating a prefix-sum matrix SUM of the range-Doppler matrix A according to the prefix-sum principle; S24. Calculate the target a to be detected through the prefix sum matrix SUM obtained in step S23 i,j The data r of the surrounding reference unit R i,j Sum ∑ i,j∈R r i,j , and obtain their arithmetic mean where K is the number of reference units K; S25. Multiply the arithmetic mean value Z calculated in step S24 k by the threshold product factor α to calculate the estimated value of the detection threshold which is the estimated value of the background clutter; S26, compare the target a to be detected i,j with the threshold estimate value T. If the value of the target a to be detected i,j is greater than the threshold estimate value T, it is determined that a target exists; if the value of the target a to be detected ij is less than the threshold estimate value T, it is determined that no target exists; S27, increasing the values of i and j one by one, and repeating steps S22 - S26. After detecting all the elements in the range-Doppler matrix A, the target detection ends.

2. The method for radar CFAR target detection based on the prefix sum algorithm according to claim 1, wherein, In step S23, the calculation formula of the prefix-sum matrix SUM is: sum i,j = sum i-1,j + sum i,j-1 - sum i-1,j-1 + a i,j where sum i,j and a i,j are the values of the i-th row and j-th column of the prefix sum matrix SUM and the range-Doppler matrix A, respectively.

3. A radar CFAR target detection method based on the prefix sum algorithm according to claim 1, characterized in that, In step S24, the sum of the data of the reference units around the target a to be detected is calculated using the prefix sum matrix SUM i,j The difference between matrix A(i-M:i+M,j-N:j+N) and matrix A(i-m:i+m,j-n:j+n) is the sum of the data of the reference units around A(i - M:i + M, j - N:j + N) - A(i - m:i + m, j - n:j + n) = sum i+M,j+N - sum i-M-1,j+N - sum i+M,j-N-1 + sum i-M-1,j-N-1 -(sum i+m,j+n - sum i-m-1,j+n - sum i+m,j-n-1 + sum i-m-1,j-n-1 ) where M and N are the lengths and widths of the reference cells respectively, and m and n are the lengths and widths of the guard cells respectively.

Images