Implementation method of improved two-dimensional ordered statistical constant false alarm rate detector based on CPU

Abstract

The invention discloses an improved two-dimensional ordered statistical constant false alarm rate detector implementation method based on a CPU. The implementation steps are as follows: inputting a two-dimensional data matrix; performing initialization setting; for each detection unit, judging whether the sample value of the detection unit is the maximum value in a nine-block grid taking the detection unit as the center or not before detection; if yes, judging whether the sample value of the detection unit is greater than k arbitrary values of all reference unit sample values multiplied by a threshold factor or not through parallel processing; if yes, the detection unit is judged as a target, and the adjacent undetected unit does not need to be detected subsequently. According to the method, the number of detection units is greatly reduced, the processing time is shortened, meanwhile, in the detection process, a CPU does not need to open up an extra memory space to store all reference units to be sorted, data sorting work which is complex in traditional two-dimensional ordered statistics constant false alarm rate and long in consumed time is ingeniously avoided, the time complexity and the space complexity of an algorithm are reduced, and the detection efficiency is improved. And the real-time performance of target detection is ensured.

Description

Technical Field

[0001] The invention belongs to the technical field of radar signal processing, and in particular relates to a method for realizing an improved two-dimensional ordered statistical constant false alarm rate detector based on a CPU. Background Art

[0002] The traditional two-dimensional ordered statistical constant false alarm rate detection method mainly performs sorting operations to find the kth value after sorting. On the one hand, although it improves the detection performance of the radar system, on the other hand, the amount of calculation is greatly increased, which brings great pressure to the real-time requirements of the radar system. Let N be the total number of all reference units in the upper, lower, left and right windows. Even if quick sorting is used, its time complexity reaches O(NlogN). Since the original order information of the data will be lost after internal sorting, it cannot meet the requirements of data stream processing. Therefore, the internal sorting method cannot be used. The data must be copied and the original position of the sampling point must be remembered before sorting. Therefore, even if the internal sorting method such as the bubble method is used, the space complexity of the algorithm is still R+O(1). The new technical method should solve the problems of large amount of sorting operations, long time consumption, and low real-time performance, and reduce the time complexity and space complexity of two-dimensional ordered statistical constant false alarm rate detection as much as possible to reduce the detection time consumption, thereby meeting the real-time requirements of the radar system. Summary of the invention

[0003] The purpose of the present invention is to provide a method for realizing an improved two-dimensional ordered statistical constant false alarm rate detector based on a CPU.

[0004] The solution to achieve the purpose of the present invention is: a method for implementing an improved two-dimensional ordered statistical constant false alarm rate detector based on a CPU, the method comprising:

[0005] Step 1: Input a two-dimensional range-Doppler matrix of size M*N, where M is the number of rows of the matrix, representing the number of range units in the range dimension, and N is the number of columns of the matrix, representing the number of Doppler dimension units in the Doppler dimension. Each element of the matrix represents each detection unit.

[0006] Step 2: Initialize the settings. The reference window of the detector is selected as a cross window. The number of reference units of the upper, lower, left, and right windows in the range dimension and Doppler dimension is set to R, the number of protection units of the upper, lower, left, and right windows is set to P, the threshold factor is T, and the sequence value is k.

[0007] Step 3: Detection begins. The two-dimensional data matrix is ​​continuously input from left to right in rows, that is, starting from the first range unit number, the corresponding Doppler unit test data 1 to N are input in sequence until the last Doppler unit data in the last M-th range unit number is input, and the detection ends. When each detection unit is determined to be a target or clutter, the next data is input, and the detection of the next detection unit begins;

[0008] Step 4: Determine whether all the sample values of the detected unit samples within the 3*3 matrix centered on the current detection unit are the maximum values in the 3*3 matrix centered on themselves. If not, execute Step 3. Otherwise, the current detection unit is directly determined to be clutter, and Step 2 is re-executed for the next detection unit;

[0009] Step 5: Determine whether the sample value of the current detection unit is zero. If not, execute Step 4. Otherwise, the current detection unit is directly determined to be clutter, and Step 2 is re-executed for the next detection unit;

[0010] Step 6: Determine whether the sample value of the current detection unit is the maximum value in the 3*3 matrix centered on itself. If so, execute Step 5. Otherwise, the current detection unit is directly determined to be clutter, and Step 2 is re-executed for the next detection unit;

[0011] Step 7: Multiply the sample values of all the reference units x i (i = 1, 2,..., 4R) within the upper, lower, left, and right windows of the current detection unit by the threshold factor T;

[0012] Step 8: Determine whether the sample value of the current detection unit is greater than any k of the product results in Step 7. If it is greater, the current detection unit is determined to be a target, and Step 2 is re-executed for the next detection unit. Otherwise, it is determined to be clutter, and Step 2 is re-executed for the next detection unit;

[0013] Step 9: For the detection units determined to be targets in Step 6, output the distance and speed of the detection unit, and calculate information such as the azimuth angle and elevation angle of the target through sum-difference processing.

[0014] Compared with the prior art, the significant advantages of the present invention are: 1) The present invention greatly reduces the number of detection units and shortens the processing time; 2) The present invention cleverly avoids the complex and time-consuming data sorting work in the traditional two-dimensional ordered statistic constant false alarm rate detection process, reduces the time complexity and space complexity, and ensures the real-time performance of target detection. Brief Description of the Drawings

[0015] Figure 1 It is a flowchart of an implementation method of an improved two-dimensional ordered statistic constant false alarm rate detector based on CPU of the present invention.

[0016] Figure 2 It is a schematic diagram of a cross reference window.

[0017] Figure 3 It is a schematic diagram of screening the sample of the current detection unit.

[0018] Figure 4 It is a flowchart of an improved two-dimensional ordered statistic constant false alarm rate algorithm processing. Detailed implementation mode

[0019] The present invention will be further described below in conjunction with the accompanying drawings.

[0020] An implementation method of an improved two-dimensional ordered statistic constant false alarm rate detector based on a CPU according to the present invention is as Figure 1 shown, and the method includes:

[0021] Step 1: Input a two-dimensional range-Doppler matrix of size M*N, where M is the number of rows of the matrix, representing the range cells in the range dimension, and N is the number of columns of the matrix, representing the Doppler cells in the Doppler dimension. Each element of the matrix represents each detection cell.

[0022] Step 2: Initialize the settings. The reference window of the detector is selected as a cross window as Figure 2 shown. The purple part is the reference cell, and the gray part is the guard cell. Set the number of reference cells in the upper, lower, left, and right windows of the range dimension and the Doppler dimension to be R each, the number of guard cells in the upper, lower, left, and right windows to be P each, the threshold factor to be T, and the order value to be k;

[0023] Step 3: Start the detection. The two-dimensional data matrix is continuously input row by row from left to right, that is, the data to be measured for the corresponding Doppler cells 1 to N are input sequentially starting from the first range cell number until the data of the last Doppler cell in the last Mth range cell number is input, and the detection ends. When each detection cell is determined to be a target or clutter, the next data is input, that is, the detection of the next detection cell starts;

[0024] Step 4: As Figure 3 shown, judge whether the detected cells 1, 2, 3, 8 adjacent to the current detection cell x i in space are the maximum values in the 3*3 matrix centered on themselves. If not, execute Step 3; otherwise, the current detection cell is directly determined to be clutter, and Step 2 is re-executed for the next detection cell;

[0025] Step 5: Judge whether the sample value of the current detection cell is zero. If not, execute Step 4; otherwise, the current detection cell is directly determined to be clutter, and Step 2 is re-executed for the next detection cell;

[0026] Step 6: Judge whether the sample value of the current detection cell is the maximum value in the 3*3 matrix centered on itself. If so, execute Step 5, and there is no need to detect the undetected cells 4, 5, 6, 7 as Figure 3 shown subsequently; otherwise, the current detection cell is directly determined to be clutter, and Step 2 is re-executed for the next detection cell;

[0027] Step 7: All the reference cells x within the upper, lower, left, and right windows of the current detection celli Multiply the sample values of (i = 1, 2, …, 4R) by the threshold factor T;

[0028] Step 8: Determine whether the sample value of the current detection unit is greater than any k of all the product results in Step 7. If it is greater, the current detection unit is determined to be a target, and Step 2 is re-executed for the next detection unit. Otherwise, it is determined to be clutter, and Step 2 is re-executed for the next detection unit;

[0029] Step 9: For the detection units determined to be targets in Step 6, output the distance and velocity of the detection unit, and calculate information such as the azimuth angle and elevation angle of the target through sum-difference processing.

[0030] The present invention will be further described below in conjunction with embodiments.

[0031] Step 1: Input a two-dimensional range-Doppler matrix of size M*N, where M = 256 is the number of rows of the matrix, representing the range cells in the range dimension, and N = 1024 is the number of columns of the matrix, representing the Doppler cells in the Doppler dimension. Each element of the matrix represents each detection unit.

[0032] Step 2: Initialize the settings. The reference window of the detector is selected as a cross window. Set the number of reference cells for the upper, lower, left, and right windows in the range and Doppler dimensions to be R = 8 each, the number of guard cells for the upper, lower, left, and right windows to be P = 2 each, the threshold factor to be T = 6, and the order value to be k = 24;

[0033] Step 3: Start the detection. Input the two-dimensional data matrix row by row from left to right, that is, input the data to be measured for the corresponding Doppler cells 1 to N starting from the first range cell number until the last Doppler cell data in the last Mth range cell number is input, and the detection ends. When each detection unit is determined to be a target or clutter, input the next data, that is, start detecting the next detection unit;

[0034] Step 4: Determine whether all the sample values of the detected units within the 3*3 matrix centered on the current detection unit are the maximum values in the 3*3 matrix centered on themselves. If not, execute Step 3. Otherwise, the current detection unit is directly determined to be clutter, and Step 2 is re-executed for the next detection unit;

[0035] Step 5: Determine whether the sample value of the current detection unit is zero. If not, execute Step 4. Otherwise, the current detection unit is directly determined to be clutter, and Step 2 is re-executed for the next detection unit;

[0036] Step 6: Determine whether the sample value of the current detection unit is the maximum value in the 3*3 matrix centered on itself. If so, execute Step 5; otherwise, the current detection unit is directly determined as clutter, and Step 2 is re-executed for the next detection unit.

[0037] Step 7: Multiply the sample values of all reference units x i (i = 1, 2,..., 32) within the upper, lower, left, and right windows of the current detection unit by the threshold factor.

[0038] Step 8: Determine whether the sample value of the current detection unit is greater than any k of the product results in Step 7. If it is greater, the current detection unit is determined as a target, and Step 2 is re-executed for the next detection unit; otherwise, it is determined as clutter, and Step 2 is re-executed for the next detection unit.

[0039] Step 9: For the detection units determined as targets in Step 6, output the distance and speed of the detection units, and calculate information such as the azimuth angle and elevation angle of the target through sum and difference processing.

[0040] For a two-dimensional range-Doppler data matrix with a data size of 256*1024, the comparison of the execution times for processing all detection units using the improved and unimproved two-dimensional ordered statistic constant false alarm rate detection is shown in the following table.

[0041] Before improvement After improvement The 1st test / ms 14.07 5.22 The 2nd test / ms 14.52 5.45 The 3rd test / ms 13.16 5.78 The 4th test / ms 11.39 5.89 The 5th test / ms 12.73 5.31 Average time consumption / ms 13.174 5.53

[0042] As can be seen from the table, the time consumption of the improved two-dimensional ordered statistic constant false alarm rate detection is significantly reduced, indicating that the improved method is practical and feasible.

[0043] In the embodiment, the implementation method of the improved two-dimensional ordered statistic constant false alarm rate detector based on the CPU greatly reduces the number of detection units, shortens the processing time, skillfully avoids the complex and time-consuming data sorting work in the traditional two-dimensional ordered statistic constant false alarm rate detection process, reduces the time complexity and space complexity of the algorithm, ensures the real-time performance of target detection, and has certain engineering application value.

Claims

1. An implementation method of an improved two-dimensional ordered statistic constant false alarm rate detector based on CPU, characterized in that, The method includes: Step 1: Input a two-dimensional range-Doppler matrix of size M*N, where M is the number of rows of the matrix, representing the number of range cells in the range dimension, and N is the number of columns of the matrix, representing the number of Doppler cells in the Doppler dimension. Each element of the matrix represents each detection unit. Step 2: Initialize the settings. The reference window of the detector is selected as a cross window. Set the number of reference cells for the upper, lower, left, and right windows in both the range and Doppler dimensions to R, the number of guard cells for the upper, lower, left, and right windows to P, the threshold factor to T, and the order value to k. Step 3: Start the detection. Input the two-dimensional data matrix row by row from left to right, that is, input the data to be measured for the corresponding Doppler cells 1 to N starting from the first range cell number until the last Doppler cell data in the last Mth range cell number is input, and the detection ends. When each detection unit is determined to be a target or clutter, input the next data, that is, start detecting the next detection unit. Step 4: Determine whether all the sample values of the detected units within the 3*3 matrix centered on the current detection unit are the maximum values in the 3*3 matrix centered on themselves. If not, execute Step 3. Otherwise, the current detection unit is directly determined to be clutter, and Step 2 is re-executed for the next detection unit. Step 5: Determine whether the sample value of the current detection unit is zero. If not, execute Step 4. Otherwise, the current detection unit is directly determined to be clutter, and Step 2 is re-executed for the next detection unit. Step 6: Determine whether the sample value of the current detection unit is the maximum value in the 3*3 matrix centered on itself. If so, execute Step 5. Otherwise, the current detection unit is directly determined to be clutter, and Step 2 is re-executed for the next detection unit. Step 7: Multiply the sample values of all reference cells x i i (i = 1, 2,..., 4R) within the upper, lower, left, and right windows of the current detection unit by the threshold factor T; Step 8: Determine whether the sample value of the current detection unit is greater than any k of the product results in Step 7. If it is greater, the current detection unit is determined to be a target, and Step 2 is re-executed for the next detection unit. Otherwise, it is determined to be clutter, and Step 2 is re-executed for the next detection unit. Step 9: For the detection units determined to be targets in Step 6, output the range and velocity of the detection unit, and calculate information such as the azimuth angle and elevation angle of the target through sum-difference processing.

2. The implementation method of an improved two-dimensional ordered statistic constant false alarm rate detector based on CPU according to claim 1, characterized in that, The method for screening detection units in Step 4 specifically includes: First, determine whether the sample value is the maximum value in the 3*3 matrix centered on the current detection unit. If it is the maximum value, perform subsequent two-dimensional ordered statistical constant false alarm rate detection on the current detection unit and mark it. When sliding the window again, the remaining undetected units within the 3*3 matrix centered on the current detection unit will no longer be the maximum values in their respective nine-square grids. Therefore, these detection units are unlikely to have targets, so subsequent two-dimensional ordered statistical constant false alarm rate detection will not be performed on them, and they are directly determined to be clutter.

3. The implementation method of an improved two-dimensional ordered statistic constant false alarm rate detector based on CPU according to claim 1, characterized in that, The method for selecting the reference window and reference units in Step 7 specifically includes: The detector starts detecting from the first detection unit at the upper left corner of the two-dimensional data matrix. At this time, the data of the two windows, the lower window and the right window, are used as reference units. Similarly, when the detection unit is located at the corner position of the range-Doppler matrix, two or three effective reference windows are selected as the reference background. When the detection unit is located in the middle of the range-Doppler matrix, that is, when the data in all four reference windows are valid, these four windows are used as the reference background.

4. An implementation method of an improved two-dimensional ordered statistic constant false alarm rate detector based on a CPU according to claim 1, characterized in that, The improved method for reducing the large computational workload of the sorting operation in the traditional two-dimensional ordered-statistic constant false alarm rate (OS-CFAR) detection in step 8 specifically includes: This method is based on the parallel processing mode of a multi-core CPU. First, the sample values X(n) of all reference cells in the upper, lower, left, and right windows of the current detection cell are taken out simultaneously i (i = 1, 2,..., 4R). These values are multiplied by the threshold factor T, and then multi-threaded parallel programming is used for processing. The sample value D of the current detection cell is compared one by one with the values X(n)*T obtained by multiplying the sample values of all reference cells by the threshold factor. If D is greater than X(n)*T, the count is incremented by 1. Finally, it is judged whether the count result is greater than k. If it is greater than k, it means that the current detection cell is the target; otherwise, it is clutter.

5. An implementation method of an improved two-dimensional ordered statistic constant false alarm rate detector based on a CPU according to claim 1, characterized in that, The method for outputting information such as the target range, velocity, azimuth angle, and elevation angle in step 9 specifically includes: After obtaining the range bin and velocity bin where the target in the sum channel is located through two-dimensional ordered-statistic constant false alarm rate detection, the target range and velocity are calculated according to the lengths of the range bin and velocity bin designed in the radar system. Combining with the corresponding positions in the range-Doppler two-dimensional data matrix of the azimuth difference and elevation difference channels, sum-difference processing is performed to provide operation data for the sum-difference amplitude comparison angle measurement work in the subsequent radar data processing system.

Images