A multi-target detection method and system based on millimeter wave radar

Abstract

The application discloses a kind of multi-target detection method and system based on millimeter wave radar, comprising: obtaining the original echo data collected by millimeter wave radar, the original echo data includes the target data point of multiple targets;Select a preset number of data points from the target data point of the multiple targets, and the selected data point is used as effective data point;Based on the effective data point, the multiple targets are detected.The application uses non-strict threshold screening constant false alarm rate algorithm combined with double index determination method and spatial distance judgment algorithm, realizes multi-target detection while reducing the influence of multipath effect on result data, improves the accuracy of target point detection, greatly reduces the output data volume of frequency domain data processing, reduces the operation amount of logic processing, which is beneficial to the implementation of the algorithm on edge mobile device.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of radar signal processing, in particular to a multi-target detection method and system based on millimeter wave radar. BACKGROUND

[0002] With the rapid growth of the automatic driving and intelligent transportation industry, there is a strong demand for high precision and real-time environmental perception capability, especially for terminal devices such as unmanned vehicles and unmanned aerial vehicles. Millimeter wave radar has been widely used in these fields due to its excellent stability and reliable detection performance. However, millimeter wave radar algorithms face problems such as high computational complexity, strict real-time requirements, and traditional algorithms that are not friendly to hardware configurations in data processing.

[0003] Patent CN113253228B discloses a multi-size target detection method based on radar condensation algorithm, including determining a radar condensation threshold; reserving a space variable; if there is a target, then the points in each frame under the angle of the target that exceed the detection threshold are formed into a detection point group; the number of target points that have not been output, the average angle of the target points, and the maximum amplitude of the target points in the distance unit where the detection point group is located are counted and numbered; the output variable is analyzed and the condensation target point is output; if there is a new frame of detection data reported, then return to detection until all frames of scanning are completed.

[0004] The above patents all have the problems mentioned in the background art: traditional radar algorithms do not fully adapt to the hardware characteristics of FPGA, and due to the high computational complexity of traditional radar algorithms, a large amount of computing resources are required, which is not conducive to implementation on edge mobile devices; to solve the above problems, the present application proposes a multi-target detection method and system based on millimeter wave radar. SUMMARY

[0005] In view of the deficiencies of the prior art, the main purpose of the present application is to provide a multi-target detection method and system based on millimeter wave radar, which can effectively solve the problems in the background art. The specific technical solution of the present application is as follows:

[0006] A multi-target detection method based on millimeter wave radar, comprising:

[0007] Obtaining raw echo data collected by a millimeter wave radar, the raw echo data including target data points of multiple targets;

[0008] Selecting a preset number of data points from the target data points of the multiple targets, and taking the selected data points as valid data points;

[0009] Detecting the multiple targets based on the valid data points.

[0010] Specifically, the selecting a preset number of data points from the target data points of the plurality of targets and taking the selected data points as valid data points comprises:

[0011] The constant false alarm rate algorithm using non-strict threshold screening selects a preset number of data points from the target data points of the plurality of targets to obtain first data points;

[0012] The first data points are screened by using a double-index determination method to obtain second data points;

[0013] The second data points are screened by using a spatial distance judgment algorithm to obtain valid data points.

[0014] Specifically, the constant false alarm rate algorithm using non-strict threshold screening selects a preset number of data points from the target data points of the plurality of targets to obtain first data points, comprising:

[0015] The target data points of the plurality of targets are subjected to FIR operation to obtain environmental interference data around each target data point;

[0016] The environmental interference data is removed from the target data point data to obtain de-interference data;

[0017] The de-interference data is sorted and a preset number of data point indexes are selected from large to small to obtain first data points.

[0018] Specifically, the removing the environmental interference data from the target data point data to obtain de-interference data comprises:

[0019] The target data point data is subjected to fast Fourier transform of a preset data amount, and a module is obtained to obtain a first module result;

[0020] The environmental interference data is subjected to module operation to obtain a second module result;

[0021] The first module result is subtracted from the second module result to obtain de-interference data.

[0022] Specifically, the screening the first data points by using a double-index determination method to obtain second data points comprises:

[0023] According to the multi-dimensional information points of the first data points, data points that exist in distance dimension and speed dimension information points at the same time in the first data points are screened out as second data points.

[0024] Specifically, the screening the second data points by using a spatial distance judgment algorithm to obtain valid data points comprises:

[0025] By comparing the size between the distance dimension indexes and the size between the speed dimension indexes of the second data points, a plurality of target points pointing to the same to-be-detected physical target are obtained, wherein the plurality of target points pointing to the same to-be-detected physical target are points close in the distance dimension index and the speed dimension index.

[0026] Among the plurality of target points, a point with a larger value in the angle dimension fast Fourier result data is selected as an effective data point.

[0027] Specifically, the detection of the plurality of targets based on the effective data point comprises:

[0028] According to the effective data point, the information of each target is calculated from multiple dimensions to obtain dynamic information of the plurality of targets, and the multi-target detection based on the millimeter wave radar data is realized.

[0029] A system for realizing the multi-target detection method based on the millimeter wave radar comprises:

[0030] The millimeter wave radar, the central processing unit, the acceleration core, the DDR, the FPGA and the display terminal.

[0031] The millimeter wave radar is used for collecting echo data.

[0032] The central processing unit is used for realizing the small-point operation and the logical judgment required by the multi-target detection method.

[0033] The acceleration core is used for realizing the digital signal processing operation required by the multi-target detection method.

[0034] The DDR uses double data transmission to improve the data transmission rate.

[0035] The FPGA is used for integrating the central processing unit, the acceleration core and the DDR.

[0036] The display terminal is used for displaying the calculation result of the multi-target detection method.

[0037] Specifically, the acceleration core is used for realizing the digital signal processing operation required by the multi-target detection method, comprising:

[0038] The acceleration core realizes the digital signal processing operation required by the multi-target detection method through one-dimensional sequence FFT operation, one-dimensional sequence FIR operation, one-dimensional sequence point multiplication operation, one-dimensional sequence modulus operation and one-dimensional sequence addition and subtraction operation.

[0039] Compared with the prior art, the present application has the following beneficial effects:

[0040] The application selects a certain number of effective data points as algorithm output according to millimeter wave radar collection configuration selection by using a non-strict threshold screening constant false alarm rate algorithm, so as to greatly reduce the output data amount of frequency domain data processing, reduce the operation amount of subsequent logic processing, and the certain number of output data amounts can reduce the complexity of algorithm task deployment, which is beneficial to the implementation of the algorithm on the edge end mobile device; the combination of the different dimension non-strict threshold screening constant false alarm rate algorithm double index determination method and the spatial distance algorithm only compared with the dimension index can reduce the influence of multipath effect on the result data; the FPGA hardware with a DSP acceleration core and a CPU is selected on the system, and the batch digital signal calculation is operated by using the DSP acceleration core in the task deployment, and other operations such as logic judgment are operated by the CPU, so that the calculation mode has higher expansibility and flexibility, and supports the improvement of the algorithm without changing the hardware platform in subsequent practice. BRIEF DESCRIPTION OF DRAWINGS

[0041] Figure 1 A work flow diagram of a multi-target detection method based on a millimeter wave radar of the application;

[0042] Figure 2 An algorithm task flow deployment schematic diagram of embodiment 1 of the application;

[0043] Figure 3 A constant false alarm rate algorithm schematic diagram in embodiment 1 of the application;

[0044] Figure 4 A non-strict threshold screening constant false alarm rate algorithm schematic diagram in embodiment 1 of the application;

[0045] Figure 5 A double index determination method schematic diagram in embodiment 1 of the application;

[0046] Figure 6 A spatial distance judgment algorithm schematic diagram in embodiment 1 of the application;

[0047] Figure 7 A hardware platform architecture diagram in embodiment 2 of the application. DETAILED DESCRIPTION

[0048] In order to make the above-mentioned purposes, features and advantages of the application more obvious and easy to understand, the specific embodiments of the application will be described in detail below with reference to the accompanying drawings.

[0049] In the following description, many specific details are set forth in order to provide a thorough understanding of the application, but the application can also be implemented in other ways different from those described herein, and those skilled in the art can make similar generalizations without departing from the connotation of the application, therefore the application is not limited by the specific embodiments disclosed below.

[0050] Third, the "one embodiment" or "an embodiment" referred to herein means a specific feature, structure, characteristic, or combination of features and / or characteristics described herein that can be included in at least one implementation of the present application. The various appearances of "in one embodiment" or "an embodiment" in the specification do not all refer to the same embodiment, although they can.

[0051] Embodiment 1

[0052] The embodiment provides a multi-target detection method based on millimeter wave radar, and a specific scheme is as follows: Figure 1 As shown in the figure, the multi-target detection method based on millimeter wave radar comprises the following steps:

[0053] S101, acquiring original echo data collected by a millimeter wave radar, wherein the original echo data comprises target data points of multiple targets;

[0054] S102, selecting a preset number of data points from the target data points of the multiple targets, and taking the selected data points as effective data points;

[0055] S103, detecting the multiple targets based on the effective data points.

[0056] In the embodiment, after performing three-dimensional fast Fourier transform on the echo data collected by the millimeter wave radar, a hardware-friendly non-strict threshold screening constant false alarm rate algorithm is used to realize multi-target data detection, a double-index determination method is used to further screen data points, and a spatial distance judgment algorithm is used to acquire target point data information. Figure 2 Because the constant false alarm rate algorithm with strict threshold screening will screen different amounts of target data information due to different radar collected data, the hardware-friendly non-strict threshold screening constant false alarm rate algorithm is used to obtain a certain amount of multi-target data information, and then the spatial distance judgment algorithm is used to further screen target data information.

[0057] Specifically, the time domain signal is converted into a frequency domain signal by performing multi-dimensional fast Fourier transform on the original echo data, the multi-dimensional fast Fourier transform includes a distance fast Fourier transform, a speed fast Fourier transform and an angle fast Fourier transform, and the fast Fourier transform can improve the signal processing speed and separate information of different dimensions; the frequency domain information data is processed by using a constant false alarm rate algorithm for non-strict threshold screening, the detection threshold can be adaptively set, the false alarm rate is kept constant, the detection probability is maximized, different background noise environments can be adapted to, the detection sensitivity is improved, and the false alarm rate is reduced; the first effective information point is screened by using a double index determination method, the distance dimension information and the speed dimension information are used for screening, the information of multiple dimensions is combined to improve the detection accuracy, the false alarm can be further reduced, and the reliability of target detection is improved; target point data information is obtained by using a spatial distance judgment algorithm, multiple targets are separated, the remaining data point information is further screened, and the accuracy of target recognition is improved; and dynamic information of multiple to-be-detected physical targets is calculated according to the target point data information, the dynamic parameters such as the position, speed and acceleration of the target are calculated based on the target point information obtained in the foregoing steps, multi-target tracking is realized, and finally the detection of multiple targets and the acquisition of dynamic information are realized.

[0058] The application uses a constant false alarm rate algorithm for non-strict threshold screening, selects a certain number of effective data points as algorithm outputs according to millimeter wave radar acquisition configuration, greatly reduces the output data amount of frequency domain data processing, reduces the operation amount of subsequent logic processing, and the certain number of output data amounts can reduce the complexity of algorithm task deployment, which is beneficial to the implementation of the algorithm on an edge mobile device; the combination of the double index determination method of the constant false alarm rate algorithm for non-strict threshold screening of different dimensions and the spatial distance algorithm for comparison of only dimension indexes can reduce the influence of multipath effect on result data; the FPGA hardware with a DSP acceleration core and a CPU is selected on the system, batch digital signal calculation is operated by using the DSP acceleration core when segmentation is performed during task deployment, and other operations such as logic judgment are operated by the CPU, the calculation mode has higher expansibility and flexibility, and supports improvement of the algorithm without changing the hardware platform in subsequent practice.

[0059] Further, the pre-set number of data points are selected from the target data points of the multiple targets, and the selected data points are used as effective data points, and the method comprises the following steps.

[0060] S201, a constant false alarm rate algorithm for non-strict threshold screening is used to select a pre-set number of data points from the target data points of the multiple targets, and first data points are obtained;

[0061] S202, a double index determination method is used to screen the first data points, and second data points are obtained.

[0062] S203, filtering the second data points by using a spatial distance judgment algorithm to obtain effective data points.

[0063] In the embodiment, the frequency domain information data is processed by using the non-strict threshold filtering constant false alarm rate algorithm, the detection threshold can be adaptively set, the false alarm rate is kept constant, the detection probability is maximized, different background noise environments can be adapted, the detection sensitivity is improved, and the false alarm rate is reduced; the first data points are filtered by using the double index determination method, at least two dimensions of information are used for filtering, the detection accuracy can be improved by combining multiple dimensions of information, the false alarm can be further reduced, and the reliability of target detection is improved; target point data information is obtained by using the spatial distance judgment algorithm, multiple targets are separated, the remaining data point information is further filtered, and the accuracy of target recognition is improved; dynamic information of multiple to-be-detected physical targets is calculated according to the target point data information, the position, speed, acceleration and other dynamic parameters of the target are calculated based on the target point information obtained in the previous steps, information calculation of multiple targets is realized, and finally the detection of multiple targets and the acquisition of dynamic information are realized.

[0064] Further, the non-strict threshold filtering constant false alarm rate algorithm selects a preset number of data points from the target data points of the multiple targets to obtain first data points, including:

[0065] S301, performing FIR operation on the target data points of the multiple targets to obtain environmental interference data around each target data point;

[0066] S302, removing the environmental interference data from the target data point data to obtain de-interference data;

[0067] S303, sorting the de-interference data, and selecting a preset number of data point indexes from large to small to obtain first data points.

[0068] In the embodiment, the background noise of the signal is estimated by FIR operation, the data within a certain range around each data point is analyzed, the environmental interference data around each data point is obtained, the frequency domain information data is subtracted by the environmental interference data, the estimated environmental interference can be removed, and a more pure signal can be extracted, the influence of noise on the target signal is eliminated, which is helpful to improve the contrast of signal and noise and enhance the detectability of effective signal; the de-interference data after processing is sorted from large to small to facilitate subsequent selection of effective information points, the first N (fixed number) largest values are selected from the sorted de-interference data, wherein N can be set by the user according to the calculation requirement to obtain the first effective information point, by selecting the first N data points, it can be ensured that the selected signal is the most significant, thereby improving the reliability of target detection.

[0069] Specifically, the non-strict threshold screening constant false alarm rate algorithm uses the FIR operation of a fixed number of FIR kernel data to calculate the threshold information in the traditional constant false alarm rate algorithm, and obtains comparison information by subtracting the source data of the input constant false alarm rate from the calculated threshold information, and then obtains the output result of the non-strict threshold screening constant false alarm rate algorithm by sorting and screening a predetermined number of data points of the subtraction operation result.

[0070] As Figure 3 , the constant false alarm rate (CFAR) algorithm schematic diagram, the most common CFAR algorithm principle schematic diagram in radar data processing, whether a data point is a valid information point is determined by comparing the value of the point with the data threshold obtained by operating the values of the surrounding reference cells. Obviously, when the millimeter wave radar works in different environments or faces different detection targets, the algorithm will generate different amounts of data results; and in general, the amount of result data generated by the algorithm is very large, close to the amount of original data collected by the radar, and the present application improves the traditional CFAR algorithm in view of this feature.

[0071] As Figure 4 , the non-strict threshold screening constant false alarm rate algorithm of the present application, the algorithm improves the traditional CFAR algorithm, and fully utilizes the advantages of the hardware platform in the algorithm deployment stage to realize a hardware-friendly algorithm. The algorithm uses the determined FIR operation kernel data obtained according to the collection configuration of the millimeter wave radar, mainly the size of Samples and Chirps, calls the DSP acceleration core to perform FIR operation to obtain the environmental interference data around each data point. On this basis, the algorithm performs batch subtraction on the source data points and the corresponding environmental interference data, and sorts the result data from large to small, and only selects a fixed number of data point indexes as the result data of the algorithm. Since the target data points finally screened by the radar algorithm only account for a very small part of the original data amount, and the data points with larger data values also mean that the echo signal is more intense and more likely to be the desired target point, therefore under such premise, this non-strict threshold screening algorithm will not cause obvious data error.

[0072] Further, the environmental interference data is removed from the target data point data to obtain the interference-removed data, comprising:

[0073] S401, performing fast Fourier transform on the target data point data by a predetermined data amount, and taking the modulus to obtain a first modulus result;

[0074] S402, taking the modulus of the environmental interference data to obtain a second modulus result;

[0075] S403, subtract the first modulus result from the second modulus result to obtain the interference-removed data.

[0076] In the embodiment, the frequency distribution characteristics of the signal can be observed by performing fast Fourier transform on the target data point data by a preset data amount and obtaining a first modulus result, which is suitable for processing a large amount of data; the second modulus result is obtained by performing modulus operation on the environmental interference data, the frequency spectrum amplitude of the interference signal is obtained by performing modulus operation on the interference signal, and the frequency characteristics of the interference are obtained; the interference-removed data is obtained by subtracting the second modulus result from the first modulus result, which is equivalent to filtering out the interference component in the frequency domain, and the background interference in the environment can be effectively removed and the main characteristics of the target signal can be retained.

[0077] Further, the first data points are filtered by using the double-index determination method to obtain second data points, including:

[0078] According to the multi-dimensional information points of the first data points, the data points that exist in the distance dimension and the speed dimension information points at the same time are filtered out from the first data points as second data points.

[0079] In the embodiment, the double-index determination method filters the constant false alarm rate algorithm result according to different non-strict threshold values, compares and judges whether the data points are considered as valid data points in the results of the two dimensions, so as to increase the credibility of the data points. The non-strict threshold constant false alarm rate algorithm is operated on the distance dimension and the speed dimension respectively and obtains two groups of data information. By comparing the corresponding indexes of the two groups of data points in the millimeter wave radar echo data tensor, the double-index determination reduces the multipath effect to a certain extent. After the operation in this stage, the design further filters the remaining data point information by using the spatial distance algorithm, compares the indexes of the data points and the credible target points in the data tensor in the distance dimension and the speed dimension, excludes the data points with indexes too close to each other to further reduce the influence of the multipath effect on the result data, and obtains the credible target points.

[0080] Specifically, for each first valid information point, the validity thereof in different dimensions (such as distance, angle, and speed) is checked, and only the information point that is considered as valid in both the distance dimension and the speed dimension is retained. Through multi-dimensional verification, the false detection rate is greatly reduced, the points that are seemingly valid in a single dimension but are actually noise or interference can be effectively filtered out, the information point is required to be valid in multiple dimensions, the detection of false targets is significantly reduced, which is helpful for separating and identifying multiple targets in a complex scene.

[0081] As Figure 5The present invention provides a dual index determination method for a constant false alarm rate (CFAR) algorithm using a non-strict threshold screening method. This algorithm performs a non-strict threshold screening CFAR algorithm on both the distance dimension and the Doppler dimension of the frequency domain data tensor. If a data point appears in the results of the non-strict threshold screening CFAR algorithm in the distance dimension, it is considered a valid information point in the distance dimension. If the data point also appears in the results of the non-strict threshold screening CFAR algorithm in the Doppler dimension, it is considered a valid information point in both dimensions. This screening step can further enhance the credibility of the algorithm results and control multipath effects to a certain extent. The amount of data generated by the non-strict threshold screening CFAR algorithm is much smaller than that of the traditional CFAR algorithm, so the actual amount of computation required for this screening step is very small.

[0082] Furthermore, the method of screening the second data points using a spatial distance judgment algorithm to obtain valid data points includes:

[0083] S501: Obtain multiple target points pointing to the same physical target to be detected by comparing the sizes of the distance dimension indexes and the sizes of the speed dimension indexes of the second data points, wherein the multiple target points pointing to the same physical target to be detected are points with similar distance dimension indexes and similar speed dimension indexes;

[0084] S502: Select a point with a larger value in the fast Fourier transform result data in the angle dimension from among the multiple target points as a valid data point.

[0085] In this embodiment, the algorithm judges the spatial position index corresponding to the target information in the result of the constant false alarm rate algorithm based on the non-strict threshold, and selects the information points according to the distance size of the valid information points in the data space. By comparing the size between the distance dimension index and the size between the speed dimension index of the second valid information point, the distance and speed differences between different valid information points are calculated. If the distance and speed of multiple points are similar, it can be reasonably inferred that they may point to the same physical target to be detected, and multiple observation points of the same physical target can be effectively identified, thereby improving data utilization. Among the multiple identified target points, it is necessary to further analyze their performance in the angle dimension and select points with larger values ​​in the fast Fourier results of the angle dimension. Usually, these points represent more obvious target signals. Among the multiple target points, points with larger values ​​in the fast Fourier results data of the angle dimension are selected to obtain target point data information. The angle information can be used to further verify the target position and enhance the ability to confirm the target. By screening out points with large values, the amount of data for subsequent processing is reduced, thereby improving the processing efficiency of the system.

[0086] like Figure 6The spatial distance judgment algorithm of the application, after the double index determination method of the non-strict threshold screening constant false alarm rate algorithm, the result data generated by the method is input, the index information of the effective information point is judged in the three-dimensional space index of the corresponding data point in the echo data tensor, whether the two effective information points point to the same physical target to be detected is judged by comparing the size between the distance dimension indexes of the effective information points and the size between the speed dimension indexes, for the two effective information points pointing to the same physical target to be detected, the point with larger value in the angle fast Fourier result data is selected as the final target, so as to control the multipath effect.

[0087] Further, the method further comprises:

[0088] According to the target point data information, the dynamic information of the physical target to be detected is calculated from multiple dimensions, and the dynamic information of the multiple physical targets to be detected is obtained, so as to realize the multi-target detection based on the millimeter wave radar data.

[0089] In the embodiment, the dynamic information of each physical target to be detected is calculated by the screened target point data information, and the position information of the current target can be obtained based on the comprehensive calculation of distance, speed and time. The relative speed of the target can be obtained by using the Doppler effect and combining the speed component of the target point. After the operation of the radar principle formula, the ranging, velocity and angle information of the multiple targets is finally obtained. By comprehensively integrating the data of multiple dimensions, more comprehensive and accurate dynamic state of the target can be provided, and multiple physical targets can be detected and tracked at the same time, so as to adapt to the complex scene demand.

[0090] Embodiment 2

[0091] A system for realizing the multi-target detection method based on the millimeter wave radar, comprising:

[0092] The millimeter wave radar, the central processing unit, the acceleration core, the DDR, the FPGA and the display terminal;

[0093] The millimeter wave radar is used for collecting echo data;

[0094] The central processing unit is used for realizing the small point operation and logical judgment required by the multi-target detection method;

[0095] The acceleration core is used for realizing the digital signal processing operation required in the multi-target detection method;

[0096] The DDR uses double data transmission to improve the data transmission rate;

[0097] The FPGA is used for integrating the central processing unit, the acceleration core and the DDR;

[0098] The display terminal is used to display the calculation results of the multi-target detection method.

[0099] In this embodiment, the millimeter wave radar system transmits millimeter wave signals and receives echo data, and detects the position, speed and other characteristics of the object by measuring the reflection changes of the signal, and can achieve high-precision detection on short distances and relatively small objects; the central processing unit is responsible for executing program instructions, performing logical judgment and control system operations, processing basic small-point operations, managing data flows and control signals between various modules, coordinating the operation of the entire system, and can execute different algorithms and processing tasks to adapt to different detection requirements; the acceleration core is a computing unit designed specifically for processing digital signals, which can quickly execute complex calculations and processing tasks related to multi-target detection, has parallel processing capabilities, and can simultaneously It handles complex signal processing in multiple data streams and optimizes specific algorithms, significantly improving computing speed and shortening target detection latency. DDR memory supports double data transmission, transmitting data simultaneously on both the rising and falling edges of each clock cycle, significantly increasing data transmission rates and providing fast data storage and access capabilities, avoiding data processing bottlenecks and improving overall system performance. FPGA integrates CPU, acceleration core, and DDR. Through the parallel processing capabilities of FPGA, it can effectively execute complex signal processing algorithms, realize parallel signal processing, and shorten response time. The display terminal is used to present real-time calculation results of multi-target detection and visualize the processed data.

[0100] like Figure 7 This hardware configuration fully considers the digital signal processing requirements of the millimeter-wave radar algorithm. Digital signal processing acceleration hardware is used to accelerate the extensive DSP operations required by the millimeter-wave radar algorithm, while the CPU is configured to perform the smaller number of other operations and logical judgments required by the millimeter-wave radar algorithm. Through the coordinated operation of the millimeter-wave radar, central processing unit, acceleration core, DDR, FPGA, and display terminal, an efficient, flexible, and real-time multi-target detection platform is constructed. The meticulous coordination between these modules ensures that the system can accurately and quickly detect and track multiple targets in dynamic environments, thereby improving overall performance and application effectiveness.

[0101] Furthermore, the acceleration core is used to implement the digital signal processing operations required in the multi-target detection method, including:

[0102] The acceleration core realizes the digital signal processing operations required in the multi-target detection method through one-dimensional sequence FFT operation, one-dimensional sequence FIR operation, one-dimensional sequence point multiplication operation, one-dimensional sequence modulus operation and one-dimensional sequence addition and subtraction operation.

[0103] In this embodiment, the acceleration core implements the digital signal processing required in the multi-target detection method through various one-dimensional sequence operations, converts the time-domain signal into a frequency-domain representation through one-dimensional sequence FFT operations, implements digital filtering using convolution operations through one-dimensional sequence FIR operations, maintains the time characteristics of the signal, removes environmental interference, multiplies the corresponding elements of two vectors and then sums them through one-dimensional sequence point multiplication operations, calculates the amplitude of a complex sequence to obtain signal strength information through one-dimensional sequence modulus operations, and performs element-wise addition or subtraction operations on two sequences through one-dimensional sequence addition and subtraction operations to achieve basic signal combination and comparison. By implementing these basic one-dimensional sequence operations in the acceleration core, the system can efficiently process radar signals and achieve fast and accurate multi-target detection. This method not only improves the performance of the system but also maintains sufficient flexibility to adapt to different application scenarios and algorithm requirements.

[0104] The above shows and describes the basic principles and main features of the present application and the advantages of the present application. Those skilled in the art should understand that the present application is not limited by the above embodiments, and the above embodiments and descriptions in the specification are only to illustrate the principles of the present application. Without departing from the spirit and scope of the present application, various changes and improvements can be made to the present application, and these changes and improvements all fall within the scope of the claimed present application. The scope of protection of the present application is defined by the appended claims and their equivalents.

Claims

1. A method for multi-target detection based on millimeter wave radar, characterized in that, The method comprises the following steps: acquiring original echo data collected by a millimeter wave radar, the original echo data comprising target data points of multiple targets; performing FIR operation on the target data points of the multiple targets to obtain environmental interference data around each target data point; removing the environmental interference data from the target data point data to obtain de-interference data; sorting the de-interference data and selecting a preset number of data points from large to small to obtain first data points; screening data points in the first data points that exist in both distance dimension information points and speed dimension information points as second data points according to multi-dimensional information points of the first data points; obtaining multiple target points pointing to the same physical target to be detected by comparing the sizes of distance dimension indexes and the sizes of speed dimension indexes of the second data points, wherein the multiple target points pointing to the same physical target to be detected are points with close distance dimension indexes and speed dimension indexes; selecting points with larger values in angle dimension fast Fourier result data from the multiple target points as effective data points; detecting the multiple targets based on the effective data points.

2. The method of claim 1, wherein, The step of removing the environmental interference data from the target data point data to obtain de-interference data comprises the following steps: performing fast Fourier transform on the target data point data with a preset data amount and taking modulus to obtain first modulus result; taking modulus of the environmental interference data to obtain second modulus result; subtracting the second modulus result from the first modulus result to obtain de-interference data.

3. The method of claim 1, wherein, The step of detecting the multiple targets based on the effective data points comprises the following steps: calculating information of each target from multiple dimensions according to the effective data points to obtain dynamic information of the multiple targets, thereby realizing multi-target detection based on millimeter wave radar data.

4. A multi-target detection system based on millimeter wave radar, for implementing a multi-target detection method based on millimeter wave radar according to any one of claims 1-3, characterized in that, The system comprises a millimeter wave radar, a central processing unit, an acceleration core, a DDR, an FPGA and a display terminal. The millimeter wave radar is used to collect echo data. The central processing unit is used to realize small-point operation and logical judgment required by the multi-target detection method. The acceleration core is used to realize digital signal processing operation required by the multi-target detection method. The DDR is used to improve data transmission rate by double data transmission. The FPGA is used to integrate the central processing unit, the acceleration core and the DDR. The display terminal is used to display calculation results of the multi-target detection method. The acceleration core is used to realize digital signal processing operation required by the multi-target detection method, which comprises the following steps:

5. The multi-target detection system based on millimeter wave radar according to claim 4, characterized in that, The acceleration core realizes digital signal processing operation required by the multi-target detection method through one-dimensional sequence FFT operation, one-dimensional sequence FIR operation, one-dimensional sequence point multiplication operation, one-dimensional sequence modulus operation and one-dimensional sequence addition and subtraction operation. ​

Images