A Multipath Ghost Recognition Method for Vehicle-Mounted Millimeter-Wave Radar Based on Reflection Point Inversion Search

Abstract

This invention discloses a multipath identification method for vehicle-mounted millimeter-wave radar based on reflection point inversion search, applied in the field of vehicle-mounted millimeter-wave multipath signal processing technology. It addresses the problem of ineffective identification and suppression of multipath ghosting in existing technologies. This invention generates a radar point cloud by preprocessing the MIMO millimeter-wave radar echo; it separates the radar point cloud into static and dynamic points, then clusters the resulting dynamic point cloud, acquiring the center coordinates and multi-domain information of each cluster; it associates cluster centers within the scene through target association; it calculates the reflection point coordinates under three assumptions; it initially identifies multipath ghosting through point cloud search steps; and finally, it further determines multipath ghosting based on multi-domain information.

Description

Technical Field

[0001] This invention belongs to the field of vehicle-mounted millimeter-wave multipath signal processing technology, and specifically relates to a multipath ghosting recognition technology. Background Technology

[0002] When using vehicle-mounted millimeter-wave radar to perceive the environment, the reflection and diffraction propagation of electromagnetic signals between different objects often cause complex multipath signals. When the multipath signal and the real target are located in the same resolution cell in the range-Doppler spectrum, the echo energy of the real signal is reduced, leading to missed detections. When the multipath signal and the real target are located in different resolution cells in the range-Doppler spectrum, the multipath signal forms a ghost image, resulting in false alarms. In autonomous driving scenarios, false alarms (multipath ghosting) caused by multipath signals are more common, and the interference they cause is more severe.

[0003] In recent years, research institutions both domestically and internationally have conducted related technical research on the multipath ghosting interference problem of vehicle-mounted millimeter-wave radar. In 2020, C. Liu et al. proposed a multipath ghosting identification method based on small distance differences to address the multipath ghosting interference problem (C. Liu, S. Liu, C. Zhang, Y. Huang and H. Wang, "Multipath propagation analysis and ghost target removal for FMCW automotive radars," IET International Radar Conference (IET IRC 2020), Online Conference, 2020, pp. 330-334, doi:10.1049 / icp.2021.0554.). However, the application scenarios of this method are very limited. When the target is close to the radar, the difference between the multipath ghost and the real target relative to the radar distance is large, and the above method cannot effectively identify multipath ghosts. Furthermore, it is also easy to misidentify two real targets that are close in distance but at different angles in adjacent lanes. In 2023, Yunda Li et al. proposed a multipath ghosting identification method based on DOA / DOD estimation (Y.Li and X.Shang, "Multipath Ghost Target Identification for Automotive MIMO Radar," 2022 IEEE 96th Vehicular Technology Conference (VTC2022-Fall), London, United Kingdom, 2022, pp.1-5, doi:10.1109 / VTC2022-Fall57202.2022.10012904.). This method identifies multipath ghosting caused by electromagnetic propagation paths with different transmission and reception directions by distinguishing the differences between the antenna's transmission and reception directions. However, this method not only fails to solve multipath ghosting when the transmission and reception directions are equal, but is also severely affected by the antenna array configuration.

[0004] In summary, the methods described above all have certain drawbacks and lack universality. However, if multipath ghosting cannot be effectively identified and suppressed, the false alarms caused by multipath ghosting will severely interfere with the detection of real targets in the scene, hindering the widespread application of vehicle-mounted millimeter-wave radar. Therefore, in-depth research into more universal and effective multipath ghosting interference identification methods for vehicle-mounted millimeter-wave radar is of great value. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a method for multipath ghosting identification of vehicle-mounted millimeter-wave radar based on reflection point inversion search.

[0006] The technical solution adopted in this invention is: a method for multipath ghosting identification of vehicle-mounted millimeter-wave radar based on reflection point inversion search, comprising:

[0007] S1. Generate radar point cloud based on the ADC data of the original echo in the current frame;

[0008] S2. Separate the radar point cloud into static and dynamic point clouds, then cluster the separated dynamic point clouds, and finally extract the center coordinates of each cluster and the multi-domain information representing the center coordinates.

[0009] S3. Take any two points from the set of cluster center coordinates obtained in step S2 and associate them to form a matching set;

[0010] S4. Extract a pair of associated points from the matching set, and solve the search area based on the reflection points.

[0011] S5. Search the search area obtained in step S4 for the existence of the radar point cloud obtained in step S1. If it exists, proceed to step S6; otherwise, return to step S4.

[0012] S6. For the pair of associated points processed in step S5, multipath ghosting judgment is performed based on multi-domain information.

[0013] S7. Traverse each pair of associated points in the matching set and repeat steps S4 to S6 to complete the multipath ghosting recognition of all point clouds under the current frame data.

[0014] The beneficial effects of this invention: This invention proposes a multipath ghosting identification method for vehicle-mounted millimeter-wave radar based on reflection point inversion, which can effectively identify multipath ghosting in vehicle-mounted millimeter-wave radar echoes. Through preprocessing of MIMO millimeter-wave radar echoes, a radar point cloud is generated; through radar point cloud processing steps, dynamic and static point cloud separation, dynamic point cloud clustering, cluster center coordinates, and multi-domain information acquisition are achieved; through target association, cluster centers within the scene are associated; reflection point coordinates under three assumptions are calculated; multipath ghosting is initially identified through point cloud search steps; and finally, multi-domain information is used to further determine multipath ghosting. Experimental results show that the method of this invention can effectively identify multipath ghosting in vehicle-mounted millimeter-wave radar echoes. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of multipath signal propagation in a typical vehicle-mounted millimeter-wave radar.

[0016] Figure 2 Example diagram of a reference coordinate system for separating the static and dynamic states of a point cloud;

[0017] Figure 3 Schematic diagrams of the geometric features of different types of multipath ghosting;

[0018] Where (a) is G S Schematic diagram of the geometric features of T and P, (b) is G S G M2 A schematic diagram of the geometric features of T', where (c) represents O, P, T, and G. M1 A schematic diagram of the geometric features;

[0019] Figure 4 This is a diagram of the experimental scene.

[0020] Figure 5 This is the result of experimental data processing;

[0021] Among them, (a) is the point cloud accumulated from multiple frames in the experimental scene, (b) is the point cloud of the moving target and multipath ghosting, (c) is the multipath ghosting recognition result, and (d) is the multipath ghosting elimination result. Detailed Implementation

[0022] To facilitate understanding of the technical content of this invention by those skilled in the art, the following description, in conjunction with the accompanying drawings, further illustrates the invention.

[0023] Figure 1 This is a schematic diagram of multipath signal propagation in a typical vehicle-mounted millimeter-wave radar; the radar position is point O, the actual target position is point T, and there is a reflection point in the scene, point P; generally, the electromagnetic wave reflection is assumed to be specular reflection; therefore, the propagation paths of electromagnetic waves mainly fall into the following categories:

[0024] Direct line of sight: First, there is a two-way direct line of sight for the target, O→T→O; second, there is also a two-way direct line of sight caused by backscattering from the reflection point P, O→P→O.

[0025] First-order multipath: A first-order path refers to the electromagnetic propagation path that passes through the reflection point P once. In the above scenario, there are two first-order multipaths: one is O→T→P→O, and the other is O→P→T→O.

[0026] Second-order multipath: Second-order multipath refers to the electromagnetic propagation path that passes through the reflection point P twice, specifically: O→P→T→P→O;

[0027] Higher-order multipath propagation refers to electromagnetic propagation paths that pass through the reflection point P multiple times. Generally, because electromagnetic reflection significantly attenuates the signal energy, higher-order paths with multiple reflections can be ignored due to the weak signal energy.

[0028] The relationship between the electromagnetic propagation path and the detection target is shown in Table 1:

[0029] Table 1 Relationship between Electromagnetic Propagation Path and Detection Target

[0030] Path name transmission path The goal formed Real Target / Ghost Path-0 O→T→O T Real target Path-1 O→P→O P Real target Path-2 O→P→T→O <![CDATA[G M1 ]]> Ghost Path-3 O→T→P→O <![CDATA[G M2 ]]> Ghost Path-4 O→P→T→P→O <![CDATA[G S ]]> Ghost

[0031] Experimental scenarios such as Figure 4 As shown, the vehicle-mounted millimeter-wave radar continuously emits electromagnetic wave signals into the scene, and the receiving antenna receives the echo signals. After mixing, filtering, and digital sampling, the raw analog-to-digital converter (ADC) data is obtained. Figure 1 As shown, in a typical automotive millimeter-wave radar application scenario, the multipath ghosting generated by the reflection point P mainly includes:

[0032] Type-1 Secondary Reflection Multipath Ghost G M1 The corresponding electromagnetic wave propagation path is O→P→T→O;

[0033] Typle-2 type secondary reflection multipath ghosting G M2 The corresponding electromagnetic wave propagation path is O→T→P→O;

[0034] Typle-2 Type Triple Reflection Multipath Ghost G s The corresponding electromagnetic wave propagation path is O→P→T→P→O;

[0035] Those skilled in the art will understand that in this invention, Type-1 indicates that the electromagnetic wave is ultimately reflected back to the radar from the actual target, and Type-2 indicates that the electromagnetic wave is ultimately reflected back to the radar from the reflection point. Figure 1 The arrows shown indicate the direction of electromagnetic wave propagation.

[0036] The method of the present invention includes the following steps:

[0037] S1: Signal preprocessing;

[0038] The signal preprocessing step generates a radar point cloud from the raw echo ADC data. The main steps include fast-time Fast Fourier Transform (FFT), slow-time Fast Fourier Transform (FFT), constant false alarm rate (CFAR) detection, angle of arrival (ADR) estimation, and coordinate transformation. Assuming that after processing one frame of radar ADC data, the generated radar point cloud set is P = {p0, p1, p2, ..., p...} NP}, where N P p represents the number of points in the point cloud. i ={x i ,y i ,v i RCS i} represents the i-th radar point cloud and its multi-domain information.

[0039] Furthermore, x i ,yi ,r i ,v i RCS i These represent the x-coordinate, y-coordinate, distance of the radar point cloud relative to the radar, radial velocity of the radar point cloud relative to the radar, and calculated radar cross-section of the radar point cloud, respectively, in the vehicle coordinate system. The multi-frame point cloud accumulation result obtained from the preprocessing is shown below. Figure 5 As shown in (a).

[0040] S2: Radar point cloud processing;

[0041] 2.1 Separation of static and dynamic point clouds

[0042] The radar point cloud obtained in step S1 includes both point clouds of moving targets and point clouds of still objects in the scene. The point cloud separation step separates the original radar point cloud into moving and still objects to facilitate subsequent processing. For details, refer to... Figure 2 The vehicle coordinate system shown has the vehicle's forward direction as the y-axis and the horizontal direction perpendicular to the y-axis as the x-axis. Assuming the target is at the same height as the radar, theoretically, the radial velocity of a static object (stationary relative to the ground) measured by the vehicle-mounted millimeter-wave radar... for:

[0043]

[0044] Among them, (x r ,y r Let θ be the position of the radar in the vehicle coordinate system. r v is the angle between the radar normal direction and the vehicle coordinate system. C,x ,v C,y , w represent the lateral and longitudinal velocity components and rotational speed of the moving car, respectively. When the point cloud velocity measured by radar is... Significant differences indicate that the point cloud is moving relative to the ground. Therefore, the basis for separating the static and dynamic states of a point cloud is:

[0045]

[0046] Where |·| represents the absolute value, S and D represent the static point cloud set and the dynamic point cloud set, respectively, and △v sd As a speed threshold, in this invention, the speed threshold Δv sd An empirical value of 0.8 m / s was used. The results of separating the static and dynamic point clouds are as follows: Figure 5 As shown in (b), from Figure 5 As can be seen, in addition to the true point cloud representing the moving target, there are also many messy multipath ghostings.

[0047] 2.2 Dynamic Point Cloud Clustering

[0048] Road targets such as moving vehicles are extended targets for vehicle-mounted millimeter-wave radar. Therefore, clustering processing of the radar's dynamic point cloud D is necessary. Considering that vehicles are rigid moving targets, this invention adopts a velocity-based improved DBSCAN clustering method to cluster the dynamic point cloud. Specifically, in the neighborhood search process of the core point in the traditional DBSCAN algorithm, a further velocity check is performed on the points in the neighborhood of a core point. Points with a velocity significantly different from that of the core point are excluded from the neighborhood of that core point. In this embodiment, Δv is taken as... DBSCAN The value is 1 m / s, meaning points whose velocity difference from the core point's velocity is greater than 1 m / s are excluded from the core point's neighborhood. The remaining processing steps are the same as the traditional DBSCAN algorithm. In this invention, the two key parameters of the DBSCAN algorithm are: neighborhood search radius Eps = 2 m, and density (minimum number of points in the neighborhood) Minpts = 1. Furthermore, the velocity threshold used for core point velocity checking is taken as an empirical value Δv. DBSCAN = 1 m / s. Let the result of clustering a frame of radar data be CLU = {clu0, clu1, clu2, ..., clu...}. Ncluster}, where N cluster Indicates the number of clusters after clustering, clu i ={p j ,...},p j ∈D represents the i-th cluster.

[0049] 2.3 Acquisition of Cluster Center Coordinates and Multi-Domain Information

[0050] After clustering, extract the cluster name from each cluster. i ={p j The center coordinates of ,...}, and the multi-domain information (velocity, distance, RCS) representing the center point.

[0051] First, extract the center coordinates of the cluster. For general targets, we take the geometric center of the point cloud within the cluster as the center coordinates (x, y, z) of that cluster. c ,y c ):

[0052]

[0053] Among them, (x i ,y iLet represent the coordinates of the point cloud within the cluster, and N represent the number of point clouds within that cluster. Specifically, for large targets such as trucks, since the point clouds representing these targets have a long spatial span, the geometric center of the point clouds cannot accurately represent the areas where multipath ghosting occurs. Considering that areas with larger RCS have a higher probability of causing multipath ghosting, this invention extracts the center coordinates of clusters with an internal point cloud span exceeding 10m based on the RCS information of the point clouds. Specifically, the maximum calculated RCS value of the point clouds in the i-th cluster is first determined. Then, search the RCS calculation value and A point cloud set whose difference is within 10 dBsm is denoted as clus′. i , Finally, extract clus' i The geometric center coordinates of the interior point cloud are used as the cluster clu i The center coordinates.

[0054] Furthermore, multi-domain features representing the cluster center are extracted. This invention uses the Euclidean distance from the cluster center coordinates to the radar as the distance r from the cluster center to the radar. c The average velocity of the point cloud within the cluster is taken as the velocity v of the cluster center. c The maximum RCS value of the point cloud within the cluster is taken as the RCS of the cluster center, denoted as RCS. c .

[0055] Furthermore, record the number of point clouds within the cluster, Nup. Let the result of step 2.3 be... in

[0056] S3: Target association, associate any two moving targets in the scene;

[0057] Cluster center generated from S2 Let any two distinct points be selected from the set and associated to form a matching set. Let the association operation be AS(·), then the matching set Γ is:

[0058] Γ=AS(S)

[0059] To reduce computational load, the above steps can be further optimized. As can be seen from multipath characteristics, the relative radar distance difference between the real target and the multipath ghost is relatively small. Therefore, target points within a certain distance difference can be selected for correlation to improve the algorithm's real-time performance. Those skilled in the art should note that this "certain distance difference" can be taken as 5m, 10m, and 15m under long-range, medium-range, and short-range radar conditions, respectively.

[0060] S4: Solving for the inversion of reflection points;

[0061] Take a pair of associated points Γ0 = {c} from the matching set Γ. A ,c B},Γ0∈Γ. Where, c A ={x a ,y a ,r a ,v a RCS a ,Nup a}, c B ={x b ,y b ,r b ,v b RCS b ,Nup b Since the distance from a multipath target to the radar is always longer than the distance from the real target to the radar, we can make a further assumption, r. a <r b That is to say, c A For the true objective, c B This is a multipath ghosting. Let the radar's location be o(x). o ,y o Solve for the reflection point under three assumptions.

[0062] 4.1 Realistic Ghosting and Type-2 Triple Reflection Multipath Ghosting G S Solving for the reflection point under the assumptions

[0063] like Figure 3 As shown in (a), multipath ghosting G S The real target T is mirror-symmetric with respect to the reflecting surface where the reflection point is located. Therefore, the real target T and the multipath ghost G... S The perpendicular bisector of the connecting line segment and the multipath ghost G S The intersection of the line connecting to radar O is exactly the location of the reflection point. Assume c... A For the true target T, c B For multipath ghosting G S The specific process for solving the reflection point is as follows:

[0064] First, calculate the line.

[0065]

[0066] Furthermore, calculate the line segment perpendicular line l mid :A mid x+B mid y+C mid =0:

[0067]

[0068] Finally, the perpendicular bisector l mid With a straight line The intersection point (x) bm ,y bm That is, the reflection point P under this assumption. TS The theoretical coordinates are:

[0069]

[0070] like Figure 3 As shown, line segment Equivalent to Figure 3 (a) T, G S The line segment it is located on;

[0071] Due to factors such as target measurement accuracy and signal noise, the target positioning has errors, which in turn affect the accuracy of the reflection point solution. Therefore, we set the reflection point search area to be centered on reflection point P. TS The region centered at a circle and with radius △R is denoted as . In this embodiment, the value of △R is 1m.

[0072] 4.2 Realistic Ghosting and Type-2 Secondary Reflection Multipath Ghosting G M2 Solving for the reflection point under the assumptions

[0073] like Figure 3 As shown in (b), multipath ghosting G M2 For line segment T'G S Since the midpoint is the center of the path, we can first calculate the coordinates of T', then determine the second-order multipath ghost S according to the midpoint coordinate formula, and finally use the method described in step 4.1 to determine the reflection point. Assume c A For the true target T, c B For multipath ghosting G M2 The specific process for solving the reflection point is as follows:

[0074] First, calculate the line segment. and line segments The included angle is:

[0075]

[0076] Furthermore, point c A Rotate θ around point o ∠AOB The intersection point c can be obtained. A′ That is to say Figure 3 In (b), T' is calculated as follows:

[0077]

[0078] Furthermore, the multipath ghosting G can be calculated using the midpoint formula. S coordinate:

[0079]

[0080] Therefore, the reflection point is calculated using the method in step 4.1. Similarly, the present invention sets the reflection point search area to be based on the reflection point. The region centered at a circle and with radius △R is denoted as .

[0081] 4.3 Realistic Ghosting and Type-1 Secondary Reflection Multipath Ghosting G M1 Solving for the reflection point under the assumptions

[0082] like Figure 3 As shown in (c), multipath ghosting G M1 There are two characteristics between the two points and the real target: 1) Ghost G M1 1) The real target T and the radar coordinates are collinear; 2) This causes the multipath ghosting G M1 The reflection point falls on the elliptical locus with foci at points O and T, and the sum of the distances from the point to the two focal lengths is |OT| + |PT|. Assume c... A For the true target T, c B For multipath ghosting G M2 The specific process for solving the reflection point is as follows:

[0083] First, obtain the calculation line segment. and line segments The included angle θ ∠AOB Due to factors such as radar angle measurement accuracy and noise, multipath ghosting G... M1 The actual target T and the reflection point P are not necessarily collinear after target detection. Therefore, we set a reasonable angular deviation threshold to control the angle θ between the two line segments. ∠AOB Below the threshold value Δε θ At that time, that is:

[0084] |θ ∠AOB |≤△ε θ

[0085] It is believed that o, c A c B The three points are approximately collinear. In this embodiment, Δε θ The value is 1.5°.

[0086] Furthermore, find the locus of the ellipse described above. The semi-major axis *a* of the ellipse... E , short half-axis b E Half focal length c E The parameters are as follows:

[0087]

[0088] Furthermore, the angle between the ellipse and the positive x-axis is:

[0089]

[0090] Finally, the elliptical trajectory can be represented as:

[0091]

[0092] Among them, (x cc ,y cc () is a line segment The midpoint.

[0093] After obtaining the elliptical trajectory E, similarly, we set the reflection point search area to an elliptical annular region centered on the elliptical trajectory with scaling factors of scale1 and scale2, denoted as . In this embodiment, the scaling factors scale1 and scale2 are 0.87 and 1.15, respectively.

[0094] When |θ ∠AOB |>△ε θ If so, then exclude that assumption and let

[0095] S5: Based on the search domain solved in S4, search for reflection points in the dynamic and static point clouds to verify the hypothesis in S4.

[0096] After solving for the reflection point region of S4, we can obtain three search regions. Therefore, within the aforementioned region, the dynamic and static point cloud set P obtained in step S1 is searched to determine if a point cloud exists. If it does, then based on the aforementioned geometric constraints, Γ0={c A ,c B} is a domain that falls within the search domain Other strong RCS targets within the area cause real target and multipath ghosting correlation point pairs, and c A For the true objective, c B It is a multi-path ghost image.

[0097] S6: Multipath ghosting decision based on multi-domain information (distance, velocity, RCS and number of point clouds).

[0098] To improve the accuracy of multipath ghosting identification and avoid false identification, multi-domain features are used to further determine multipath ghosting. As known from step S4, c A ={x a ,y a ,r a ,v a RCS a ,Nup a} and c B ={x b ,y b ,r b ,vb RCS b ,Nup b The decision-maker considers the following three factors:

[0099] Factor δ1: velocity. There is a special relationship between the velocity of the multipath ghost and the velocity of the real target.

[0100]

[0101] Where, ε v These are the speed thresholds, which are taken as an empirical value of 1 m / s in this embodiment.

[0102] Factor δ2: RCS. The reflection of electromagnetic waves will cause the echo energy to attenuate, so the calculated RCS value of the real target will be greater than the calculated RCS value of the multipath ghost.

[0103]

[0104] Factor δ3: Point cloud count, indicating that the number of point clouds representing the real target is not less than the number of point clouds representing the multipath ghost image corresponding to the real target. In this embodiment, the number of point clouds corresponding to the target is the number of point clouds within the cluster corresponding to the cluster center of the target.

[0105]

[0106] In summary, the multipath target decision maker can be denoted as:

[0107]

[0108] when If the calculation result is equal to 3, then determine c. B For multipath ghosting; when If the calculated result is less than 3, then c is determined to be less than 3. B Not for the many paths and ghostly shadows;

[0109] If a cluster center is identified as a multipath ghost, then all point clouds within that cluster are also marked as multipath ghosts.

[0110] S7: Traverse each pair of associations in the matching set obtained in step S3, and repeat steps S4 to S6 to complete the multipath recognition task for all point clouds in the current frame data.

[0111] The final recognition result is as follows Figure 5 As shown in (c), the radar point cloud marked with a box is identified as multipath ghosting, from... Figure 5 As can be seen in (c), after processing with the proposed inventive technique, multipath ghosting can be effectively identified, and the result after eliminating multipath ghosting is as follows: Figure 5 As shown in (d), the experimental results verified the feasibility and effectiveness of the present invention.

[0112] Those skilled in the art will recognize that the embodiments described herein are intended to help the reader understand the principles of the invention, and should be understood that the scope of protection of the invention is not limited to such specific statements and embodiments. Various modifications and variations can be made to the invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the invention should be included within the scope of the claims of the invention.

Claims

1. A method for multipath identification of vehicle-mounted millimeter-wave radar based on reflection point inversion search, characterized in that, include: S1. Generate radar point cloud based on the ADC data of the original echo in the current frame; S2. Perform dynamic and static point cloud separation on the radar point cloud, then cluster the obtained dynamic point cloud, and finally extract the center coordinates of each cluster and the multi-domain information representing the center coordinates; the multi-domain information includes: the distance from the cluster center coordinates to the radar, the velocity of the cluster center coordinates, the radar cross-section of the cluster center coordinates, and the number of point clouds corresponding to the cluster center. S3. Select any two points from the set of cluster center coordinates obtained in step S2 and associate them to form a matching set; S4. Extract a pair of associated points from the matching set, and solve for the search region based on the reflection point inversion. Step S4 specifically includes: Take a pair of related points from the matching set and compare the distances of the two points in the pair to the radar. The point with the smaller distance is assumed to be the real target, and the other point is assumed to be a multipath ghost. Solve for the coordinates of the reflection points for the following three hypothetical scenarios, and obtain the search area for each scenario based on the coordinates of the reflection points: Assumption 1: The multipath ghosting is specifically a Typle-2 type triple-reflection multipath ghosting. ; Assumption 2: The multipath ghosting specifically refers to Type-2 secondary reflection multipath ghosting. ; Assumption 3: The multipath ghosting specifically refers to Type-1 secondary reflection multipath ghosting. ; S5. Search the search area obtained in step S4 for the existence of the radar point cloud obtained in step S1. If it exists, proceed to step S6; otherwise, return to step S4. S6. For the pair of associated points processed in step S5, multipath ghosting is determined based on multi-domain information. If multipath ghosting exists in the pair of associated points, all point clouds in the cluster corresponding to the multipath ghosting are marked as multipath ghosting. Step S6 is specifically as follows: S61. Calculate the following three factors respectively: velocity factor The calculation formula is: ； RCS factor The calculation formula is: ； Point cloud number factor The calculation formula is: ； S62. Based on the three factors calculated in step S61, perform multipath ghosting recognition; specifically: calculate the velocity factor. RCS factor Point cloud number factor The sum of the three values; if the result equals 3, then multipath ghosting is determined to exist; otherwise, it does not exist. S7. Repeat steps S4, S5, and S6 until the multipath ghosting recognition task for all point clouds in the current frame data is completed.

2. The method for multipath identification of vehicle-mounted millimeter-wave radar based on reflection point inversion search according to claim 1, characterized in that, The center coordinates of the cluster are specifically: For a cluster corresponding to a target with a point cloud span of less than or equal to 10m, the geometric center of the point cloud is taken to represent the cluster center. For a cluster corresponding to a target with a point cloud span exceeding 10m, the specific center coordinates of the cluster are as follows: A1. Calculate the maximum radar cross-section of the point cloud in this cluster. ; A2. Search for the calculated radar cross-section value within this cluster and... The point cloud set with differences within 10 dBsm; A3. Extract the geometric center coordinates of the point cloud within the point cloud set obtained in step A2 as the center coordinates of the cluster.

3. The method for multipath identification of vehicle-mounted millimeter-wave radar based on reflection point inversion search according to claim 2, characterized in that, When the center of a cluster is identified as a multipath ghost, all point clouds within that cluster are also labeled as multipath ghosts.

Images