Multipath ghosting suppression method for millimeter wave radar tunnel scene

Abstract

The invention discloses a multipath ghosting suppression method for a millimeter-wave radar tunnel scene, which is applied to the field of millimeter-wave radar target detection and aims to solve the problem of poor multipath ghosting suppression effect in the tunnel scene in the prior art. The method comprises the following steps: firstly, constructing a confidence criterion by using multi-dimensional information point cloud data generated by radar original echoes, and performing density-based clustering on the point cloud data; performing linear fitting on the zero Doppler point cloud to obtain slope information of the tunnel wall, matching the slope of the perpendicular bisector of the dynamic point cloud pair with the slope set, and judging the target pair with lower confidence in the successfully associated target pair as a multipath ghosting; and finally, in consideration of motion characteristics of targets in the tunnel and a multipath reflection geometric model, performing transverse and longitudinal double-rectangular region division on the dynamic point cloud, performing Doppler information constraint on point cloud data in the region, and screening out the targets which meet conditions and are relatively low in confidence coefficient and judging the targets as multipath ghosts. According to the invention, vehicle targets in a semi-closed environment can be detected, and multipath ghosts generated by reflection of tunnel walls, handrails and the like can be effectively identified and inhibited while real targets are reserved.

Description

Technical Field

[0001] The invention belongs to the field of millimeter wave radar target detection, and in particular relates to a multipath ghost suppression technology. Background Art

[0002] Millimeter wave radar target detection technology is to detect, locate and track the target by using the generated echo signal through the target reflecting the transmitted signal. In the intelligent transportation system, millimeter wave radar is an important sensing unit. With its unique advantages such as strong anti-interference ability in complex road environments and not being affected by weather, it plays an important role in target detection in traffic scenes, autonomous driving environment perception, and traffic information collection. In the process of radar detection, due to the reflection and diffraction of electromagnetic waves, in addition to the direct signal, there are multipath signals formed by multiple reflections and diffractions. In the process of environmental perception and target detection, the virtual ghosts formed by multipath signals not only cause certain interference to the detection and tracking of real targets, but also increase the amount of calculation in the process of data processing, resulting in unnecessary waste of computing resources. Especially in the scene of tunnel traffic radar detection, due to the presence of multiple walls and railing reflective surfaces in the semi-enclosed space, there will be a large number of multipath ghosts caused by static reflective surfaces, as well as multipath ghosts caused by multiple reflections of electromagnetic waves between different vehicles in the tunnel scene, which greatly interfere with the detection and tracking of real targets. Therefore, it is very important to suppress multipath ghosts in tunnel scenes.

[0003] Many research institutions at home and abroad have conducted research on multipath ghost suppression methods in tunnel environments. The team of Xidian University proposed a method for removing multipath ghosts in tunnels based on multi-radar videos in the patent "Method for Removing Multipath Ghosts in Tunnels Based on Multi-radar Videos". It uses millimeter-wave radar and visual information for fusion to obtain more comprehensive target information, and continues the system track ID number and visual information through target matching in the overlapping area of ​​adjacent radars, and uses whether the visual information exists in the target to remove multipath ghosts. The above method achieves the suppression of multipath interference through the method of multi-radar and visual fusion perception. The visual information in this method is greatly affected by light, and the use of multiple radars for target matching requires a high cost. The team of Tongji University proposed a method for removing mirror images of tunnel trajectory data based on millimeter-wave radar in the patent "A method for removing mirror images of tunnel trajectory data based on millimeter-wave radar". This method estimates the tunnel wall to eliminate the mirror image vehicle trajectory data outside the tunnel wall range, and at the same time establishes a trajectory credibility evaluation index. By scoring the credibility of the target trajectory, the vehicle trajectory data with a score lower than the lane credibility score is eliminated, thereby realizing the identification and suppression of multipath ghosts. This method only identifies and suppresses the mirror image ghosts caused by the tunnel wall, and cannot suppress the multipath interference caused by multiple reflections in the tunnel. At the same time, for this scenario, researchers from Xidian University proposed a method for removing multipath interference in tunnel tracks based on zero Doppler clutter points in the patent "Method for removing multipath interference in tunnel tracks based on zero Doppler clutter points". This method obtains the distance and speed of the target in the tunnel, uses the maximum likelihood algorithm to obtain the angles of the left multipath and right multipath of the target, generates the zero channel clutter point vector at each moment, and uses the target area and multipath area to remove multipath interference in the track. This method divides the multipath area according to the target information, uses certain target prior information, and at the same time cannot interfere with and suppress the ghost in the tunnel. Summary of the invention

[0004] In order to solve the above technical problems, the present invention proposes a multipath ghost suppression method for a millimeter wave radar tunnel scene.

[0005] The technical solution adopted by the present invention is: a multipath ghost suppression method for a millimeter wave radar tunnel scene, comprising:

[0006] S1: The original echo data received by the traffic millimeter wave radar is sampled by ADC (Analog to Digital Converter), and the data is subjected to two-dimensional fast Fourier transform, constant false alarm target detection and arrival angle estimation to obtain radar point cloud data;

[0007] S2: Perform spatial projection transformation on the radar point cloud data, taking into account the installation height of the radar, and realize the coordinate and radial velocity projection transformation from three-dimensional space to two-dimensional space;

[0008] S3: using a density-based clustering algorithm to complete point aggregation on the point cloud data unified into a two-dimensional space in step S2;

[0009] S4: Construct the confidence evaluation criteria of the radar point cloud based on multi-dimensional information and calculate the confidence of each point cloud data;

[0010] S5: For the point cloud data after the point traces are condensed in step S4, zero Doppler point cloud separation is performed based on radial velocity information to obtain dynamic point cloud clusters and static point cloud clusters respectively; linear fitting is performed on the static point cloud clusters to obtain a reflection boundary slope set;

[0011] S6: Based on the multipath propagation geometry model, dynamic target matching is performed according to the reflection boundary slope set obtained in step S5, confidence evaluation is performed on target point pairs that meet the conditions, and multipath ghosts are identified; and the identified multipath ghosts are eliminated to obtain a dynamic point cloud cluster set;

[0012] S7: divide the dynamic point cloud clustering set into candidate areas; and use the speed feature information to constrain and complete the confidence parameter evaluation of the point cloud data, so as to identify the multipath ghosts in the dynamic point cloud clustering set again, and eliminate the identified multipath ghosts to obtain the point cloud data after suppressing the multipath ghosts.

[0013] Beneficial effects of the present invention: The present invention proposes a multipath ghost suppression method for millimeter-wave radar tunnel scenes, which can effectively identify second-order multipath ghosts caused by reflective surfaces such as walls and railings in semi-closed tunnel scenes. At the same time, based on the motion characteristics of vehicle targets in the tunnel, the recognition and suppression of first-order multipath ghosts in the tunnel are realized. Based on the constructed confidence criterion, multipath ghosts can be suppressed without causing the loss of real targets. Compared with the multipath suppression method mentioned in the technical background, the present invention, under the detection conditions of a single radar sensor, does not rely on the auxiliary information of other sensors, perceives the tunnel environment based on radar point cloud data, and identifies and suppresses multipath ghosts caused by reflections inside and outside the tunnel. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 This is a schematic diagram of the tunnel scene;

[0015] Figure 2 It is a top view of electromagnetic wave reflection in tunnel environment;

[0016] Figure 3 It is the reflection geometry model of the left wall of the tunnel;

[0017] Figure 4 It is the reflection geometry model of the right wall of the tunnel;

[0018] Figure 5 It is a top view of electromagnetic wave reflection in tunnel environment;

[0019] Figure 6 It is a schematic diagram of spatial projection transformation;

[0020] Wherein, (a) is a top view; (b) is a side view;

[0021] Figure 7 is the reflection geometry model in the tunnel;

[0022] Figure 8 is the result of processing the measured data;

[0023] Among them, (a) is the experimental scene; (b) is the multi-frame accumulation result after clustering the original point cloud; (c) is the multi-frame multipath ghost recognition result; (d) is the multi-frame multipath ghost suppression result. DETAILED DESCRIPTION

[0024] To facilitate those skilled in the art to understand the technical content of the present invention, the present invention is further explained below with reference to the accompanying drawings.

[0025] The present invention provides a multipath ghost suppression method for a millimeter wave radar tunnel scene. First, point cloud data is generated according to the original radar echo, and a confidence criterion is constructed based on the multidimensional information of the point cloud data, and density-based clustering is implemented for the data. A tunnel multipath ghost suppression method based on slope set matching is proposed for the second-order multipath ghosts generated by the tunnel walls and railings. Specifically, the slope of each cluster cluster is obtained by linear fitting of the zero Doppler point cloud, and the midpoint perpendicular slope of any target point pair is obtained in the dynamic point cloud and matched with the slope set, and the target point pairs with low confidence in the successful association are identified as multipath ghosts; secondly, for the multipath ghosts formed by multiple reflections of electromagnetic waves between vehicles and walls in the tunnel, considering the motion characteristics of the vehicles in the tunnel, the target is divided into double rectangular areas in the horizontal and vertical directions, and the point cloud data in the area is constrained by Doppler information, and the candidate targets that meet the conditions are evaluated for confidence, thereby realizing the identification and suppression of multipath ghosts in the tunnel.

[0026] like Figure 1 The tunnel scene diagram is shown in Figure 1, and the top view obtained by geometric modeling of the scene diagram is shown in Figure 2. Figure 2As shown in the figure. Radar R is located at the tunnel entrance, and there are pedestrian railings in the semi-enclosed tunnel. The left wall, right wall and pedestrian railings of the tunnel are strong reflective surfaces. In addition to the direct line of sight signal for the vehicle target in the tunnel, the electromagnetic wave signal also has multipath ghosts formed by multiple reflections between the strong reflective surfaces. The electromagnetic wave propagation path model is established based on ray tracing theory.

[0027] The left wall of the tunnel is used as a reflective surface. Figure 3 As shown, the specific propagation path of the electromagnetic wave signal is:

[0028] Direct line of sight path: R→T→R, the electromagnetic wave signal is transmitted from the radar R to the real target T and then returns to the radar R;

[0029] First-order multipath: R→T→Q 1 →R, after the electromagnetic wave signal is transmitted from the radar R to the real target T, it passes through the left wall Q of the tunnel 1 Reflected back to the radar R, generating a first-order multipath ghost S 1 ;

[0030] Second-order multipath: R→Q 1 →T→Q 1 →R, the electromagnetic wave signal is emitted from the radar R and reflected to the left wall Q of the tunnel 1 , after reaching the real target T, pass through the left wall of the tunnel Q 1 Back to the radar R, a second-order multipath ghost S is generated;

[0031] The right wall of the tunnel is used as a reflective surface. Figure 4 As shown, the specific propagation path of the electromagnetic wave signal is:

[0032] Direct line of sight path: R→T→R, the electromagnetic wave signal is transmitted from the radar R to the real target T and then returns to the radar R;

[0033] First-order multipath: R→T→Q 3 →R, the electromagnetic wave signal is transmitted from the radar R to the real target T, and then passes through the right wall Q of the tunnel 1 Reflected back to the radar R, generating a first-order multipath ghost L 1 ;

[0034] Second-order multipath: R→Q 3 →T→Q 3 →R, the electromagnetic wave signal is emitted from the radar R and reflected to the right wall Q of the tunnel 3 , after reaching the real target T, pass through the right wall Q of the tunnel 3 Back to the radar R, a second-order multipath ghost L is generated;

[0035] The railings in the tunnel serve as a reflective surface, and the specific propagation path of the electromagnetic wave signal is:

[0036] First-order multipath: R→T→Q 2→R, the electromagnetic wave signal is transmitted from the radar R to the real target T, and then passes through the tunnel railing Q 2 Reflected back to the radar R, generating a first-order multipath ghost G 1 ;

[0037] Second-order multipath: R→Q 2 →T→Q 2 →R, the electromagnetic wave signal is emitted from the radar R and reflected to the tunnel railing Q 2 , after reaching the real target T, pass through the right wall Q of the tunnel 2 Back to the radar R, a second-order multipath ghost G is generated;

[0038] Considering the strong energy attenuation of millimeter wave radar signals after multiple reflections, the energy of high-order path signals with multiple reflections is weak and can be ignored. Only the real targets formed by the first and second reflections, the first-order and second-order ghosts are considered. The top view of electromagnetic wave reflection in the tunnel environment is shown in the figure below. Figure 5 shown.

[0039] Based on the modeling of electromagnetic wave propagation signals in the above radar detection scenario, the specific implementation steps of the present invention are given below:

[0040] Step 1: Radar point cloud data generation

[0041] The received millimeter-wave radar echo signal is sampled by ADC to obtain a data matrix represents the matrix, N R Indicates the number of ADC sampling points, N D Indicates the number of millimeter-wave radar chirp signals, N A Indicates the number of echo data antenna channels, and performs N r Perform a fast Fourier transform in the time dimension to obtain a data matrix containing target distance information. At the same time, N d Perform a fast Fourier transform of the time dimension to obtain the range Doppler data matrix Considering that the energy of radar echo is weak in multiple reflections of electromagnetic waves, the obtained range Doppler data matrix is ​​incoherently accumulated in the antenna channel dimension to obtain a two-dimensional data matrix

[0042]

[0043] The matrix is ​​subjected to unit average constant false alarm detection to obtain the target's distance and Doppler frequency information. Finally, the target's arrival angle is estimated to obtain the target's azimuth information, thereby generating a matrix containing the target distance R, radial velocity v D , azimuth angle θ, radar scattering cross-sectional area rcs, point cloud data signal-to-noise ratio snr, and radar point cloud data of received power power.

[0044] Step 2: Space Projection Transformation

[0045] In road traffic scenes, targets usually move in a two-dimensional plane, and the vertical resolution requirement is low. Therefore, most millimeter-wave radars used to detect such scenes are uniform linear arrays, which cannot obtain information about the pitch angle and have a certain installation height. The point cloud data obtained is based on the radar coordinate system, so it is necessary to perform a projection transformation from three-dimensional space to two-dimensional space on the point cloud data. Figure 6 Assume that the installation height of the radar is h, realize the coordinate transformation from three-dimensional space to two-dimensional space for the radar point cloud, project the radial distance R in three-dimensional space into two-dimensional space, and obtain the radial distance r and altitude angle α as follows:

[0046]

[0047] Based on the azimuth angle θ of the target point cloud detected by the radar and the radial distance r projected into the two-dimensional space, the position coordinates (x, y) of the target are obtained as follows:

[0048]

[0049] The radial velocity v in three-dimensional space D' Projected into two-dimensional space:

[0050] v d =v D' sinα

[0051] By performing spatial projection transformation on the radar point cloud data, its spatial position information is unified into a two-dimensional plane to obtain the point cloud data set P:

[0052]

[0053] Among them, x i Indicates the lateral distance of the i-th point cloud, y i represents the longitudinal distance of the i-th point cloud, θ i Indicates the azimuth, rcs i represents the target scattering area, power i Indicates signal power, snr i represents the signal-to-noise ratio, Represents the set of rational numbers.

[0054] Step 3: Density-based spatial clustering of point cloud data

[0055] Millimeter-wave radar has the characteristics of short wavelength, large bandwidth, and miniaturized antenna, so it has a smaller distance resolution and higher spatial angle resolution, can more accurately distinguish the details of the target, and distinguish different reflection points in a smaller spatial range. Since the vehicles, railings, walls and other targets in the road environment are composed of multiple reflection surfaces, the millimeter-wave radar with high resolution can clearly distinguish multiple parts of a target, so a single target will generate multiple reflection points to form point cloud data. The point cloud data set P after the spatial projection transformation in step 2 is clustered using a density-based spatial clustering method. Determine the clustering parameters: neighborhood radius eps and neighborhood minimum point number Minpts, and cluster the point cloud set by setting reasonable clustering parameters. In this embodiment, the neighborhood radius is set to 2m according to the width of the vehicle in the tunnel. Considering that the millimeter-wave radar has the characteristics of high resolution, so that the measured target has a certain extension, combined with the actual data, in this embodiment, the minimum number of neighborhood points is taken as 5. Assume that after the point cloud data is spatially clustered based on density, a total of l clusters are obtained, and the cluster set generated by the radar data of a single frame is:

[0056] L = {cluster 1 ,cluster 2 ,cluster 3 ,...cluster l}

[0057] Among them, cluster represents a cluster, and the cluster contains point cloud data that meets the clustering conditions. At the same time, for each point cloud cluster after clustering i Select the cluster center. Considering that the distribution of points in the clusters of each target in point cloud clustering is uneven, directly using the mean as the cluster center is easily affected by the deviation of abnormal points in the cluster. Therefore, the center point is selected as the cluster center in the cluster. The sum of the distances from this point to all other points in the cluster is the smallest, and it is a real data point. Assume that cluster i The cluster contains N i 6-dimensional data points, Define any two points p in the cluster i and p j The Euclidean distance is:

[0058] d(p i ,p j )=||p i -p j ||

[0059] For each point p in the cluster i ∈cluster i , calculate the sum of the distances D(p i )for:

[0060]

[0061] You can get the i-th cluster cluster i The sum of the distances between each point in the vector D and the rest of the points i for:

[0062] D i ={D(p 1 ),D(p 2 ),...,D(p Ni )}

[0063] Select the point with the smallest distance and the smallest distance as the center point p of the cluster medoid :

[0064]

[0065] Among them, p medoid ={x i ,y i ,θ i ,rcs i ,power i ,snr i}Use the multidimensional vector of the center point to represent the cluster center of the cluster, and record the number of point clouds in the cluster num i , as a type of feature of the cluster center, is used to describe the density of the cluster. The updated cluster center point is:

[0066]

[0067] Step 4: Construct confidence evaluation criteria

[0068] Based on the multi-dimensional feature information of point cloud data, taking into account the large differences between the real target and the multipath ghost in radar scattering cross-section area rcs, point cloud data signal-to-noise ratio snr, and received power power, the radar equation is used to determine that the multi-dimensional features have a certain correlation. Therefore, a confidence evaluation criterion is constructed to further judge the candidate multipath ghost.

[0069] Define each point cloud cluster i The multidimensional feature vector M of the midpoint cloud i =[rcs i ,snr i ,power i ] T Obey a joint multidimensional Gaussian distribution:

[0070] M i =[rcs i ,snri ,power i ]~N(μ i ,Σ i )

[0071] Among them, μ i and Σ i are the characteristic mean vector and covariance matrix of the cluster respectively. For each point cloud cluster, the corresponding characteristic mean vector can be calculated for:

[0072]

[0073] At the same time, for point cloud cluster i The data points in the table calculate the correlation between different features and obtain the covariance matrix Σ of the Gaussian distribution. i for:

[0074]

[0075] Among them, the covariance is defined as:

[0076]

[0077] For each cluster cluster i The joint probability density function of multidimensional information definition is:

[0078]

[0079] Wherein, d represents the dimension of the multidimensional vector, and in this embodiment, the value is 3, Σ i and μ i Respectively represent the mean and covariance of the multidimensional features of the cluster point cloud data, and define the obtained probability density value f(i) as the confidence of the cluster. For each point cloud cluster cluster i (i=1,2,3...N i ) calculates the confidence level to describe whether the cluster is a real target, and uses it as the cluster center p of the cluster medoid New one-dimensional features, the updated cluster center points are:

[0080]

[0081] Step 5: Tunnel Environment Perception

[0082] The semi-closed tunnel environment contains static building structures such as continuous walls and railings, but their surfaces are not completely smooth, but have a certain degree of roughness, which will cause the scattering of electromagnetic wave reflection signals and form multiple reflection points. Considering factors such as radar beam width, the perception of static environments often presents discrete point cloud data. The static environment contains a large number of reflective surfaces, which causes multipath ghosts in the detection process of real targets. Therefore, the set L obtained after clustering in step 3 is separated by zero Doppler point cloud. Considering that there is a certain speed measurement error in the detection process of radar, the Doppler velocity error △v is defined. D Perform point cloud separation to obtain dynamic point cloud clusters L dy And static point cloud cluster L st Theoretically, the Doppler velocity measurement of a static object by a millimeter wave radar is 0 m / s. However, considering that a certain velocity measurement error exists in the millimeter wave radar, the Doppler velocity error △v is selected in the present invention. D It is 0.05m / s.

[0083] There are k static point cloud clusters in a single frame of data Cluster each static point cloud The least square method is used for linear fitting. Assume that the straight line fitted by the cluster is expressed as:

[0084]

[0085] in, is the slope of the fitted line for the cluster, is the intercept of the fitted line. Define the loss function of the slope and intercept of the clustering cluster for:

[0086]

[0087] The loss function is respectively about the variable a k , b k The optimal solution obtained by partial derivative is:

[0088]

[0089] Among them, (x i ,y i ) is the cluster The static point coordinates in the static point cloud set are linearly fitted to each cluster in the static point cloud set to form a slope set:

[0090] Step 6: Matching and associating moving targets

[0091] For the dynamic point cloud cluster L obtained in step 5 dyThe cluster center points are matched and associated with each other. The moving target association matrix is ​​defined as:

[0092]

[0093] in, are the centers of any two clusters in the dynamic point cloud cluster. According to the typical multipath propagation geometry model, the real target and the second-order multipath ghost are mirror-symmetric about the reflection surface. Calculate the dynamic target association pair The slope of the perpendicular bisector k ij for:

[0094]

[0095] Determine the slope k of the perpendicular bisector of the associated pair ij Whether k is satisfied ij ∈A s If it belongs to this set, it satisfies the "real target-multipath ghost" association pair, traverses the association matrix composed of all cluster center points of the dynamic point cloud cluster clusters for judgment, and associates the target association pairs that meet the conditions The confidence judgment is performed to obtain the multipath ghost recognition result Re:

[0096]

[0097] in, are the confidences of the associated target pairs respectively, and the dynamic point cloud cluster set obtained by eliminating the identified multipath ghosts is L' dy .

[0098] Step 7: Candidate region division and feature constraints

[0099] There are multiple static reflective surfaces such as walls and railings in a semi-enclosed tunnel environment. When detecting a real vehicle target, the electromagnetic wave is reflected multiple times between the static building structure and the real target, which is easy to form multipath ghosts. However, due to the lateral distance R between the reflective surface and the vehicle target in the tunnel environment, th When the distance is small, according to the multipath propagation geometric model, when the distance is close, the real target and the first-order multipath ghost formed by the reflection are geometrically overlapped in the distance measurement. At the same time, combined with the characteristics of the vehicle moving in a straight line in the tunnel scene and the lane width R road Information, the dynamic point cloud cluster obtained in step 6 The candidate regions are divided into Figure 7 shown.

[0100] Assume that the center of the dynamic point cloud cluster Define the width w for the geometric center 1 and length h 1 The rectangular area for:

[0101]

[0102] Similarly, define the width as w 2 and length h 2 The rectangular area for:

[0103]

[0104] In the actual scene, combined with experimental data, considering the size and movement characteristics of the vehicle itself in the tunnel, a rectangular area is set The width and length are 4m and 3m respectively. Considering the geometric characteristics of multipath ghost distribution caused by multiple reflections of electromagnetic waves in the tunnel, a rectangular area is set. The width and length are 2m and 6m respectively.

[0105] There is a correlation between the real target and the different multipath ghost velocities formed by multiple reflections, which can be satisfied through theoretical derivation and simulation verification:

[0106] |v real -v ghost |≤△v th

[0107] Among them, v real , v ghost Denote the speed of the real target and multipath ghost respectively, △v th is the threshold of the speed difference. In this embodiment, a simulation experiment is carried out based on a typical multipath propagation scenario. The simulation conditions are: reflection boundary x=2m, real target position x=3m, y=6m, real target speed is 3m / s, and 30 frames are simulated. It is found that the maximum Doppler speed difference between the real target and the first-order ghost is 0.3m / s, and the maximum Doppler speed difference between the real target and the second-order ghost is 0.7m / s. In the tunnel scenario, taking into account the certain speed measurement error of the millimeter-wave radar and the actual scenario characteristics, the speed threshold is set to 1m / s in this embodiment.

[0108] To satisfy the dynamic point cloud cluster center Candidate area Point cloud data is used to constrain the velocity feature dimension and evaluate the confidence of the points that meet the conditions. By comparing the confidence, points with low confidence that are both located in the candidate area and meet the velocity feature dimension constraints will be regarded as multipath ghosts. That is, when the point cloud data The following judgment relationship is met for multipath ghost:

[0109]

[0110] The proposed algorithm is verified using measured data. The experimental scenario is as follows: Figure 8 As shown in (a), Figure 8 (b) is the result after clustering the original radar point cloud data; Figure 8 (c) is the recognition result of multipath ghost in the tunnel scene; Figure 8 (d) From Figure 8 (b) Remove Figure 8 (c) The suppression result after identifying the multipath ghost.

[0111] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and should be understood that the scope of protection of the present invention is not limited to such specific statements and embodiments. For those skilled in the art, the present invention may have various changes and variations. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of the claims of the present invention.

Claims

1. A multipath ghost suppression method for a millimeter wave radar tunnel scene, characterized in that: include: S1: ADC samples the raw echo data received by the traffic millimeter-wave radar, and performs two-dimensional fast Fourier transform, constant false alarm target detection and arrival angle estimation on the data to obtain radar point cloud data; S2: Perform spatial projection transformation on the radar point cloud data, taking into account the installation height of the radar, and realize the coordinate and radial velocity projection transformation from three-dimensional space to two-dimensional space; S3: using a density-based clustering algorithm to complete point aggregation on the two-dimensional point cloud data obtained in step S2; S4: Construct the confidence evaluation criteria of radar point cloud based on multi-dimensional information and calculate the confidence of each cluster; S5: For the point cloud data after the point traces are condensed in step S3, zero Doppler point cloud separation is performed based on radial velocity information to obtain dynamic point cloud clusters and static point cloud clusters respectively; linear fitting is performed on the static point cloud clusters to obtain a reflection boundary slope set; S6: Based on the multipath propagation geometry model, dynamic target matching is performed according to the reflection boundary slope set obtained in step S5, confidence evaluation is performed on target point pairs that meet the conditions, and multipath ghosts are identified; and the identified multipath ghosts are eliminated to obtain a dynamic point cloud cluster set; S7: divide the dynamic point cloud clustering set into candidate areas; and use the speed feature information to constrain and complete the confidence parameter evaluation of the point cloud data, so as to identify the multipath ghosts in the dynamic point cloud clustering set again, and eliminate the identified multipath ghosts to obtain the point cloud data after suppressing the multipath ghosts.

2. The multipath ghost suppression method for millimeter wave radar tunnel scene according to claim 1 is characterized in that: The two-dimensional fast Fourier transform in step S1 includes: performing a fast time-dimensional fast Fourier transform on the sampled data matrix to obtain a data matrix N containing target distance information r , for N r Perform a slow-time fast Fourier transform to obtain the range Doppler data matrix N d .

3. The multipath ghost suppression method for millimeter wave radar tunnel scene according to claim 2 is characterized in that: It also includes performing incoherent accumulation of the range Doppler data matrix in the antenna channel dimension to obtain a two-dimensional data matrix N c : Among them, N A Indicates the number of antenna channels.

4. The multipath ghost suppression method for millimeter wave radar tunnel scene according to claim 3 is characterized in that: Step S2 specifically includes: Assuming that the installation height of the radar is h, the coordinate transformation of the radar point cloud from three-dimensional space to two-dimensional space is realized, and the radial distance R in three-dimensional space is projected into two-dimensional space, and the radial distance r and altitude angle α are obtained as follows: Based on the azimuth angle θ of the target point cloud detected by the radar and the radial distance r projected into the two-dimensional space, the position coordinates (x, y) of the target are obtained as follows: The radial velocity v in three-dimensional space D' Projected into two-dimensional space: v d =v D' sinα By performing spatial projection transformation on the radar point cloud data, its spatial position information is unified into a two-dimensional plane to obtain the point cloud data set P:

5. The multipath ghost suppression method for millimeter wave radar tunnel scene according to claim 4 is characterized in that: Step S3 selects a central point in the cluster as the cluster center, the sum of the distances from this point to all other points in the cluster is the smallest, and it is a real data point.

6. The multipath ghost suppression method for the millimeter wave radar tunnel scene according to claim 5 is characterized in that: The implementation process of step S4 is: Define each point cloud cluster i The multidimensional feature vector M of the midpoint cloud i =[rcs i ,snr i ,power i ] T Obey a joint multidimensional Gaussian distribution: M i =[rcs i ,snr i ,power i ]~N(μ i ,S i ) Among them, μ i ,Σ i are the characteristic mean vector and covariance matrix of the cluster respectively; For each point cloud cluster, the corresponding feature mean vector can be calculated for: At the same time, for point cloud cluster i The data points in the table calculate the correlation between different features and obtain the covariance matrix Σ of the Gaussian distribution. i for: Among them, the covariance is defined as: For each cluster cluster i The joint probability density function of multidimensional information definition is: Where d represents the dimension of the multidimensional vector, Σ i and μ i Respectively represent the mean and covariance of the multidimensional features of the cluster point cloud data; The obtained probability density value f(i) is defined as the confidence of the cluster.

7. The multipath ghost suppression method for millimeter wave radar tunnel scene according to claim 6 is characterized in that: In step S5, linear fitting is performed on the static point cloud clusters to obtain a set of reflection boundary slopes. The specific process is as follows: The fitted straight line is expressed as: in, is the slope of the fitted straight line for the static point cloud clustering cluster, is the intercept of the fitted straight line; Define the loss function with slope and intercept for clustering for: The loss function is respectively about the variable a k , b k The optimal solution obtained by partial derivative is: Among them, (x i ,y i ) is a static point cloud cluster The static point coordinates in ; By performing linear fitting on each cluster in the static point cloud set, the slope set is constructed as 8. The multipath ghost suppression method for millimeter wave radar tunnel scene according to claim 7 is characterized in that: The implementation process of step S6 is: The cluster center points of the dynamic point cloud clusters obtained in step S5 are matched and associated with each other; the dynamic target association matrix is ​​defined as: in, are any two cluster center points in the dynamic point cloud cluster, i≠j; According to the typical multipath propagation geometric model, the real target and the second-order multipath ghost are mirror-symmetric about the reflection surface; the dynamic target association pair is calculated. The slope of the perpendicular bisector k ij for: Traverse the association matrix composed of all cluster center points of the dynamic point cloud cluster to make a judgment and select the points that meet the condition k ij ∈A s Target association pair After making a confidence decision, the multipath ghost recognition result Re is obtained as: in, are the confidences of the associated target pairs respectively; The dynamic point cloud cluster set obtained by eliminating the identified multipath ghosts is L' dy .

9. The multipath ghost suppression method for millimeter wave radar tunnel scene according to claim 8 is characterized in that: The specific implementation process of step S7 is: Considering the size and motion characteristics of the target itself, the center of the dynamic point cloud cluster As the geometric center, define the first rectangular area Considering the geometric characteristics of multipath ghost distribution caused by multiple reflections of electromagnetic waves in the tunnel, the center of the dynamic point cloud cluster is used as the As the geometric center, define the second rectangular area The real target and the different multipath ghost velocities formed by multiple reflections are correlated, satisfying: |v real -v ghost |≤△v th Among them, v real , v ghost Denote the speed of the real target and multipath ghost respectively, △v th is the threshold of speed difference; To satisfy the dynamic point cloud cluster center Candidate area The point cloud data is used to constrain the velocity feature dimension and to evaluate the confidence of the points that meet the conditions. The multipath condition satisfies the following decision relationship:

Images