CN114137509B - Millimeter wave Lei Dadian cloud clustering method and device - Google Patents
Millimeter wave Lei Dadian cloud clustering method and device Download PDFInfo
- Publication number
- CN114137509B CN114137509B CN202111447566.5A CN202111447566A CN114137509B CN 114137509 B CN114137509 B CN 114137509B CN 202111447566 A CN202111447566 A CN 202111447566A CN 114137509 B CN114137509 B CN 114137509B
- Authority
- CN
- China
- Prior art keywords
- track
- point cloud
- target
- detected
- dadian
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 36
- 230000001343 mnemonic effect Effects 0.000 claims description 5
- 238000004891 communication Methods 0.000 claims description 3
- 230000000694 effects Effects 0.000 abstract description 9
- 238000010586 diagram Methods 0.000 description 7
- 238000004590 computer program Methods 0.000 description 6
- 230000008569 process Effects 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000007547 defect Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
- G01S13/66—Radar-tracking systems; Analogous systems
- G01S13/72—Radar-tracking systems; Analogous systems for two-dimensional tracking, e.g. combination of angle and range tracking, track-while-scan radar
- G01S13/723—Radar-tracking systems; Analogous systems for two-dimensional tracking, e.g. combination of angle and range tracking, track-while-scan radar by using numerical data
- G01S13/726—Multiple target tracking
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
- G01S13/88—Radar or analogous systems specially adapted for specific applications
Landscapes
- Engineering & Computer Science (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Physics & Mathematics (AREA)
- Computer Networks & Wireless Communication (AREA)
- General Physics & Mathematics (AREA)
- Electromagnetism (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
The application provides a millimeter wave Lei Dadian cloud clustering method and device, which are used for acquiring current frame Lei Dadian cloud information of a target to be detected and track information tracked and output by the previous frame; then, according to a preset algorithm, the current frame Lei Dadian cloud information and the track information output by the previous frame tracking are associated to obtain track coordinates, course angles and vehicle type classification of the target to be detected; and finally, taking the elliptic wave gate as priori knowledge and classifying the track coordinates, the course angle and the vehicle type of the target to be detected as posterior knowledge, and carrying out point cloud clustering on the target to be detected. According to the application, according to the priori knowledge that the vehicle is rectangular, the wave gate used for calculating the core object in the clustering is set as an ellipse, the point cloud sample obtains the course angle and the vehicle type classification through the associated track, and then the direction and the size of the elliptical wave gate are changed through the course angle and the vehicle type classification, so that the clustering effect of the point cloud clustering is improved, and the target tracking quality is improved.
Description
Technical Field
The application relates to the technical field of data processing, in particular to a millimeter wave Lei Dadian cloud clustering method and device.
Background
The millimeter wave radar is basically divided into two steps in traffic tracking: the method comprises the steps of point cloud clustering and target tracking, wherein the output of the point cloud clustering is used as the input of the target tracking, and therefore the quality of the target tracking is directly affected by the quality of the point cloud clustering.
Currently, density clustering algorithms (e.g., DBSCAN) are typically used due to sparsity of millimeter wave Lei Dadian clouds and unknowns of target numbers. However, the point cloud characteristics of the millimeter wave radar in the traffic scene are: the small targets and the distant targets have a small number of points or even only one point, the large targets have a plurality of points but have larger dispersibility, and the near radar endpoint cloud of the same target is dense and far from the endpoint cloud is sparse, so when a DBSCAN clustering algorithm is used, in order to ensure that the small targets can be clustered normally, the minimum clustering point number needs to be set to be 1, the epsilon in epsilon-field is very difficult to set, if the small targets are too large, the small targets are clustered into one cluster, and if the small targets are too small, vehicles with a small point are clustered into a plurality of clusters; both of these situations can degrade the quality of the subsequent tracking.
Therefore, how to improve the clustering effect of the point cloud clusters, so as to improve the quality of target tracking, is a problem to be solved by those skilled in the art.
Disclosure of Invention
In view of the above, the application provides a millimeter wave Lei Dadian cloud clustering method and device for improving the clustering quality of point cloud clustering, thereby improving the target tracking quality.
In order to achieve the above object, the present application provides the following technical solutions:
a millimeter wave Lei Dadian cloud-based clustering method, comprising:
acquiring current frame Lei Dadian cloud information of a target to be detected and track information tracked and output by the previous frame, wherein the radar point cloud information comprises positions and time stamps of the target to be detected, and the track information comprises coordinates, speeds, classifications and time stamps;
correlating the current frame Lei Dadian cloud information with the track information tracked and output by the previous frame according to a preset algorithm to obtain track coordinates, course angles and vehicle type classification of the target to be detected;
and carrying out point cloud clustering on the object to be detected by taking the elliptic wave gate as priori knowledge and taking the track coordinates, the course angle and the vehicle type of the object to be detected as posterior knowledge.
Further, the associating the current frame Lei Dadian cloud information with the track information output by the previous frame tracking according to the preset algorithm to obtain track coordinates, a course angle and a vehicle type classification of the target to be detected includes:
predicting track coordinates of the current moment of the track according to a first preset step for track information tracked and output by the previous frame;
and the current frame Lei Dadian cloud information endows the point cloud with course angles and vehicle type classification according to a second preset step.
Further, the first preset step includes:
predicting the current position of the track according to the previous frame track and the uniform motion model;
the course angle of the track at the current moment directly uses the course angle of the track of the upper frame to obtain the track set at the current moment.
Further, the second preset step includes:
determining a sample o according to a first preset formula i In the nearest neighbor track of the track set T', the first preset formula is thatWherein (1)>
Determining a sample o according to a second preset formula i The second preset formula is thatWhere inf is a maximum mnemonic, the sample may be from a radar echo of a new target;
determining a sample o according to a third preset formula i The third preset formula is thatWhere 0 indicates no classification attribute and the sample may come from a radar echo of a new target.
Further, the classifying the to-be-detected target with the elliptic wave gate as a priori knowledge and the track coordinates, the course angle and the vehicle type of the to-be-detected target as posterior knowledge, includes:
s1: taking an elliptic wave gate as priori knowledge and classifying track coordinates, course angles and vehicle types of the target to be detected as posterior knowledge to obtain a point cloud sample set D' = { (o) of comprehensive posterior knowledge i ，θ i ，c i )|o i ＝(x i ，y i )，i∈[1，...，m]}；
S2: initializing a core object set Ω=Φ, clustering cluster number k=0, unvisited sample set Γ=d', and cluster division c=Φ;
s3: for a point cloud sample set D' fused with posterior knowledge, finding out all core objects according to a preset step;
s4: if the core object set Ω=Φ, then the set c= { C is partitioned according to clusters 1 ，C 2 ，...，C k Samples in } calculate centroid coordinates for each clusterObtaining a cluster centroid coordinate set S= { S 1 ，s 2 ，...，s k }；
S5: if the core object set Ω is not equal to φ, randomly selecting one core object o from the core object set Ω, and initializing the current cluster core object queue Ω cur = { o }, initializing a class sequence number k=k+1, initializing a current cluster sample set C k = { o }, update the unvisited sample set Γ = Γ - { o };
s6: if the current cluster core object queue omega cur Cluster C =φ k After the generation is completed, updating cluster division C= { C 1 ，C 2 ，...，C k Update core object set Ω=Ω -C k S4, switching to S4; otherwise update the core object set Ω=Ω -C k ；
S7: at the current cluster core object queue Ω cur A core object o 'is taken out, a sample set N (o') in an o 'wave gate is found, delta=N (o') ∈Γ is set, and a current cluster sample set C is updated k ＝C k Update non-accessed sample set Γ=Γ - Δ, update Ω cur ＝Ω cur U (delta. U omega) -o', and S6;
s8: outputting the clustered cluster number k and a cluster centroid coordinate set S= { S 1 ，s 2 ，...，s k }。
A millimeter wave Lei Dadian cloud-based cluster device, comprising:
the first processing unit is used for acquiring current frame Lei Dadian cloud information of a target to be detected and track information tracked and output by the previous frame, wherein the radar point cloud information comprises positions and time stamps of the target to be detected, and the track information comprises coordinates, speeds, classifications and time stamps;
the second processing unit is used for associating the current frame Lei Dadian cloud information with the track information tracked and output by the previous frame according to a preset algorithm to obtain track coordinates, course angles and vehicle type classification of the target to be detected;
and the third processing unit is used for carrying out point cloud clustering on the object to be detected by taking the elliptic wave gate as priori knowledge and classifying the track coordinates, the course angle and the vehicle type of the object to be detected as posterior knowledge.
Further, the second processing unit is configured to:
predicting track coordinates of the current moment of the track according to a first preset step for track information tracked and output by the previous frame;
and the current frame Lei Dadian cloud information endows the point cloud with course angles and vehicle type classification according to a second preset step.
Further, the second processing unit is configured to:
predicting the current position of the track according to the previous frame track and the uniform motion model;
the course angle of the track at the current moment directly uses the course angle of the track of the upper frame to obtain the track set at the current moment.
A storage medium comprising a stored program, wherein the program, when run, controls a device in which the storage medium resides to perform the millimeter wave Lei Dadian cloud clustering based method as described above.
An electronic device comprising at least one processor, and at least one memory, bus connected to the processor; the processor and the memory complete communication with each other through the bus; the processor is configured to invoke the program instructions in the memory to perform the millimeter wave Lei Dadian cloud based clustering method as described above.
According to the millimeter wave Lei Dadian cloud clustering method and device, the current frame Lei Dadian cloud information of the target to be detected and the track information tracked and output by the previous frame are obtained (the radar point cloud information comprises the position and the time stamp of the target to be detected, and the track information comprises coordinates, speeds, classifications and time stamps); then, according to a preset algorithm, the current frame Lei Dadian cloud information and the track information output by the previous frame tracking are associated to obtain track coordinates, course angles and vehicle type classification of the target to be detected; and finally, taking the elliptic wave gate as priori knowledge and classifying the track coordinates, the course angle and the vehicle type of the target to be detected as posterior knowledge, and carrying out point cloud clustering on the target to be detected. According to the embodiment of the application, the wave gate used for calculating the core object in the clustering is set as an ellipse according to the priori knowledge that the vehicle is rectangular, the point cloud sample obtains the course angle and the vehicle type classification through the associated track, and the direction and the size of the elliptical wave gate are changed through the course angle and the vehicle type classification, so that the clustering effect of the point cloud clustering is improved, and the target tracking quality is improved.
Drawings
In order to more clearly illustrate the embodiments of the application or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the application, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a schematic flow chart of a cloud clustering method based on millimeter waves Lei Dadian disclosed in the embodiment of the application;
fig. 2 is a schematic clustering diagram of a cloud clustering concrete implementation based on millimeter waves Lei Dadian disclosed in the embodiment of the application;
fig. 3 is a schematic structural diagram of a cloud clustering device based on millimeter waves Lei Dadian disclosed in the embodiment of the application;
fig. 4 is a schematic structural diagram of an electronic device according to an embodiment of the present application.
Detailed Description
The applicant finds that epsilon in the DBSCAN clustering algorithm in the prior art adopts a circular wave gate, and the circular wave gate can provide better robustness under the condition of unknown target attribute, but is not applicable when the vehicle point clouds are all long in a traffic scene; and the posterior knowledge of the course angle, classification and the like of the vehicle which are output by target tracking is not effectively utilized.
Therefore, aiming at the defects of a DBSCAN clustering algorithm in a traffic scene, the application provides a millimeter wave Lei Dadian cloud clustering method and device, and the aim is to: and the clustering effect of the point cloud clustering is improved, so that the target tracking quality is improved.
The following description of the embodiments of the present application will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present application, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.
Referring to fig. 1, a schematic flow chart of a cloud clustering method based on millimeter waves Lei Dadian is provided in an embodiment of the present application. As shown in fig. 1, the embodiment of the application provides a cloud clustering method based on millimeter waves Lei Dadian, which comprises the following steps:
s101: and acquiring current frame Lei Dadian cloud information of the target to be detected and track information tracked and output by the previous frame, wherein the radar point cloud information comprises the position and the time stamp of the target to be detected, and the track information comprises coordinates, speeds, classifications and time stamps.
In the embodiment of the application, the premise of carrying out point cloud clustering is that the current frame Lei Dadian cloud information of a target to be detected and track information tracked and output by the previous frame are required to be acquired, wherein the current frame radar point cloud information is point cloud information corresponding to the current frame in a point cloud sample set of the current frame output by a radar, and comprises information such as position (Cartesian coordinate system), time stamp and the like; the track information of the last frame of tracking output comprises coordinates, speeds, classifications and time stamps, wherein the coordinates and the speeds are respectively set into Cart (large-sized motor vehicles such as buses and large trucks), medium-sized vehicles (saloon cars, vans, SUVs, minibus and the like) and small vehicles (non-machine small targets such as people, battery cars, motorcycles, tricycles and the like) by Cart coordinate systems.
It should be noted that, according to the description of step 101, the input in the embodiment of the present application includes: this frame Lei Dadian cloud sample set d= { (o) i ，time cur )|o i ＝(x i ，y i )，i∈[1，...，m]-a }; track set for tracking and outputting upper frameWherein->Algorithm parameters (μ: number of samples clustered into clusters threshold, γ -circular wave gate radius, α c∈[1，2，3] Elliptic wave gate minor half-axis value (where α 1 ，α 2 ，α 3 Short half-shaft values of the large car, the medium car and the small car respectively), beta c∈[1，2，3] Elliptic wave gate major half axis value (where β 1 ，β 2 ，β 3 Long half axle values of the large car, the medium car and the small car respectively), and delta-sample is related to the distance threshold value of the track.
S102: and correlating the current frame Lei Dadian cloud information with the track information tracked and output by the previous frame according to a preset algorithm to obtain track coordinates, course angles and vehicle type classification of the target to be detected.
In the embodiment of the present application, the associating the current frame Lei Dadian cloud information with the track information output by the previous frame tracking according to a preset algorithm to obtain track coordinates, course angles and vehicle type classifications of the target to be detected includes: predicting track coordinates of the current moment of the track according to a first preset step for track information tracked and output by the previous frame; and the current frame Lei Dadian cloud information endows the point cloud with course angles and vehicle type classification according to a second preset step.
It should be noted that the first preset step includes: predicting the current position of the track according to the previous frame track and the uniform motion model; the course angle of the track at the current moment directly uses the course angle of the track of the upper frame to obtain the track set at the current moment.
It should be noted that the second preset step includes: determining a sample o according to a first preset formula i In the nearest neighbor track of the track set T', the first preset formula is thatWherein,,determining a sample o according to a second preset formula i The second preset formula is +.>Where inf is a maximum mnemonic, the sample may be from a radar echo of a new target; determining a sample o according to a third preset formula i The third preset formula is thatWhere 0 indicates no classification attribute and the sample may come from a radar echo of a new target.
S103: and carrying out point cloud clustering on the object to be detected by taking the elliptic wave gate as priori knowledge and taking the track coordinates, the course angle and the vehicle type of the object to be detected as posterior knowledge.
In the embodiment of the present application, the performing the point cloud clustering on the object to be detected by using the elliptic wave gate as the priori knowledge and classifying the track coordinates, the course angle and the vehicle type of the object to be detected as the posterior knowledge includes:
s1: taking an elliptic wave gate as priori knowledge and classifying track coordinates, course angles and vehicle types of the target to be detected as posterior knowledge to obtain a point cloud sample set D' = { (o) of comprehensive posterior knowledge i ，θ i ，c i )|o i ＝(x i ，y i )，i∈[1，...，m]}；
S2: initializing a core object set Ω=Φ, clustering cluster number k=0, unvisited sample set Γ=d', and cluster division c=Φ;
s3: for a point cloud sample set D' fused with posterior knowledge, finding out all core objects according to a preset step;
s4: if the core object set Ω=Φ, then the set c= { C is partitioned according to clusters 1 ，C 2 ，...，C k Samples in } calculate centroid coordinates for each clusterObtaining a cluster centroid coordinate set S= { S 1 ，s 2 ，...，s k }；
S5: if the core object set Ω is not equal to φ, randomly selecting one core object o from the core object set Ω, and initializing the current cluster core object queue Ω cur = { o }, initializing a class sequence number k=k+1, initializing a current cluster sample set C k = { o }, update the unvisited sample set Γ = Γ - { o };
s6: if the current cluster core object queue omega cur Cluster C =φ k After the generation is completed, updating cluster division C= { C 1 ，C 2 ，...，C k Update core object set Ω=Ω -C k S4, switching to S4; otherwise update the core object set Ω=Ω -C k ；
S7: at the current cluster core object queue Ω cur A core object o 'is taken out, a sample set N (o') in an o 'wave gate is found, delta=N (o') ∈Γ is set, and a current cluster sample set C is updated k ＝C k Update the unvisited sample set Γ=Γ - Δ,updating omega cur ＝Ω cur U (delta. U omega) -o', and S6;
s8: outputting the clustered cluster number k and a cluster centroid coordinate set S= { S 1 ，s 2 ，...，s k }。
In a specific embodiment, as shown in fig. 2, the output of the embodiment of the present application is the clustered cluster number k and the cluster centroid coordinate set S. The specific implementation mode is as follows:
1) For the upper frame track set T, j=1, 2,..n, the track current time coordinates and heading angle are predicted as follows.
a) According to the current position of the track predicted by the previous frame track according to the uniform motion model (because the radar frame interval is about 70 milliseconds, uniform motion can be assumed; of course other motion models may be selected), as follows:
b) The course angle of the track at the current moment directly uses the course angle of the track of the upper frame (the course angle of the upper frame can be approximately taken without great change of the course angle of the interval 70 milliseconds), and the specific expression is as follows:
obtaining a track set at the current moment
2) And for the point cloud sample set D, imparting posterior information such as heading angle and the like to the point cloud according to the following steps.
a) Find sample o i Nearest neighbor track in track set T
Wherein,,
b) Sample o i Course angle of (2)
Where inf is a maximum mnemonic, the sample may be from a radar echo of a new target.
c) Sample o i Classification of (2)
Where 0 indicates no classification attribute and the sample may come from a radar echo of a new target.
Point cloud sample set D' = { (o) for obtaining comprehensive posterior knowledge i ，θ i ，c i )|o i ＝(x i ，y i )，i∈[1，...，m]}。
3) Initializing a core object set Ω=Φ, clustering the number of clusters k=0, not accessing the sample set Γ=d', clustering the c=Φ.
4) For the point cloud sample set D' fusing posterior knowledge, all core objects are found as follows.
a) Find sample o i Is a sub-sample set N (o) i )
b) If the number of sub-sample set samples satisfies |N (o i ) I is not less than mu, sample o i Adding a core object sample set: Ω=Ω { o } i }。
5) If the core object set Ω=Φ, go to step 9), otherwise go to step 6).
6) In the core object set omega, randomly selecting one core object o and initializingCurrent cluster core object queue Ω cur = { o }, initializing a class sequence number k=k+1, initializing a current cluster sample set C k = { o }, update the unvisited sample set Γ = Γ - { o }.
7) If the current cluster core object queue omega cur Cluster C =φ k After the generation is completed, updating cluster division C= { C 1 ，C 2 ，...，C k Update core object set Ω=Ω -C k Turning to step 5); otherwise update the core object set Ω=Ω -C k 。
8) At the current cluster core object queue Ω cur A core object o 'is taken out, a sample set N (o') in an o 'wave gate is found, delta=N (o') ∈Γ is set, and a current cluster sample set C is updated k ＝C k Update non-accessed sample set Γ=Γ - Δ, update Ω cur ＝Ω cur U.S. (delta. U.OMEGA) -o', go to step 7).
9) Dividing the set c= { C according to the cluster 1 ，C 2 ，...，C k Samples in } calculate centroid coordinates for each clusterObtaining a cluster centroid coordinate set S= { S 1 ，s 2 ，...，s k }。
10 Outputting the clustered cluster number k and a cluster centroid coordinate set S= { S 1 ，s 2 ，...，s k }。
In the specific embodiment shown in fig. 2, the left graph is an image of a real environment, the middle is the clustering effect of the embodiment of the present application, and the right is the DBSCAN clustering effect. It should be noted that the present application tracks no maneuvering target, as some vehicles in fig. 2 have no radar point cloud data (the stationary point cloud has been filtered out) because of the equal red light, and the last remaining radar echo points have radial velocity. According to the technical scheme, priori knowledge (elliptic wave doors) and posterior knowledge (point cloud associated tracks are used for obtaining course angles and vehicle type classification) are added, and the technical scheme can be used for accurately clustering buses, cars and battery cars; while the right DBSCAN algorithm uses circular wave gates (circle radius e=2.5, minpoints=1) to find that the bus and one of the cars are clustered into two clusters, respectively.
By comparison, the clustering effect of the technical scheme of the application is better than that of DBSCAN, if epsilon of the DBSCAN is increased to solve the problem that one target is clustered into a plurality of clusters, but epsilon is too large to cause a plurality of small targets to be clustered into one cluster, for example, when the distance between a plurality of battery cars is smaller than 1 meter or vehicles running side by side when the vehicle passes a pedestrian crosswalk, a plurality of targets can be clustered into one cluster, so epsilon is not easy to set too large.
Although DBSCAN can cluster some targets of unknown properties, its drawbacks are also apparent in traffic scenarios. According to the technical scheme, priori knowledge and posterior knowledge (elliptic wave doors with different sizes are selected according to the vehicle types and the direction of the elliptic wave doors is changed according to the course angle) are fully used, and the problem that a large vehicle is clustered into a plurality of clusters or a plurality of small objects close to the large vehicle are clustered into one cluster can be well solved.
The embodiment of the application provides a millimeter wave Lei Dadian cloud clustering method, which is characterized in that the current frame Lei Dadian cloud information of a target to be detected and track information (the radar point cloud information comprises the position and time stamp of the target to be detected and the track information comprises coordinates, speed, classification and time stamp) which are tracked and output by the previous frame are obtained; then, according to a preset algorithm, the current frame Lei Dadian cloud information and the track information output by the previous frame tracking are associated to obtain track coordinates, course angles and vehicle type classification of the target to be detected; and finally, taking the elliptic wave gate as priori knowledge and classifying the track coordinates, the course angle and the vehicle type of the target to be detected as posterior knowledge, and carrying out point cloud clustering on the target to be detected. According to the embodiment of the application, the wave gate used for calculating the core object in the clustering is set as an ellipse according to the priori knowledge that the vehicle is rectangular, the point cloud sample obtains the course angle and the vehicle type classification through the associated track, and the direction and the size of the elliptical wave gate are changed through the course angle and the vehicle type classification, so that the clustering effect of the point cloud clustering is improved, and the target tracking quality is improved.
Referring to fig. 3, based on the millimeter wave Lei Dadian cloud clustering method disclosed in the above embodiment, the present embodiment correspondingly discloses a millimeter wave Lei Dadian cloud clustering device, which includes:
the first processing unit 301 is configured to obtain current frame Lei Dadian cloud information of a target to be detected and track information tracked and output by a previous frame, where the radar point cloud information includes a position and a timestamp of the target to be detected, and the track information includes coordinates, a speed, a classification and a timestamp;
the second processing unit 302 is configured to correlate, according to a preset algorithm, the current frame Lei Dadian cloud information with the track information tracked and output by the previous frame, to obtain track coordinates, a course angle, and a vehicle type classification of the target to be detected;
and the third processing unit 303 is configured to perform point cloud clustering on the target to be detected by using the elliptic wave gate as a priori knowledge and classifying the track coordinates, the course angle and the vehicle type of the target to be detected as posterior knowledge.
Further, the second processing unit 302 is configured to:
predicting track coordinates of the current moment of the track according to a first preset step for track information tracked and output by the previous frame;
and the current frame Lei Dadian cloud information endows the point cloud with course angles and vehicle type classification according to a second preset step.
Further, the second processing unit 302 is configured to:
predicting the current position of the track according to the previous frame track and the uniform motion model;
the course angle of the track at the current moment directly uses the course angle of the track of the upper frame to obtain the track set at the current moment.
The millimeter wave Lei Dadian cloud clustering device comprises a processor and a memory, wherein the first processing unit, the second processing unit, the third processing unit and the like are stored in the memory as program units, and the processor executes the program units stored in the memory to realize corresponding functions.
The processor includes a kernel, and the kernel fetches the corresponding program unit from the memory. The kernel can be provided with one or more than one kernel, and the clustering effect of the point cloud clusters is improved by adjusting kernel parameters, so that the aim of improving the target tracking quality is fulfilled.
The embodiment of the application provides a storage medium, wherein a program is stored on the storage medium, and the program is executed by a processor to realize the cloud clustering method based on millimeter waves Lei Dadian.
The embodiment of the application provides a processor which is used for running a program, wherein the millimeter wave Lei Dadian cloud clustering method is executed when the program runs.
An embodiment of the present application provides an electronic device, as shown in fig. 4, where the electronic device 40 includes at least one processor 401, and at least one memory 402 and a bus 403 connected to the processor; wherein the processor 401 and the memory 402 complete communication with each other through the bus 403; the processor 401 is configured to invoke the program instructions in the memory 402 to execute the above-mentioned cloud clustering method based on millimeter waves Lei Dadian.
The electronic device herein may be a server, a PC, a PAD, a mobile phone, etc.
The application also provides a computer program product adapted to perform, when executed on a data processing device, a program initialized with the method steps of:
acquiring current frame Lei Dadian cloud information of a target to be detected and track information tracked and output by the previous frame, wherein the radar point cloud information comprises positions and time stamps of the target to be detected, and the track information comprises coordinates, speeds, classifications and time stamps;
correlating the current frame Lei Dadian cloud information with the track information tracked and output by the previous frame according to a preset algorithm to obtain track coordinates, course angles and vehicle type classification of the target to be detected;
and carrying out point cloud clustering on the object to be detected by taking the elliptic wave gate as priori knowledge and taking the track coordinates, the course angle and the vehicle type of the object to be detected as posterior knowledge.
Further, the associating the current frame Lei Dadian cloud information with the track information output by the previous frame tracking according to the preset algorithm to obtain track coordinates, a course angle and a vehicle type classification of the target to be detected includes:
predicting track coordinates of the current moment of the track according to a first preset step for track information tracked and output by the previous frame;
and the current frame Lei Dadian cloud information endows the point cloud with course angles and vehicle type classification according to a second preset step.
Further, the first preset step includes:
predicting the current position of the track according to the previous frame track and the uniform motion model;
the course angle of the track at the current moment directly uses the course angle of the track of the upper frame to obtain the track set at the current moment.
Further, the second preset step includes:
determining a sample o according to a first preset formula i In the nearest neighbor track of the track set T', the first preset formula is thatWherein (1)>
Determining a sample o according to a second preset formula i The second preset formula is thatWhere inf is a maximum mnemonic, the sample may be from a radar echo of a new target;
determining a sample o according to a third preset formula i The third preset formula is thatWhere 0 indicates no classification attribute and the sample may come from a radar echo of a new target.
Further, the classifying the to-be-detected target with the elliptic wave gate as a priori knowledge and the track coordinates, the course angle and the vehicle type of the to-be-detected target as posterior knowledge, includes:
s1: taking an elliptic wave gate as priori knowledge and classifying track coordinates, course angles and vehicle types of the target to be detected as posterior knowledge to obtain a point cloud sample set D' = { (o) of comprehensive posterior knowledge i ，θ i ，c i )|o i ＝(x i ，y i )，i∈[1，...，m]}；
S2: initializing a core object set Ω=Φ, clustering cluster number k=0, unvisited sample set Γ=d', and cluster division c=Φ;
s3: for a point cloud sample set D' fused with posterior knowledge, finding out all core objects according to a preset step;
s4: if the core object set Ω=Φ, then the set c= { C is partitioned according to clusters 1 ，C 2 ，...，C k Samples in } calculate centroid coordinates for each clusterObtaining a cluster centroid coordinate set S= { S 1 ，s 2 ，...，s k }；
S5: if the core object set Ω is not equal to φ, randomly selecting one core object o from the core object set Ω, and initializing the current cluster core object queue Ω cur = { o }, initializing a class sequence number k=k+1, initializing a current cluster sample set C k = { o }, update the unvisited sample set Γ = Γ - { o };
s6: if the current cluster core object queue omega cur Cluster C =φ k After the generation is completed, updating cluster division C= { C 1 ，C 2 ，...，C k Update core object set Ω=Ω -C k S4, switching to S4; otherwise update the core object set Ω=Ω -C k ；
S7: at the current cluster core object queue Ω cur A core object o ' is taken out, a sample set N (o ') in a 0' wave gate is found out, and the following steps are carried outDelta=n (o')Γ, updating the current cluster sample set C k =C k Update non-accessed sample set Γ=Γ - Δ, update Ω cur ＝Ω cur U (delta. U omega) -o', and S6;
s8: outputting the clustered cluster number k and a cluster centroid coordinate set S= { S 1 ，s 2 ，...，s k }。
The present application is described in terms of methods, apparatus (systems), computer program products, flowcharts, and/or block diagrams in accordance with embodiments of the present application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
In one typical configuration, the device includes one or more processors (CPUs), memory, and a bus. The device may also include input/output interfaces, network interfaces, and the like.
The memory may include volatile memory, random Access Memory (RAM), and/or nonvolatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM), among other forms in computer readable media, the memory including at least one memory chip. Memory is an example of a computer-readable medium.
Computer readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of storage media for a computer include, but are not limited to, phase change memory (PRAM), static Random Access Memory (SRAM), dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium, which can be used to store information that can be accessed by a computing device. Computer-readable media, as defined herein, does not include transitory computer-readable media (transmission media), such as modulated data signals and carrier waves.
It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises an element.
It will be appreciated by those skilled in the art that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The foregoing is merely exemplary of the present application and is not intended to limit the present application. Various modifications and variations of the present application will be apparent to those skilled in the art. Any modification, equivalent replacement, improvement, etc. which come within the spirit and principles of the application are to be included in the scope of the claims of the present application.
Claims (10)
1. The cloud clustering method based on the millimeter waves Lei Dadian is characterized by comprising the following steps of:
acquiring current frame Lei Dadian cloud information of a target to be detected and track information tracked and output by the previous frame, wherein the radar point cloud information comprises positions and time stamps of the target to be detected, and the track information comprises coordinates, speeds, classifications and time stamps;
correlating the current frame Lei Dadian cloud information with the track information tracked and output by the previous frame according to a preset algorithm to obtain track coordinates, course angles and vehicle type classification of the target to be detected;
and taking an elliptic wave gate as priori knowledge and classifying the track coordinates, the course angle and the vehicle type of the target to be detected as posterior knowledge to obtain a point cloud sample set of comprehensive posterior knowledge, finding out a core object in the point cloud sample set, and then carrying out cluster division on the point cloud in the point cloud sample set according to the core object to obtain a cluster centroid coordinate set so as to carry out point cloud clustering on the target to be detected.
2. The method of claim 1, wherein associating the current frame Lei Dadian cloud information with the track information output by the previous frame tracking according to a preset algorithm to obtain track coordinates, a course angle and a vehicle type classification of the target to be detected comprises:
predicting track coordinates of the current moment of the track according to a first preset step for track information tracked and output by the previous frame;
and the current frame Lei Dadian cloud information endows the point cloud with course angles and vehicle type classification according to a second preset step.
3. The method according to claim 2, wherein the first preset step comprises:
predicting the current position of the track according to the previous frame track and the uniform motion model;
the course angle of the track at the current moment directly uses the course angle of the track of the upper frame to obtain the track set at the current moment.
4. The method according to claim 2, wherein the second preset step comprises:
determining a sample o according to a first preset formula i In the nearest neighbor track of the track set T', the first preset formula is thatWherein (1)>Delta is the distance threshold of the sample associated track;
determining a sample o according to a second preset formula i The second preset formula is thatWhere inf is a maximum mnemonic, the sample may be from a radar echo of a new target;
determining a sample o according to a third preset formula i The third preset formula is thatWhere 0 indicates no classification attribute and the sample may come from a radar echo of a new target.
5. The method according to claim 4, wherein the classifying the track coordinates, the course angle and the vehicle type of the object to be detected with elliptic wave gate as a priori knowledge and with the track coordinates, the course angle and the vehicle type of the object to be detected as a posterior knowledge, obtaining a point cloud sample set of comprehensive posterior knowledge, finding out a core object in the point cloud sample set, and then performing cluster division on point clouds in the point cloud sample set according to the core object, to obtain a cluster centroid coordinate set, so as to perform point cloud clustering on the object to be detected, includes:
s1: taking an elliptic wave gate as priori knowledge and classifying track coordinates, course angles and vehicle types of the target to be detected as posterior knowledge to obtain a point cloud sample set D' = { (o) of comprehensive posterior knowledge i ,θ i ,c i )|o i ＝(x i ,y i ),i∈[1,…,m]}；
S2: initializing a core object set Ω=Φ, clustering cluster number k=0, unvisited sample set Γ=d', and cluster division c=Φ;
s3: for a point cloud sample set D' fused with posterior knowledge, finding out all core objects according to a preset step;
s4: if the core object set Ω=Φ, then the set c= { C is partitioned according to clusters 1 ,C 2 ,…,C k Samples in } calculate centroid coordinates for each clusterObtaining a cluster centroid coordinate set S= { S 1 ,s 2 ,…,s k }；
S5: if the core object set Ω is not equal to φ, randomly selecting one core object o from the core object set Ω, and initializing the current cluster core object queue Ω cur = { o }, initializing a class sequence number k=k+1, initializing a current cluster sample set C k = { o }, update the unvisited sample set Γ = Γ - { o };
s6: if the current cluster core object queue omega cur Cluster C =φ k After the generation is completed, updating cluster division C= { C 1 ,C 2 ,…,C k Update core object set Ω=Ω -C k S4, switching to S4; otherwise update the core object set Ω=Ω -C k ；
S7: at the current cluster core object queue Ω cur A core object o 'is taken out, a sample set N (o') in an o 'wave gate is found, delta=N (o') ∈Γ is set, and a current cluster sample set C is updated k ＝C k Update non-accessed sample set Γ=Γ - Δ, update Ω cur ＝Ω cur U (delta. U omega) -o', and S6;
s8: outputting the clustered cluster number k and a cluster centroid coordinate set S= { S 1 ,s 2 ,…,s k }。
6. Millimeter wave Lei Dadian cloud clustering party based device, which is characterized by comprising:
the first processing unit is used for acquiring current frame Lei Dadian cloud information of a target to be detected and track information tracked and output by the previous frame, wherein the radar point cloud information comprises positions and time stamps of the target to be detected, and the track information comprises coordinates, speeds, classifications and time stamps;
the second processing unit is used for associating the current frame Lei Dadian cloud information with the track information tracked and output by the previous frame according to a preset algorithm to obtain track coordinates, course angles and vehicle type classification of the target to be detected;
and the third processing unit is used for taking an elliptic wave gate as priori knowledge and classifying the track coordinates, the course angle and the vehicle type of the target to be detected as posterior knowledge to obtain a point cloud sample set of comprehensive posterior knowledge, finding out a core object in the point cloud sample set, and then dividing the point cloud in the point cloud sample set into clusters according to the core object to obtain a cluster centroid coordinate set so as to perform point cloud clustering on the target to be detected.
7. The apparatus of claim 6, wherein the second processing unit is configured to:
predicting track coordinates of the current moment of the track according to a first preset step for track information tracked and output by the previous frame;
and the current frame Lei Dadian cloud information endows the point cloud with course angles and vehicle type classification according to a second preset step.
8. The apparatus of claim 7, wherein the second processing unit is configured to:
predicting the current position of the track according to the previous frame track and the uniform motion model;
the course angle of the track at the current moment directly uses the course angle of the track of the upper frame to obtain the track set at the current moment.
9. A storage medium comprising a stored program, wherein the program, when run, controls a device in which the storage medium is located to perform the millimeter wave Lei Dadian cloud-based clustering method of any one of claims 1 to 5.
10. An electronic device comprising at least one processor, and at least one memory, bus coupled to the processor; the processor and the memory complete communication with each other through the bus; the processor is configured to invoke program instructions in the memory to perform the millimeter wave Lei Dadian cloud based clustering method of any one of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111447566.5A CN114137509B (en) | 2021-11-30 | 2021-11-30 | Millimeter wave Lei Dadian cloud clustering method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111447566.5A CN114137509B (en) | 2021-11-30 | 2021-11-30 | Millimeter wave Lei Dadian cloud clustering method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114137509A CN114137509A (en) | 2022-03-04 |
| CN114137509B true CN114137509B (en) | 2023-10-13 |
Family
ID=80386321
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111447566.5A Active CN114137509B (en) | 2021-11-30 | 2021-11-30 | Millimeter wave Lei Dadian cloud clustering method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114137509B (en) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114779230B (en) * | 2022-03-22 | 2024-03-08 | 福瑞泰克智能系统有限公司 | Object tracking method and device, electronic equipment and vehicle |
| CN114523902B (en) * | 2022-03-30 | 2023-08-15 | 深蓝汽车科技有限公司 | Collision early warning method and system for front vehicle door opening scene |
| CN115236674B (en) * | 2022-06-15 | 2024-06-04 | 北京踏歌智行科技有限公司 | Mining area environment sensing method based on 4D millimeter wave radar |
| CN115113205B (en) * | 2022-07-07 | 2023-10-20 | 南京慧尔视智能科技有限公司 | Holographic image method and device for road, electronic equipment and storage medium |
| CN116721246B (en) * | 2023-07-14 | 2024-03-19 | 酷哇科技有限公司 | Continuous frame point cloud rapid labeling method and system |
| CN117111053B (en) * | 2023-10-23 | 2024-02-27 | 武汉未来幻影科技有限公司 | Millimeter wave radar tracking result processing method, device and processing equipment |
| CN117111019B (en) * | 2023-10-25 | 2024-01-09 | 深圳市先创数字技术有限公司 | Target tracking and monitoring method and system based on radar detection |
| CN117647807B (en) * | 2024-01-30 | 2024-04-19 | 安徽隼波科技有限公司 | Motor vehicle size estimation method based on millimeter wave radar |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110361727A (en) * | 2019-07-22 | 2019-10-22 | 浙江大学 | A kind of millimetre-wave radar multi-object tracking method |
| CN110658531A (en) * | 2019-08-23 | 2020-01-07 | 畅加风行(苏州)智能科技有限公司 | Dynamic target tracking method for port automatic driving vehicle |
| CN112561971A (en) * | 2020-12-16 | 2021-03-26 | 珠海格力电器股份有限公司 | People flow statistical method, device, equipment and storage medium |
| CN113269889A (en) * | 2021-06-04 | 2021-08-17 | 重庆大学 | Self-adaptive point cloud target clustering method based on elliptical domain |
| CN113281735A (en) * | 2021-05-19 | 2021-08-20 | 南京慧尔视智能科技有限公司 | Method, device and system for improving target tracking performance of millimeter wave radar and storage medium |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11062454B1 (en) * | 2019-04-16 | 2021-07-13 | Zoox, Inc. | Multi-modal sensor data association architecture |
| GB2590115B (en) * | 2019-09-13 | 2023-12-06 | Motional Ad Llc | Extended object tracking using radar |
-
2021
- 2021-11-30 CN CN202111447566.5A patent/CN114137509B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110361727A (en) * | 2019-07-22 | 2019-10-22 | 浙江大学 | A kind of millimetre-wave radar multi-object tracking method |
| CN110658531A (en) * | 2019-08-23 | 2020-01-07 | 畅加风行(苏州)智能科技有限公司 | Dynamic target tracking method for port automatic driving vehicle |
| CN112561971A (en) * | 2020-12-16 | 2021-03-26 | 珠海格力电器股份有限公司 | People flow statistical method, device, equipment and storage medium |
| CN113281735A (en) * | 2021-05-19 | 2021-08-20 | 南京慧尔视智能科技有限公司 | Method, device and system for improving target tracking performance of millimeter wave radar and storage medium |
| CN113269889A (en) * | 2021-06-04 | 2021-08-17 | 重庆大学 | Self-adaptive point cloud target clustering method based on elliptical domain |
Non-Patent Citations (2)
| Title |
|---|
| Robust and Adaptive Radar Elliptical Density-Based Spatial Clustering and labeling for mmWave Radar Point Cloud Data;Renyuan Zhang et al.;《2019 53rd Asilomar Conference on Signals, Systems and Computer》;1-4 * |
| 智能车辆前方机动目标的运动状态识别方法研究;刘凡;《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》;41-44 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114137509A (en) | 2022-03-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114137509B (en) | Millimeter wave Lei Dadian cloud clustering method and device | |
| US9870512B2 (en) | Lidar-based classification of object movement | |
| JP2021523443A (en) | Association of lidar data and image data | |
| US11164051B2 (en) | Image and LiDAR segmentation for LiDAR-camera calibration | |
| CN111091023B (en) | Vehicle detection method and device and electronic equipment | |
| US11604272B2 (en) | Methods and systems for object detection | |
| WO2023273467A1 (en) | True value data determination method and apparatus, neural network training method and apparatus, and travel control method and apparatus | |
| CN113536920A (en) | Semi-supervised three-dimensional point cloud target detection method | |
| CN116432060A (en) | Target self-adaptive clustering method, device, equipment and storage medium based on radar | |
| Arunpriyan et al. | Real-time speed bump detection using image segmentation for autonomous vehicles | |
| Lim et al. | Gaussian process auto regression for vehicle center coordinates trajectory prediction | |
| Billones et al. | Visual Percepts Quality Recognition Using Convolutional Neural Networks | |
| US11698437B2 (en) | Segmentation and classification of point cloud data | |
| Jiang et al. | An obstacle detection and distance measurement method for sloped roads based on VIDAR | |
| JP7564329B2 (en) | Labeling information determination method and apparatus | |
| Bratulescu et al. | Object Detection in Autonomous Vehicles | |
| CN115320581B (en) | Real-time vehicle lane change risk assessment method, computer equipment and computer storage medium | |
| CN115457274A (en) | Vehicle-mounted view angle shielding target detection method and device based on deep learning | |
| Wang et al. | A 64‐Line Lidar‐Based Road Obstacle Sensing Algorithm for Intelligent Vehicles | |
| Fu et al. | Behavior analysis of distant vehicles using LIDAR point cloud | |
| CN115439484B (en) | Detection method and device based on 4D point cloud, storage medium and processor | |
| CN113452952B (en) | Road condition monitoring method, device and system | |
| Bencheriet et al. | Vehicle tracking and trajectory estimation for detection of traffic road violation | |
| Luo et al. | Object detection based on binocular vision with convolutional neural network | |
| Seo et al. | A Spatiotemporal Deep Learning Architecture for Road Surface Classification Using LiDAR in Autonomous Emergency Braking Systems |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| PE01 | Entry into force of the registration of the contract for pledge of patent right | ||
| PE01 | Entry into force of the registration of the contract for pledge of patent right |
Denomination of invention: A point cloud clustering method and device based on millimeter wave radar Granted publication date: 20231013 Pledgee: Bank of China Limited by Share Ltd. Nanjing Jiangning branch Pledgor: NANJING HURYS INTELLIGENT TECHNOLOGY Co.,Ltd. Registration number: Y2024980010482 |