Large target detection method, large target detection device and electronic equipment

Abstract

The present application discloses a large target detection method, a large target detection device, an electronic device, and a computer-readable storage medium. The method comprises: obtaining a point cloud to be processed, wherein the point cloud to be processed is obtained by detecting a target area through radar; clustering the point cloud to be processed to determine at least one target point set; screening out a large target point set from the at least one target point set; for each large target point set, determining a to-be-searched range of the large target point set based on target attribute information corresponding to the large target point set; searching for target splitting points within the to-be-searched range, wherein the target splitting points are points that are related to the large target point set but are not classified into the large target point set; and classifying the target splitting points into the large target point set. The present application solution can improve the radar's detection accuracy for large targets.

Description

Technical Field

[0001] The present application belongs to the field of information processing technology, and in particular relates to a large target detection method, a large target detection device, an electronic device, and a computer-readable storage medium. Background Art

[0002] When detecting a target that's too large, the radar will detect multiple reflection points. Because these reflection points are at different angles relative to the radar, the radial velocities measured by the radar vary. Consequently, the radar may interpret these reflection points as belonging to different targets. This can cause a large target to be mistakenly detected as multiple smaller ones, affecting detection effectiveness. Summary of the Invention

[0003] The present application provides a large target detection method, a large target detection device, an electronic device, and a computer-readable storage medium, which can improve the detection accuracy of radar for large targets.

[0004] In a first aspect, the present application provides a large target detection method, comprising:

[0005] Obtaining a point cloud to be processed, wherein the point cloud to be processed is obtained by detecting a target area with a radar;

[0006] Cluster the point cloud to be processed and determine at least one target point set;

[0007] In at least one target point set, a large target point set is screened out;

[0008] For each large target point set, the search range of the large target point set is determined according to the target attribute information corresponding to the large target point set;

[0009] Find the target split point in the search range, where the target split point is a point that is related to the large target point set but is not classified into the large target point set;

[0010] Classify the target split points into a large target point set.

[0011] In a second aspect, the present application provides a large target detection device, comprising:

[0012] An acquisition module is used to acquire a point cloud to be processed, wherein the point cloud to be processed is obtained by detecting a target area with a radar;

[0013] A first determination module is used to cluster the point cloud to be processed and determine at least one target point set;

[0014] A screening module, configured to screen out a large target point set from at least one target point set;

[0015] The second determining module is used to determine the to-be-searched range of the large target point set according to the target attribute information corresponding to the large target point set for each large target point set;

[0016] A search module is used to search for a target splitting point within the search range, wherein the target splitting point is a point that is related to the large target point set but is not classified into the large target point set;

[0017] The classification module is used to classify the target split points into a large target point set.

[0018] In a third aspect, the present application provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method of the first aspect when executing the computer program.

[0019] In a fourth aspect, the present application provides a computer-readable storage medium storing a computer program, which implements the steps of the method of the first aspect when executed by a processor.

[0020] In a fifth aspect, the present application provides a computer program product, which includes a computer program, and when the computer program is executed by one or more processors, it implements the steps of the method of the first aspect.

[0021] Compared with the prior art, the present application has the following advantages: the electronic device first obtains a point cloud to be processed, wherein the point cloud to be processed is obtained by detecting the target area through radar, and then clusters the point cloud to be processed to determine at least one target point set. This process realizes clustering at the point cloud data level. After that, the electronic device can filter out a large target point set from the at least one target point set, and for each large target point set, determine the search range of the large target point set based on the target attribute information corresponding to the large target point set, and then search for target splitting points within the search range, wherein the target splitting points are points related to the large target point set but not classified into the large target point set, and finally classify the target splitting points into the large target point set. This process realizes re-clustering at the target data level. Through clustering at the point cloud data level and re-clustering at the target data level, it can be ensured that the points detected by the radar for the same large target can be classified into the same large target point set as much as possible, reducing the situation where a large target is detected as multiple small targets.

[0022] It can be understood that the beneficial effects of the second to fifth aspects mentioned above can be found in the relevant description of the first aspect mentioned above, and will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the embodiments or descriptions of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0024] Figure 1 This is a schematic diagram of the implementation process of the large target detection method provided in the embodiment of the present application;

[0025] Figure 2 is an example diagram of a sub-range to be searched provided in an embodiment of the present application;

[0026] Figure 3 : is an example diagram of the deployment of various devices involved in the large target detection method provided in the embodiment of the present application;

[0027] Figure 4 This is a structural block diagram of a large target detection device provided in an embodiment of the present application;

[0028] Figure 5 It is a structural diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0029] In the following description, specific details such as specific system structures and techniques are provided for purposes of illustration rather than limitation to facilitate a thorough understanding of the embodiments of the present application. However, it will be apparent to those skilled in the art that the present application may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid obscuring the description of the present application with unnecessary detail.

[0030] In order to illustrate the technical solution proposed in this application, a specific embodiment is provided below.

[0031] It is understood that when applying the large target splicing method proposed in the embodiments of the present application, the target can be a vehicle or other objects. As an example only, the other object can be an animal or a means of transportation other than a vehicle, and the embodiments of the present application do not limit the type of the target. In the case where the target is a vehicle, the large target refers to a vehicle larger than a preset size, that is, a large vehicle. The embodiments of the present application mainly use the vehicle as an example to illustrate the proposed large target detection method.

[0032] See also Figure 1 , the implementation process of the large target detection method is detailed as follows:

[0033] Step 101: Obtain the point cloud to be processed.

[0034] Radars can be deployed on the road to detect vehicles passing through the coverage area of ​​their signals. As an example only, on highways, radars can be deployed on gantry frames. Taking into account the embodiments of the present application, the equipment used to detect vehicles is not limited to radars. Therefore, the common coverage area of ​​other equipment and radars is recorded as the target area. It can be understood that the deployment posture of the radar is known. For the signal emitted by the radar, different signal transmission (reception) angles correspond to different areas. Based on this, a large amount of raw radar data can be obtained by detecting the target area through the radar. After these raw radar data are uploaded to the electronic equipment, they are processed by the electronic equipment through one-dimensional fast FFT (range dimension FFT), two-dimensional FFT (Doppler FFT), CFAR and three-dimensional FFT (angle dimension FFT) and other related processing to obtain a point cloud to be processed. Each point in the point cloud to be processed contains information such as its radial distance and radial velocity relative to the radar.

[0035] The radar can also upload its radar cross section (RCS) to the electronic equipment. It can be understood that based on the data obtained by the radar, the electronic equipment can perform subsequent operations.

[0036] Step 102: cluster the point cloud to be processed to determine at least one target point set.

[0037] Since the vehicle is only traveling on the road, that is, it can be approximately regarded as the vehicle will only generate displacement in the horizontal plane in a short period of time, and will not generate displacement in the vertical direction. Based on this, for each point detected by the radar in the embodiment of the present application, the electronic device actually focuses on the position of each point projected on the horizontal plane; that is, the height of these points in the vertical direction can be ignored in the embodiment of the present application. It can be considered that the operations performed by the electronic device on each point in the point cloud to be processed in the embodiment of the present application are all based on the processing angle of looking down; that is, the electronic device performs a series of operations such as clustering, searching and re-clustering on the point cloud to be processed in a horizontal two-dimensional coordinate system.

[0038] To reduce the possibility of a large target being detected as multiple small targets, electronic devices can use the adaptive radius DBSCAN (Density-Based Spatial Clustering of Applications with Noise) clustering algorithm to cluster the point cloud to be processed. After clustering is completed, at least one target point set is obtained. It can be understood that the meaning of this target point set is that all points contained in a target point set have a high probability of belonging to the same target. Through this clustering algorithm, an appropriate clustering radius can be selected for clustering, making the clustering result (that is, the target point set obtained by clustering) more accurate, ensuring the accuracy of large target detection at the point cloud data level.

[0039] Step 103: Filter out a large target point set from at least one target point set.

[0040] As described above, in addition to radar, other devices may also detect vehicles on the road. For example, the other device may be a camera or a roadside unit (RSU). The present application does not limit the type of the other device. It should be noted that the other target and the radar must be synchronized before use to avoid discrepancies caused by asynchronous reporting of data.

[0041] In some application scenarios, when the other device is a camera: first, the camera can capture the image within the target area and upload the image to the electronic device. Then, the electronic device analyzes each target (i.e., each vehicle) in the image to obtain the target attribute information (i.e., vehicle attribute information) of each target. Finally, the electronic device matches and fuses the target attribute information of each target with each target point set, and can determine which target point sets represent large targets (i.e., large vehicles). These target point sets determined to represent large targets are the large target point sets.

[0042] In other application scenarios, when the other device is a roadside unit (RSU), the RSU first communicates with the on-board units (OBUs) within the target area and obtains the target attribute information (i.e., vehicle attribute information) of the corresponding targets (i.e., vehicles) reported by each OBU. The RSU then uploads this target attribute information to the electronic device. Finally, the electronic device matches and fuses the received target attribute information with each target point set to determine which target point sets represent large targets (i.e., large vehicles). These target point sets determined to represent large targets are referred to as large target point sets.

[0043] As an example only, when the target is a vehicle, the target attribute information (ie, vehicle attribute information) may include vehicle license plate, vehicle position, vehicle model, vehicle length, vehicle width, etc., which are not limited here.

[0044] Step 104 : For each large target point set, determine the to-be-searched range of the large target point set according to the target attribute information corresponding to the large target point set.

[0045] The clustering operation performed in step 102 may result in incorrect clustering, resulting in points that should actually belong to a large target point set failing to be classified into the large target point set through step 102. In order to improve the accuracy of detection, the electronic device may consider clustering the points again at the target data level so that some misclassified points near the large target point set (that is, points that originally belonged to the large target point set but were not classified into the large target point set) can be reclassified into the large target point set. Considering that the size of the target is limited, even if points are misclassified, these misclassified points will not be too far away from the large target point set to which they should belong. Based on this, in order to improve processing efficiency and avoid the electronic device from re-clustering points within an unlimited range, in an embodiment of the present application, the electronic device can determine the search range of the large target point set based on the target attribute information corresponding to the large target point set.

[0046] Step 105: Find the target splitting point within the search range.

[0047] After searching within the range to be searched, the target splitting points can be obtained. It can be understood that in an ideal scenario, these target splitting points and the points in the large target point set should all be points obtained by detecting the same large target. However, in actual scenarios, due to reasons such as misclustering, it is impossible to completely ensure that these target splitting points and the points in the large target point set are all points obtained by detecting the same large target. Based on this, in an embodiment of the present application, the electronic device may consider that these target splitting points and the points in the large target point set are very likely to be points obtained by detecting the same large target, and the target splitting point may be defined as: a point that is related to the large target point set but is not classified into the large target point set.

[0048] Step 106: Classify the target split points into a large target point set.

[0049] Through steps 104-106, each large target point set can find a corresponding target splitting point. That is, for any large target point set, the electronic device can find the target splitting point corresponding to that large target point set. The electronic device can then add the target splitting point corresponding to that large target point set to the large target point set, thereby achieving re-clustering at the target data level. In actual application scenarios, since each large target point set represents a large vehicle, by tracking each large target point set at different times, the electronic device can track the trajectory of each large vehicle within the target area.

[0050] It can be understood that through clustering at the point cloud data level and re-clustering at the target data level, it can be ensured that the points detected by the radar for the same large target can be classified into the same large target point set as much as possible, reducing the situation where a large target is detected as multiple small targets.

[0051] In some embodiments, since the angular resolution of the radar is fixed, there is non-uniformity in the radar data. On this basis, since the reflection intensity, shape and volume of different targets may be different, the point cloud density obtained when the radar detects different targets is also different. It can be seen that the point cloud density of a large target is usually different from the point cloud density of other targets (ie small targets). Based on the above premise, the embodiment of the present application proposes to use the DBSCAN clustering algorithm with an adaptive radius, and improve the accuracy of clustering by adjusting the clustering radius adopted by the DBSCAN clustering algorithm, and cluster the point cloud of large targets into one category as much as possible. The following is a detailed description of the DBSCAN clustering algorithm with an adaptive radius:

[0052] The DBSCAN clustering algorithm has two parameters: cluster radius E ps Sum point threshold min ps The adaptive radius DBSCAN clustering algorithm proposed in the embodiment of the present application refers to adjusting the clustering radius E ps , so that it can achieve better large target clustering effect.

[0053] Among them, the cluster radius E ps It can be calculated using the following formula:

[0054] E ps =αE ps,k +βE ps,RCS (1)

[0055] In the above formula, E ps is the clustering radius used for a core point; E ps,k is the radius based on the point cloud interval distance, which can be calculated by k-nearest neighbor distance; E ps,RCSThe radius is based on the RCS reported by the radar and can be obtained by converting the RCS. α and β are weight parameters that satisfy 0≤α, β≤1 and α+β=1. The specific values ​​can be determined according to the actual application scenario. It can be understood that if, in the actual application scenario, the RCS has a greater impact on the cluster radius and the point cloud spacing has a smaller impact on the cluster radius, then β can be taken as a larger value and α can be taken as a smaller value; conversely, if, in the actual application scenario, the RCS has a smaller impact on the cluster radius and the point cloud spacing has a greater impact on the cluster radius, then β can be taken as a smaller value and α can be taken as a larger value.

[0056] Specifically, E can be solved as follows ps,k :

[0057] First, the points in the point cloud to be processed are simply classified by statistical means to obtain multiple suspected targets, each of which has multiple points. The process can be briefly described as follows: large targets and small targets at various distances are detected in the test scenario to obtain the typical value of the point cloud interval distance of the large target and the typical value of the point cloud interval distance of the small target; then, based on the typical value of the point cloud interval distance of the large target and the typical value of the point cloud interval distance of the small target, the points in the point cloud to be processed are classified to obtain multiple suspected targets.

[0058] Then, perform the following operations on any suspected target:

[0059] Analyze the suspected target and obtain the number of points contained in the suspected target; based on the number, determine the k parameter to be set in the k nearest neighbor distance; based on the k parameter, calculate the distance between the suspected target and its kth nearest neighbor point; compare the distance with the set judgment threshold to calculate the E of the suspected target. ps,k , which can be expressed as follows:

[0060]

[0061] Wherein, γ1 is a preset first distance threshold, γ2 is a preset second distance threshold, γ3 is a preset third distance threshold, and γ1, γ2 and γ3 are not equal to each other; x k is the distance between the suspected target and its kth adjacent point; ε1 is the preset first judgment threshold, ε2 is the preset second judgment threshold, and ε1<ε2.

[0062] It can be understood that a corresponding E can be solved for each suspected target ps,k .

[0063] Specifically, E can be solved as follows ps,RCS :

[0064] E ps,RCS =rRCS (3)

[0065] Among them, r RCS It is an imaginary radius, which can be calculated based on the RCS reported by the radar.

[0066] Based on the above formulas (1), (2) and (3), each suspected target obtains the corresponding E ps , that is, E ps Target-based adaptation is achieved. Generally speaking, large targets usually have relatively large E ps , small targets usually have relatively small E ps In this way, when performing clustering, electronic devices can actively distinguish different targets based on the adaptive clustering radius, thereby suppressing or reducing the occurrence of misclustering of large targets at the source.

[0067] In some embodiments, for any large target point set, the electronic device may determine a suitable to-be-searched range for the large target point set in the following manner:

[0068] A1. Determine the target point in the large target point set.

[0069] When the electronic device matches and fuses the target attribute information with the target point set in step 103, it can attach the target attribute information to a random point in the matched target point set, and this point can be determined as the destination target point of the large target point set. Of course, the electronic device can also attach the target attribute information to the center point of the matched target point set, that is, the electronic device can determine the center point of the large target point set as the destination target point. In this embodiment of the present application, there is no limitation on the destination target point.

[0070] A2. Determine the search range with the target point as the center and the specified multiple of the specified target length as the radius.

[0071] The target attribute information includes attribute parameters such as target ID, target length, and target width. When the target is a vehicle, the target ID in the target attribute information is specifically the vehicle license plate; the target length in the target attribute information is specifically the vehicle length; and the target width in the target attribute information is specifically the vehicle width.

[0072] After the electronic device determines the target length indicated by the target attribute information corresponding to the large target point set as the designated target length, it can define a circular range with the target point as the center and a specified multiple of the designated target length as the radius. As an example only, the radius r can be expressed by the following formula:

[0073] r=a*L (4)

[0074] Among them, L is the specified target length; a is the specified multiple. It can be understood that the value of a should not be too large, and its value range can be: 1≤a≤3.

[0075] In some embodiments, based on the aforementioned to-be-searched range, the electronic device may search for target splitting points in an orderly manner. The search process is as follows:

[0076] B1. Divide the to-be-searched range into N non-overlapping to-be-searched sub-ranges.

[0077] The electronic device can first determine the equal division length Δr based on the radius r of the range to be searched. Then, with the target point as the center of the circle, the electronic device can draw a circle with an increment of Δr within the range to be searched; that is, still with the target point as the center of the circle and nΔr (n = 1, 2, ..., N) as the radius, thereby dividing the range to be searched into N non-overlapping sub-ranges to be searched according to the equal division length Δr. It should be noted that in order to ensure that N is an integer, Δr needs to be a divisor of r, that is, Δr can be divided by r. Then N can be expressed by the following formula:

[0078]

[0079] in:

[0080] The first sub-range to be searched is: a circular range with the target split point as the center and the equally divided length as the radius;

[0081] The i-th sub-range to be searched is: the annular range formed by the i-th circle and the i-1-th circle, the i-th circle has the target point as the center and i times the equally divided length as the radius, the i-1-th circle has the target point as the center and i-1 times the equally divided length as the radius, and i is a positive integer greater than 1 and less than N+1.

[0082] It should be noted that the value of Δr should not be too small, and the calculated N should be kept within 10 to ensure processing efficiency; the value of Δr should not be too large to avoid the existence of a large number of points belonging to other targets within a range.

[0083] To understand the division process of the sub-range to be searched, please refer to Figure 2 , Figure 2 An example of the sub-range to be searched from a bird's-eye view is given. Figure 2In the figure, there are five sub-ranges to be searched. For ease of understanding, different shades of gray are used to fill the different sub-ranges to be searched. The first sub-range to be searched has the lightest gray, the second has the second lightest gray, and so on, until the fifth has the darkest gray. It can be seen that, except for the first sub-range to be searched, which is circular, the other sub-ranges to be searched are all annular, and the width of each annular ring and the radius of the circle are equally divided by the length Δr.

[0084] B2. For each sub-range to be searched, calculate the sum of the speed differences between the split point and the destination target point within the sub-range to be searched.

[0085] The split point proposed in the embodiment of the present application refers to a point that is not classified into the large target point set. For the nth sub-range to be searched, the electronic device can calculate the sum of the speed differences between the split point in the sub-range to be searched and the target point through the following process:

[0086] First, the radial velocity of each split point within the nth sub-range to be searched and the radial velocity of the target point are obtained. The radial velocity refers to the radial velocity of the detected point relative to the radar, which can be uploaded by the radar.

[0087] Then, the absolute value of the radial velocity difference between each split point in the nth sub-range to be searched and the target point is calculated, which can be expressed by the following formula:

[0088] Δv n,j =|v0-v n,j |,j=1,2,… (6)

[0089] Where Δv n,j is the absolute value of the radial velocity difference between the jth split point and the target point in the nth sub-range to be searched; v0 is the radial velocity of the target point; v n,j is the radial velocity of the jth splitting point in the nth sub-range to be searched.

[0090] Finally, the absolute values ​​of all radial velocity differences obtained in the nth sub-range to be searched are summed to obtain the sum of the velocity differences in the nth sub-range to be searched, which can be expressed by the following formula:

[0091]

[0092] Where Δv n is the sum of the speed differences in the nth sub-range to be searched.

[0093] B3. Determine the target search sub-range based on the sum of the speed differences.

[0094] It is understood that for points belonging to the same target, the absolute value of the radial velocity difference between these points and the target point should change steadily with the distance between the points and the target point. Based on this, the electronic device can determine the target search sub-range by the sudden change of the sum of the velocity differences. The specific process is as follows:

[0095] Detect whether the sum of the speed differences of the k+1th sub-range to be searched has a sudden change compared to the sum of the speed differences of the kth sub-range to be searched, where k is initially 1;

[0096] If it is detected that the sum of the speed differences of the k+1th sub-range to be searched has a sudden change compared to the sum of the speed differences of the kth sub-range to be searched, the first k sub-ranges to be searched are all determined as target search sub-ranges;

[0097] If it is detected that the sum of the speed differences of the k+1th subrange to be searched does not undergo a sudden change compared to the sum of the speed differences of the kth subrange to be searched, k is updated to the next positive integer, and the process returns to the step of detecting whether the sum of the speed differences of the k+1th subrange to be searched undergoes a sudden change compared to the sum of the speed differences of the kth subrange to be searched, and subsequent steps, wherein the updated k is less than N.

[0098] Whether a mutation occurs can be determined as follows: first, the absolute value of the difference between the sum of the speed differences of the k+1th sub-range to be searched and the sum of the speed differences of the kth sub-range to be searched is calculated, as shown in the following formula:

[0099] δ k+1,k =|Δv k+1 -Δv k | (7)

[0100] Then the calculated δ k+1,k Compare with the preset speed difference threshold ω: If δ k+1,k ≤ω, no mutation occurs; if δ k+1,k >ω, a mutation occurs.

[0101] The above process can be understood as: the electronic device judges δ in turn 2,1 , δ 3,2 and δ 4,3 Wait until you find the first δ that is greater than ω. k+1,k Based on the first δ greater than ω k+1,k , the first k sub-ranges to be searched are determined as target search sub-ranges.

[0102] B4. Determine all splitting points in the target search sub-range as target splitting points.

[0103] The electronic device considers that the split points within the to-be-searched sub-range, when the sum of the speed differences does not change suddenly, and the target point all belong to the same target. Based on this, after determining the target search sub-range, the electronic device can determine all split points within the target search sub-range as target split points. Subsequently, in step 106, these target split points are all classified into the large target point set to which the target point belongs.

[0104] In some embodiments, the range to be searched and the sub-range to be searched can also be divided based on other shapes. The other shapes can be squares or triangles, etc., which are not limited here. As an example only, the range to be searched can be a square range with the destination target point as the center and a specified multiple of the specified target length as the side length; similarly, for N non-overlapping sub-ranges to be searched, the equal division length can be determined based on the side length of the range to be searched, then the first sub-range to be searched can be: a square range with the destination target point as the center and the equal division length as the side length; the i-th sub-range to be searched can be: the range formed by the i-th square and the i-1-th square, the i-th square with the destination target point as the center and the i-1-th equal division length as the side length, the i-1-th square with the destination target point as the center and the i-1-th equal division length as the side length.

[0105] It can be understood that the electronic device mentioned in the embodiment of the present application can be a server or other device with information processing capabilities, and the type of the electronic device is not limited here.

[0106] In some embodiments, see Figure 3 , Figure 3 The deployment examples of various devices involved in the large target detection method are given. Figure 3 In the figure, the dotted arrow indicates the driving direction, the black circle indicates the radar, the black square indicates the RSU (that is, other equipment used for detection), and the box indicates the electronic equipment. It can be understood that multiple radars and RSUs can be deployed. Figure 3 There is no limit on the number of radars and RSUs. Figure 3 It can be seen that the radar and RSU can both be deployed on the gantry of the highway; the dark gray area selected by the solid line is the target area, that is, the common detection area of ​​the radar and RSU; the light gray area selected by the gray dotted line is the overlapping detection area of ​​the radar deployed on two adjacent gantries; the white area selected by the dotted line between the two gantries is the separate detection area of ​​the radar. The radar can perform multi-target detection, and the RSU can interact with the OBUs of multiple vehicles to obtain vehicle attribute information. Of course, in actual applications, the radar, RSU and electronic equipment can be deployed according to the road planning situation. The embodiments of this application do not limit the deployment method of each device.

[0107] In some embodiments, it is understood that when a large vehicle enters a separate detection area of ​​a radar, since only the radar can detect the target, this may cause the large target to split. If this occurs, the electronic device can immediately use the adaptive radius DBSCAN clustering algorithm proposed above to cluster the target, thereby ensuring the continuity and uniqueness of the large vehicle's trajectory.

[0108] In some embodiments, a communication connection is established between two adjacent electronic devices, where the two adjacent electronic devices refer to the electronic devices connected to the radars on two adjacent gantries. It is understood that, of the two adjacent electronic devices, the upstream electronic device can output the detected track of the large vehicle to the downstream electronic device (upstream and downstream are determined based on the driving direction), thereby enabling the tracking of the large vehicle's track on different road sections, further ensuring the continuity and uniqueness of the large vehicle's track.

[0109] As can be seen from the above, through the embodiment of the present application, first, a point cloud to be processed is obtained, wherein the point cloud to be processed is obtained by detecting the target area through radar, and then the point cloud to be processed is clustered to determine at least one target point set. This process realizes clustering at the point cloud data level. After that, a large target point set can be screened out from at least one target point set, and for each large target point set, the target attribute information corresponding to the large target point set is used to determine the search range of the large target point set. Then, the target splitting point is searched within the search range, wherein the target splitting point is a point related to the large target point set but not classified into the large target point set. Finally, the target splitting point is classified into the large target point set. This process realizes re-clustering at the target data level. Through clustering at the point cloud data level and re-clustering at the target data level, it can be ensured that the points obtained by the radar detecting the same large target can be classified into the same large target point set as much as possible, thereby reducing the situation where a large target is detected as multiple small targets.

[0110] Corresponding to the large target detection method provided above, the embodiment of the present application also provides a large target detection device. Figure 4 As shown, the large target detection device 400 includes:

[0111] An acquisition module 401 is configured to acquire a point cloud to be processed, wherein the point cloud to be processed is obtained by detecting a target area with a radar;

[0112] A first determining module 402 is configured to cluster the point cloud to be processed and determine at least one target point set;

[0113] A screening module 403 is configured to screen out a large target point set from the at least one target point set;

[0114] The second determining module 404 is configured to determine, for each large target point set, a to-be-searched range of the large target point set according to target attribute information corresponding to the large target point set;

[0115] A search module 405 is configured to search for a target splitting point within the to-be-searched range, wherein the target splitting point is a point that is related to the large target point set but is not classified into the large target point set;

[0116] The classification module 406 is configured to classify the target split points into the large target point set.

[0117] Optionally, the second determining module 404 includes:

[0118] a destination target point determination unit, configured to determine a destination target point in the large target point set;

[0119] The unit for determining the range to be searched is used to determine the range to be searched with the destination target point as the center and a specified multiple of a specified target length as the radius, wherein the specified target length is the target length indicated by the target attribute information corresponding to the large target point set.

[0120] Optionally, the search module 405 includes:

[0121] a to-be-searched sub-range dividing unit, configured to divide the to-be-searched range into N non-overlapping to-be-searched sub-ranges;

[0122] a speed difference sum calculation unit, configured to calculate, for each sub-range to be searched, a sum of speed differences between a split point in the sub-range to be searched and the destination target point, wherein the split point is a point that is not classified into the large target point set;

[0123] a target search subrange determining unit, configured to determine a target search subrange according to the sum of the speed differences;

[0124] The target splitting point determining unit is configured to determine all splitting points in the target search sub-range as target splitting points.

[0125] Optionally, the unit for dividing the sub-range to be searched includes:

[0126] an equally divided length determining subunit, configured to determine the equally divided length based on the radius of the range to be searched;

[0127] a range division subunit, configured to divide the to-be-searched range into N non-overlapping to-be-searched sub-ranges according to the equally divided lengths;

[0128] in:

[0129] The first sub-range to be searched is a circular range with the target point as the center and the equal-division length as the radius;

[0130] The i-th sub-range to be searched is: the annular range formed by the i-th circle and the i-1-th circle, the i-th circle has the target point as the center and i times the equal-division length as the radius, the i-1-th circle has the target point as the center and i-1 times the equal-division length as the radius, and i is a positive integer greater than 1 and less than N+1.

[0131] Optionally, the speed difference sum calculation unit includes:

[0132] A calculation subunit, configured to calculate an absolute value of a radial velocity difference between each split point in the subrange to be searched and the target point;

[0133] The summing subunit is configured to sum the calculated absolute values ​​of all radial velocity differences within the sub-range to be searched to obtain the sum of the velocity differences.

[0134] Optionally, the target search sub-range determining unit includes:

[0135] a mutation detection subunit, configured to detect whether a mutation occurs between the sum of the speed differences of the k+1th sub-range to be searched and the sum of the speed differences of the kth sub-range to be searched, where k is initially 1;

[0136] a target search subrange determining subunit, configured to, if it is detected that the sum of the speed differences of the k+1th subrange to be searched is suddenly changed compared with the sum of the speed differences of the kth subrange to be searched, determine the first k subranges to be searched as the target search subranges;

[0137] The k value updating subunit is configured to update k to the next positive integer if it is detected that the sum of the speed differences of the k+1th subrange to be searched does not undergo a sudden change compared to the sum of the speed differences of the kth subrange to be searched, and trigger the re-execution of the above-mentioned sudden change detection subunit, wherein the updated k is less than N.

[0138] Optionally, the screening module 403 includes:

[0139] a receiving unit, configured to receive at least one piece of target attribute information uploaded by a roadside unit, the roadside unit being synchronized with the radar;

[0140] a fusion unit, configured to fuse the at least one piece of target attribute information with the at least one target point set;

[0141] The screening unit is configured to screen out a large target point set from the at least one target point set based on target attribute information corresponding to each of the target point sets.

[0142] As can be seen from the above, through the embodiment of the present application, first, a point cloud to be processed is obtained, wherein the point cloud to be processed is obtained by detecting the target area through radar, and then the point cloud to be processed is clustered to determine at least one target point set. This process realizes clustering at the point cloud data level. After that, a large target point set can be screened out from at least one target point set, and for each large target point set, the target attribute information corresponding to the large target point set is used to determine the search range of the large target point set. Then, the target splitting point is searched within the search range, wherein the target splitting point is a point related to the large target point set but not classified into the large target point set. Finally, the target splitting point is classified into the large target point set. This process realizes re-clustering at the target data level. Through clustering at the point cloud data level and re-clustering at the target data level, it can be ensured that the points obtained by the radar detecting the same large target can be classified into the same large target point set as much as possible, thereby reducing the situation where a large target is detected as multiple small targets.

[0143] Corresponding to the large target detection method provided above, the embodiment of the present application further provides an electronic device. Figure 5 The electronic device 5 in the embodiment of the present application includes: a memory 501, one or more processors 502 ( Figure 5 Only one is shown) and a computer program stored in memory 501 and executable on the processor. Memory 501 is used to store software programs and units. Processor 502 executes the software programs and units stored in memory 501 to perform various functional applications and diagnose to obtain resources corresponding to the above-mentioned preset events. Specifically, when processor 502 executes the above-mentioned computer program stored in memory 501, it implements the following steps:

[0144] Acquire a point cloud to be processed, wherein the point cloud to be processed is obtained by detecting a target area with a radar;

[0145] Clustering the point cloud to be processed to determine at least one target point set;

[0146] Filtering a large target point set from the at least one target point set;

[0147] For each large target point set, determining a to-be-searched range of the large target point set according to target attribute information corresponding to the large target point set;

[0148] Searching for a target splitting point within the to-be-searched range, wherein the target splitting point is a point that is related to the large target point set but is not classified into the large target point set;

[0149] The target split points are classified into the large target point set.

[0150] Assuming that the above is the first possible implementation, in a second possible implementation provided on the basis of the first possible implementation, determining the to-be-searched range of the large target point set according to the target attribute information corresponding to the large target point set includes:

[0151] Determining a destination target point in the large target point set;

[0152] The range to be searched is determined with the destination target point as the center and a specified multiple of a specified target length as the radius, wherein the specified target length is the target length indicated by the target attribute information corresponding to the large target point set.

[0153] In a third possible implementation provided on the basis of the second possible implementation, searching for a target splitting point within the to-be-searched range includes:

[0154] Dividing the to-be-searched range into N non-overlapping to-be-searched sub-ranges;

[0155] For each sub-range to be searched, calculating the sum of speed differences between a split point in the sub-range to be searched and the destination target point, wherein the split point is a point that is not classified into the large target point set;

[0156] determining a target search subrange according to the sum of the speed differences;

[0157] All splitting points in the target search subrange are determined as target splitting points.

[0158] In a fourth possible implementation provided on the basis of the third possible implementation, dividing the to-be-searched range into N non-overlapping to-be-searched sub-ranges includes:

[0159] Determining the length of the equal divisions based on the radius of the range to be searched;

[0160] Dividing the to-be-searched range into N non-overlapping to-be-searched sub-ranges according to the equally divided lengths;

[0161] in:

[0162] The first sub-range to be searched is a circular range with the target point as the center and the equal-division length as the radius;

[0163] The i-th sub-range to be searched is: the annular range formed by the i-th circle and the i-1-th circle, the i-th circle has the target point as the center and i times the equal-division length as the radius, the i-1-th circle has the target point as the center and i-1 times the equal-division length as the radius, and i is a positive integer greater than 1 and less than N+1.

[0164] In a fifth possible implementation provided based on the third possible implementation, calculating the sum of speed differences between the split points within the sub-range to be searched and the destination target point includes:

[0165] Calculate the absolute value of the radial velocity difference between each split point in the sub-range to be searched and the target point;

[0166] The calculated absolute values ​​of all radial velocity differences within the sub-range to be searched are summed to obtain the sum of the velocity differences.

[0167] In a sixth possible implementation provided based on the third possible implementation, determining the target search subrange according to the sum of the speed differences includes:

[0168] Detecting whether a sudden change occurs in the sum of the speed differences of the k+1th sub-range to be searched compared to the sum of the speed differences of the kth sub-range to be searched, where k is initially 1;

[0169] If it is detected that the sum of the speed differences of the k+1th sub-range to be searched is suddenly changed compared with the sum of the speed differences of the kth sub-range to be searched, the first k sub-ranges to be searched are all determined as the target search sub-ranges;

[0170] If it is detected that the sum of the speed differences of the k+1th sub-range to be searched does not undergo a sudden change compared to the sum of the speed differences of the kth sub-range to be searched, k is updated to the next positive integer, and the process returns to execute the step of detecting whether the sum of the speed differences of the k+1th sub-range to be searched does not undergo a sudden change compared to the sum of the speed differences of the kth sub-range to be searched and subsequent steps, wherein the updated k is less than N.

[0171] In a seventh possible implementation provided on the basis of the first possible implementation, the step of screening out a large target point set from the at least one target point set includes:

[0172] receiving at least one piece of target attribute information uploaded by a roadside unit, the roadside unit being synchronized with the radar;

[0173] fusing the at least one piece of target attribute information with the at least one target point set;

[0174] Based on the target attribute information corresponding to each of the target point sets, a large target point set in the at least one target point set is screened out.

[0175] It should be understood that in the embodiment of the present application, the processor 502 may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc.

[0176] The memory 501 may include a read-only memory and a random access memory, and provides instructions and data to the processor 502. A portion or all of the memory 501 may also include a non-volatile random access memory. For example, the memory 501 may also store device category information.

[0177] As can be seen from the above, through the embodiment of the present application, first, a point cloud to be processed is obtained, wherein the point cloud to be processed is obtained by detecting the target area through radar, and then the point cloud to be processed is clustered to determine at least one target point set. This process realizes clustering at the point cloud data level. After that, a large target point set can be screened out from at least one target point set, and for each large target point set, the target attribute information corresponding to the large target point set is used to determine the search range of the large target point set. Then, the target splitting point is searched within the search range, wherein the target splitting point is a point related to the large target point set but not classified into the large target point set. Finally, the target splitting point is classified into the large target point set. This process realizes re-clustering at the target data level. Through clustering at the point cloud data level and re-clustering at the target data level, it can be ensured that the points obtained by the radar detecting the same large target can be classified into the same large target point set as much as possible, thereby reducing the situation where a large target is detected as multiple small targets.

[0178] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the above-mentioned device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0179] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant description of other embodiments.

[0180] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of external device software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0181] In the embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the system embodiments described above are merely schematic. For example, the division of the above modules or units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0182] The units described above as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0183] If the above-mentioned integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present application implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the associated hardware through a computer program. The above-mentioned computer program can be stored in a computer-readable storage medium, and the computer program, when executed by the processor, can implement the steps of the above-mentioned various method embodiments. Among them, the above-mentioned computer program includes computer program code, and the above-mentioned computer program code can be in source code form, object code form, executable file or some intermediate form, etc. The above-mentioned computer-readable storage medium may include: any entity or device that can carry the above-mentioned computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer-readable memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium, etc. It should be noted that the content contained in the above-mentioned computer-readable storage medium can be appropriately increased or decreased according to the requirements of legislation and patent practices in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practices, computer-readable storage media does not include electrical carrier signals and telecommunication signals.

[0184] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present application, and should all be included in the scope of protection of the present application.

Claims

1. A large target detection method, characterized in that: include: Acquire a point cloud to be processed, wherein the point cloud to be processed is obtained by detecting a target area by a radar, and the target area is a common coverage area of ​​other devices and the radar; Clustering the point cloud to be processed to determine at least one target point set; In the at least one target point set, a large target point set is screened out by the other device, wherein the other device is synchronized with the radar; For each large target point set, determining a to-be-searched range of the large target point set according to target attribute information corresponding to the large target point set; Searching for a target splitting point within the to-be-searched range, wherein the target splitting point is a point that is related to the large target point set but is not classified into the large target point set; The target split points are classified into the large target point set.

2. The large target detection method according to claim 1, wherein: The determining, based on the target attribute information corresponding to the large target point set, a to-be-searched range of the large target point set includes: Determining a destination target point in the large target point set; The range to be searched is determined with the destination target point as the center and a specified multiple of a specified target length as the radius, wherein the specified target length is the target length indicated by the target attribute information corresponding to the large target point set.

3. The large target detection method according to claim 2, wherein: The step of searching for a target splitting point within the to-be-searched range includes: Dividing the to-be-searched range into N non-overlapping to-be-searched sub-ranges; For each sub-range to be searched, calculating the sum of speed differences between a split point in the sub-range to be searched and the destination target point, wherein the split point is a point that is not classified into the large target point set; determining a target search subrange according to the sum of the speed differences; All splitting points in the target search subrange are determined as target splitting points.

4. The large target detection method according to claim 3, wherein: The step of dividing the to-be-searched range into N non-overlapping to-be-searched sub-ranges includes: Determining the length of the equal divisions based on the radius of the range to be searched; Dividing the to-be-searched range into N non-overlapping to-be-searched sub-ranges according to the equally divided lengths; in: The first sub-range to be searched is a circular range with the target point as the center and the equal-division length as the radius; The i-th sub-range to be searched is: the annular range formed by the i-th circle and the i-1-th circle, the i-th circle has the target point as the center and i times the equal-division length as the radius, the i-1-th circle has the target point as the center and i-1 times the equal-division length as the radius, and i is a positive integer greater than 1 and less than N+1.

5. The large target detection method according to claim 3, wherein: The calculating the sum of speed differences between the split point in the sub-range to be searched and the destination target point includes: Calculate the absolute value of the radial velocity difference between each split point in the sub-range to be searched and the target point; The calculated absolute values ​​of all radial velocity differences within the sub-range to be searched are summed to obtain the sum of the velocity differences.

6. The large target detection method according to claim 3, wherein: The determining of the target search sub-range according to the sum of the speed differences includes: Detecting whether a sudden change occurs in the sum of the speed differences of the k+1th sub-range to be searched compared to the sum of the speed differences of the kth sub-range to be searched, where k is initially 1; If it is detected that the sum of the speed differences of the k+1th sub-range to be searched is suddenly changed compared with the sum of the speed differences of the kth sub-range to be searched, the first k sub-ranges to be searched are all determined as the target search sub-ranges; If it is detected that the sum of the speed differences of the k+1th sub-range to be searched does not undergo a sudden change compared to the sum of the speed differences of the kth sub-range to be searched, k is updated to the next positive integer, and the process returns to execute the step of detecting whether the sum of the speed differences of the k+1th sub-range to be searched does not undergo a sudden change compared to the sum of the speed differences of the kth sub-range to be searched and subsequent steps, wherein the updated k is less than N.

7. The large target detection method according to claim 1, wherein: The step of screening out a large target point set from the at least one target point set comprises: receiving at least one piece of target attribute information uploaded by a roadside unit, the roadside unit being synchronized with the radar; fusing the at least one piece of target attribute information with the at least one target point set; Based on the target attribute information corresponding to each of the target point sets, a large target point set in the at least one target point set is screened out.

8. A large target detection device, characterized in that: include: An acquisition module is used to acquire a point cloud to be processed, wherein the point cloud to be processed is obtained by detecting a target area by a radar, and the target area is a common coverage area of ​​other devices and the radar; A first determination module is configured to cluster the point cloud to be processed and determine at least one target point set; a screening module, configured to screen out a large target point set from the at least one target point set by using the other device, wherein the other device is synchronized with the radar; A second determining module is configured to determine, for each large target point set, a to-be-searched range of the large target point set according to target attribute information corresponding to the large target point set; a search module, configured to search for a target splitting point within the to-be-searched range, wherein the target splitting point is a point that is related to the large target point set but is not classified into the large target point set; A classification module is used to classify the target split points into the large target point set.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images