Method and system for accurate vehicle identification on upper and lower composite highways

Abstract

The embodiment of the present invention discloses a method and system for accurately identifying vehicles on a composite highway with upper and lower layers, and relates to the field of intelligent transportation technology. The method includes: using ETC antennas installed on the gantries on each layer of the highway to preliminarily identify vehicle information passing through the gantries on each layer; using a radar-visual integrated machine installed on the same gantries as each ETC antenna to exclude misidentified suspected lower-layer vehicles from the preliminary identification information of the upper-layer gantry. This embodiment provides an efficient multi-sensor data fusion technology to improve vehicle identification performance in a composite highway environment.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of intelligent transportation technology, and in particular to a method and system for accurately identifying vehicles on an upper and lower composite highway. Background Art

[0002] With the popularity of composite highways, traditional vehicle identification technology faces more and more challenges. The design of composite highways usually includes multiple lanes. This structural complexity brings special technical requirements for vehicle positioning and identification. Existing vehicle identification systems usually rely on a single type of sensor (such as radar or camera), which may be sufficient under simple road conditions, but in the multi-layer highway environment, the recognition accuracy of these systems is significantly reduced. A single camera or radar can only cover a limited area and cannot capture the complex dynamics of the interweaving upper and lower layers. Traditional systems often have delays when processing large amounts of data collected from different sensors. At the same time, under changeable weather conditions (such as rain, fog, and snow), the performance of traditional sensors may drop significantly, especially optical cameras and certain types of radars. The lack of an effective data fusion mechanism between different sensors makes it difficult to verify the data of each sensor with each other, increasing the risk of misjudgment. In the multi-layer structure of composite highways, the rapid change of vehicles between different layers increases the difficulty of identification. It is difficult for existing technologies to quickly and accurately distinguish and track the exact location of the target vehicle. In response to the above problems, existing solutions often fail to achieve ideal recognition efficiency and accuracy, especially under high traffic density and complex meteorological conditions. Therefore, developing an efficient multi-sensor data fusion technology to improve vehicle recognition performance in complex highway environments has become a technical problem that urgently needs to be solved. Summary of the invention

[0003] The embodiments of the present invention provide a method and system for accurately identifying vehicles on a composite upper and lower-level highway to solve the above-mentioned technical problems.

[0004] In a first aspect, an embodiment of the present invention provides a method for accurately identifying vehicles on a composite upper and lower-level highway, comprising:

[0005] Using the ETC antennas installed on the gantries at each level of the highway, the vehicle information passing through each level of the gantries is preliminarily identified;

[0006] Using a radar-vision integrated device installed on the same gantry as each ETC antenna, misidentified suspected lower-level vehicles are eliminated from the preliminary identification information of the upper gantry.

[0007] In a second aspect, an embodiment of the present invention provides a vehicle accurate identification system for a composite highway with upper and lower layers, including multiple gantries, multiple ETC antennas, multiple integrated radar and visual devices, and an industrial computer, wherein each layer of the highway is respectively deployed with at least one gantry, and each gantry is respectively equipped with an ETC antenna and an integrated radar and visual device;

[0008] The industrial computer is used to use each ETC antenna to preliminarily identify the vehicle information passing through each layer of the gantry; and use the radar-visual integrated machine installed on the same gantry as each ETC antenna to exclude misidentified suspected lower-level vehicles from the preliminary identification information of the upper-level gantry.

[0009] In summary, the embodiment of the present invention provides a method and system for accurate vehicle identification on a composite highway with upper and lower layers, which improves the identification accuracy and efficiency of vehicles on the composite highway through innovative sensor deployment and advanced data processing technology. By installing a radar-visual integrated machine and an ETC antenna on the upper and lower layers of the highway, respectively, accurate capture of vehicle information and hierarchical distinction can be achieved. Such a setting can perform special processing and analysis on dynamic traffic at different levels, thereby improving the accuracy and matching speed of data. The radar-visual integrated machine is used to achieve high-precision vehicle and license plate recognition. This technology uses an advanced neural network model to locate the vehicle, license plate and license plate information, ensuring that the recognition task can be performed efficiently in a complex natural environment, especially within the gantry range of about 250 meters on the highway, and can accurately capture and process vehicle information. In the face of complex environmental factors such as changes in light, weather conditions, etc., the impact of environmental noise on recognition accuracy is reduced by using high-performance sensors and optimized neural network models. Through real-time data processing technology and an adaptive feedback system, the recognition results are analyzed and adjusted in real time to ensure the timeliness and accuracy of the recognition information. For data that cannot be directly matched and fused, the system can conduct in-depth analysis and output corresponding processing results to support more complex decision-making needs. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0011] Figure 1 It is a schematic diagram of a vehicle accurate identification system for an upper and lower composite highway provided by an embodiment of the present invention;

[0012] Figure 2It is a flow chart of a method for accurately identifying vehicles on a composite upper and lower highway provided by an embodiment of the present invention;

[0013] Figure 3 It is a schematic diagram of data collection of an ETC transaction system provided by an embodiment of the present invention;

[0014] Figure 4 It is a schematic diagram of a video device of a radar-vision integrated machine identifying vehicle information provided by an embodiment of the present invention;

[0015] Figure 5 It is a schematic diagram of a roadside ETC antenna coverage provided by an embodiment of the present invention;

[0016] Figure 6 It is a schematic diagram of a vehicle accurate identification system for upper and lower composite highways provided by an embodiment of the present invention for collecting and fusing vehicle information from different types of sensors;

[0017] Figure 7 It is a schematic diagram of another vehicle accurate identification system for upper and lower composite highways provided by an embodiment of the present invention for collecting and fusing vehicle information from different types of sensors. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work belong to the scope of protection of the present invention.

[0019] In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicating the orientation or positional relationship, are based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first", "second", and "third" are used for descriptive purposes only, and cannot be understood as indicating or implying relative importance.

[0020] In the description of the present invention, it is also necessary to explain that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0021] An embodiment of the present invention provides a method for accurately identifying vehicles on an upper and lower composite highway. In order to illustrate the method, a system architecture that supports the implementation of the method is first introduced. Figure 1 is a schematic diagram of a vehicle accurate identification system for a composite upper and lower highway provided by an embodiment of the present invention, such as Figure 1 As shown in the figure, the system is deployed on a three-dimensional upper and lower highway, including multiple gantries, multiple ETC antennas, multiple integrated radar and vision machines and industrial computers. Among them, a gantry is installed at a certain distance on each level of the highway, and each gantry is equipped with an ETC antenna and an integrated radar and vision machine to identify the information of vehicles passing through each gantry. Each gantry column is installed with an industrial computer and multi-source information collection software to realize the collection and fusion of gantry, radar and vision machine and ETC antenna data.

[0022] Based on the above system architecture, Figure 2 1 is a flow chart of a method for accurately identifying vehicles on a composite highway with upper and lower layers provided by an embodiment of the present invention. The method is applicable to the case of identifying vehicles on a composite highway with upper and lower layers, and can be executed by the industrial computer in the above system. Figure 2 As shown, the method specifically includes:

[0023] S110. Unify time and space benchmarks.

[0024] In this embodiment, the industrial computer provides a unified time service to ensure that the time of the radar-visual integrated machine and the ETC antenna remains consistent, forming a unified time reference.

[0025] At the same time, highway section units are established. Each section unit has a prepared identification range for the radar-vision integrated device and the ETC antenna. The space is divided by longitude and latitude to form a unified spatial reference.

[0026] S120: Modeling effective identification units of road sections.

[0027] This step divides the electronic fence of the effective identification area of ​​the road section by installing the integrated radar and ETC antenna on the gantry. Among them, the effective road section position can be determined by the gantry, and the electronic fence of the effective identification area can be divided at the gantry position; the integrated radar and ETC antenna as sensing devices can collect information about vehicles passing through the electronic fence of the effective identification area, such as: license plate number, passing time, etc. Optionally, the gantry is a fixed installation device, and the output data mainly includes gantry ID, gantry name, longitude, latitude and other information.

[0028] S130: Using the ETC antennas installed on the gantries at each level of the expressway, preliminarily identify the information of vehicles passing through the gantries at each level.

[0029] The ETC transaction system (RSU antenna) is based on the national standard DSRC radio frequency identification technology, which can accurately and quickly identify the license plate number information at a long distance. The RSU antenna array design effectively suppresses interference from side lanes, following vehicles, and interference from the back of the antenna board. The data collected and output by the roadside ETC antenna (i.e., RSU antenna) include the license plate number and the reception time, as shown in Table 1. The entire data collection process is as follows: Figure 3 shown.

[0030] Table 1

[0031]

[0032]

[0033] S140. Using a radar-vision integrated device installed on the same gantry as each ETC antenna, misidentified suspected lower-level vehicles are excluded from the preliminary identification information of the upper-level gantry.

[0034] The upper gantry is a gantry deployed on the upper highway. In a specific implementation, the above process may include the following steps:

[0035] Step 1: Use the radar and video integrated device installed on the upper gantry to perform secondary identification of vehicle information passing through the upper gantry. The radar and video integrated device includes radar and video equipment. On the one hand, the radar can be used to perform secondary identification of vehicle targets passing through the gantry where the radar and video integrated device is located. Optionally, the radar can be used to identify the position of vehicle targets passing through the gantry where the radar and video integrated device is located; then determine the translation matrix based on the relative position of the radar and the road; determine the rotation matrix based on the angle between the radar and the road; translate and rotate the vehicle position in the radar coordinate system based on the translation matrix and the rotation matrix to obtain the vehicle position in the road coordinate system; the position and speed of each vehicle target in the road coordinate system together constitute the vehicle information secondary identified by the radar.

[0036] Specifically, the radar can obtain the x, y coordinate information of the target, but not the z coordinate information of the target. The conversion from the radar coordinate system to the road coordinate system can be regarded as the conversion of the two-dimensional XY coordinate system. Based on mathematical knowledge, the conversion relationship from the radar coordinate system to the road coordinate system can be derived. The conversion matrix is ​​composed of two parts: the rotation matrix caused by the angle, and the translation matrix generated by the translation. The translation amount of the translation matrix can be understood as the horizontal distance between the device and the origin of the road coordinate system. Therefore, the translation matrix can be obtained by adjusting the horizontal offset, and the rotation amount of the rotation matrix can be obtained by adjusting the angle of the radar. If the device is installed facing the road coordinate system, it can also be understood that the rotation matrix is ​​0, and there is no need to adjust the radar angle value. The device is required to be installed facing the center of the road as much as possible to reduce the adjustment amount of the angle and horizontal offset.

[0037] The purpose of radar calibration is mainly to make the orientation of the radar target consistent with that of the target on the actual road. For example, a radar is not installed completely in the center of the lane. If the radar is not calibrated, assuming that there is a car driving straight along lane 1, there may be a certain offset in the radar interface. In order to be consistent with the actual situation, the angle of the radar needs to be corrected so that the vehicle drives along the direction of the lane; then adjust its horizontal offset. If the vehicle that was actually driving in lane 1 after just straightening is driving in lane 2 on the radar interface, it is necessary to adjust the horizontal distance of one lane to the left. After adjusting these two parameters, the vehicle driving on the radar interface is consistent with the actual driving situation.

[0038] Assume that the point in the radar coordinate system is (x r ,y r ), the point in the road coordinate system is (x d ,y d ), the rotation angle of the radar relative to the road coordinate system is θ, and the horizontal offset is (d x ,d y ). Then the transformation matrix can be expressed as:

[0039]

[0040] The first part is the rotation matrix, which is responsible for rotating the points in the radar coordinate system to the direction of the road coordinate system according to the angle of the radar. The second part is the translation matrix, which translates the rotated points to match the origin of the road coordinate system. The rotation matrix allows the points in the radar coordinate system to maintain the correct orientation in the road coordinate system, while the translation matrix adjusts these points to match the actual physical location. This conversion is necessary, especially when the radar device may not be installed exactly in the center of the road or has a certain angle deviation.

[0041] On the other hand, video equipment can be used to perform secondary recognition of vehicle targets passing through the gantry where the integrated radar and vision device is located. Figure 4 , a video device can be used to capture images of the covered road section on one side of the gantry where the integrated radar and vision device is located; the captured images are input into the vehicle detection neural network model to identify the vehicle area in the image (i.e. Figure 4 The vehicle area is input into the license plate detection neural network model to identify the license plate area (i.e. Figure 4 The license plate area is input into the license plate information neural network model to identify the specific license plate information; the position, speed, license plate information, body color, etc. of each vehicle target together constitute the vehicle information recognized by the video device for the second time.

[0042] For the sake of distinction and description, in this embodiment, the vehicle target identified by the radar is referred to as the first vehicle target, and the vehicle target identified by the video device is referred to as the second vehicle target. After obtaining the vehicle targets provided by the two devices, each first vehicle target is matched with each second vehicle target; based on the successfully matched vehicle targets, the vehicle information recognized by the radar and video integrated device for the second time is generated.

[0043] Traditional fusion schemes and fusion strategies are mainly single-frame fusion. When waiting in line at an intersection, the distance between targets is very close, the visual frame overlap is high, and mismatching is prone to occur. Or when parking, the radar target disappears and cannot be correctly fused, which affects the continuity of the target trajectory and the accuracy of traffic information. However, this embodiment breaks the above practice. Based on the above calibration information, the spatial position and speed of the target and other tracking information, the correlation matching probability matrix between the radar target and the video target is constructed, and the matching relationship between the radar and video targets with the highest probability is determined by a fast retrieval method as the basic fusion information, and the final reliable data is output by such a comprehensive algorithm.

[0044] Optionally, the matching probability of each first vehicle target and each second vehicle target can be calculated based on the distance, speed and light intensity between each first vehicle target identified by the radar and each second vehicle target identified by the video device. Assume that there is a radar target set (R) and a video target set (V), and each target (r i ∈R) and (v j ∈V) corresponds to a probability p ij , represents the radar target r i With video target v j The probability of matching. This probability can depend on many factors such as distance, lighting, speed, etc. The element p in the probability matrix (P) ij It can be calculated by the following formula:

[0045]

[0046] Among them, d ij and ij represents the distance and speed difference between the i-th first vehicle target and the j-th second vehicle target, d ik and ik represents the distance and speed difference between the i-th first vehicle target and the k-th second vehicle target, N V represents the total number of the second vehicles, l ij and l ik represents the influence of light intensity on matching accuracy, and α, β, and γ represent three weights respectively.

[0047] Finally, the first vehicle target and the second vehicle target whose matching probability meets the conditions are identified as the same vehicle target that has been successfully matched. Based on the above operations, the data finally output by the entire radar-visual integrated device includes the successfully matched license plate number, license plate color, vehicle body color, lane number, vehicle type, speed, time, longitude, latitude, road location, gantry information, etc., as shown in Table 2:

[0048] Table 2

[0049]

[0050] Step 2: If a vehicle exists in the preliminary identification information of the upper gantry but does not exist in the secondary identification information, it is determined that the vehicle is a suspected lower-level vehicle that was misidentified.

[0051] The radar-visual integrated recognition device can effectively identify vehicles within the effective range (about 250 meters) by integrating radar and video. This step matches the information with the vehicle information identified by the ETC antenna. If the data is consistent, it is valid data, otherwise it is invalid data. In particular, since the roadside ETC antenna (RSU) emits microwaves in a fan-shaped shape, in order to better cover vehicles passing through the road, the ETC antenna is usually installed at an angle inclined to the road surface. Its microwave coverage range is as follows: Figure 5 As shown in the dotted line part. Therefore, in the upper and lower three-dimensional highway, the roadside ETC antenna of the upper highway may cover the vehicles on the lower highway, thereby misidentifying the vehicles on the lower highway as vehicles passing through the upper gantry, which is a major reason for the misidentification of vehicle information. Based on this feature, this embodiment mainly uses the ETC transaction system, assisted by the effective identification information of the integrated radar and vision device to eliminate the misidentified vehicles, greatly reducing the probability of vehicle misidentification. The data collection and fusion process of the entire method is as follows: Figure 6 , Figure 7 Of course, the misidentification here may also be caused by the misreading of the vehicle tag information of the upper vehicle, which is referred to as a suspected lower vehicle in this implementation.

[0052] Furthermore, in a specific embodiment, for the vehicle information provided by the roadside ETC antenna, radar, and video equipment (multiple identification sources), different multi-source information fusion strategies can be adopted for the upper and lower highways according to the characteristic that the ETC antenna of the upper highway in the composite highway will cover the lower vehicle. As described above, when multi-source information fusion is performed on the upper highway, it is considered that if a vehicle target is successfully identified in all multi-source information and the vehicle information is consistent, then the vehicle information is valid, otherwise the vehicle information is invalid. When multi-source information fusion is performed on the lower-level highway, the suspected lower-level vehicles that are misidentified by the upper-level ETC antenna can be first added to the list of vehicles preliminarily identified by the ETC antenna of the lower-level gantry in the same longitude and latitude area as the upper-level gantry (that is, the longitude and latitude ranges of the upper and lower road sections are the same, only the heights are different), and then the radar-vision integrated device installed on the lower-level gantry is used to perform secondary identification on the vehicle information passing through the lower-level gantry; if a vehicle exists in both the preliminary identification information and the secondary identification information of the lower-level gantry after addition, and the vehicle information is consistent, the vehicle is judged to be a valid lower-level vehicle.

[0053] In this specific implementation, different multi-source information fusion strategies are adopted for the upper and lower roads: for the upper roads, the strategy of "if a vehicle target is successfully identified in multiple sources of information and the vehicle information is consistent, then the vehicle information is valid, otherwise the vehicle information is invalid" is adopted, that is, the effective information range is continuously narrowed. For the lower roads, this embodiment first uses the suspected lower-level vehicle identified by the ETC antenna of the upper road as a supplement to the vehicle information identified by the lower-level ETC antenna, and then uses the supplementary information to match and verify the identification information of the radar-visual integrated device of the lower road, that is, first expand the effective information range and then narrow it. If the added vehicle does not exist in the lower-level highway or the added vehicle information is wrong, it cannot match the vehicle information identified by the integrated radar vision device on the lower-level highway. The vehicle information can still be filtered out (the probability that each device has an error and the error information is consistent is almost zero); but if the added vehicle and vehicle information are missed due to ETC antenna communication failure on the lower-level highway, the information can be used as a supplement to the vehicle information on the lower-level highway and successfully matched with the vehicle information identified by the integrated radar vision device on the lower-level highway, thereby improving the accuracy of vehicle identification on the lower-level highway. This method breaks the idea of ​​mutual verification and gradual exclusion in the fusion of multi-source information on single-level highways, and expands the source of vehicle information on the lower-level highway to the upper-level highway equipment, which can effectively avoid missed vehicle identification caused by ETC antenna communication failure, etc., and is particularly suitable for vehicle information identification on upper- and lower-level composite highways.

[0054] In summary, this embodiment provides a method for accurately identifying vehicles on a composite highway with upper and lower layers, and improves the recognition accuracy and efficiency of vehicles on the composite highway through innovative sensor deployment and advanced data processing technology. By installing a radar-visual integrated machine and an ETC antenna on the upper and lower layers of the highway, respectively, accurate capture of vehicle information and hierarchical distinction can be achieved. Such a setting can be used to perform special processing and analysis on dynamic traffic at different levels, thereby improving the accuracy and matching speed of data. The radar-visual integrated machine is used to achieve high-precision vehicle and license plate recognition. This technology uses an advanced neural network model to locate the vehicle, license plate, and license plate information, ensuring that the recognition task can be performed efficiently in a complex natural environment, especially within the gantry range of about 250 meters on the highway, and can accurately capture and process vehicle information. In the face of complex environmental factors such as changes in lighting, weather conditions, etc., the impact of environmental noise on recognition accuracy is reduced by using high-performance sensors and optimized neural network models. Through real-time data processing technology and an adaptive feedback system, the recognition results are analyzed and adjusted in real time to ensure the timeliness and accuracy of the recognition information. For data that cannot be directly matched and fused, the system can conduct in-depth analysis and output corresponding processing results to support more complex decision-making needs.

[0055] In addition, the embodiment of the present invention also provides a vehicle accurate identification system for upper and lower composite highways, including multiple gantries, multiple ETC antennas, multiple radar and visual integrated machines, and industrial computers. Figure 1 As shown, each level of the expressway is deployed with at least one gantry, and each gantry is equipped with an ETC antenna and a radar-vision integrated device.

[0056] The industrial computer is used to use each ETC antenna to preliminarily identify the vehicle information passing through each layer of the gantry; and use the radar-visual integrated machine installed on the same gantry as each ETC antenna to exclude misidentified suspected lower-level vehicles from the preliminary identification information of the upper-level gantry.

[0057] This embodiment and any of the above method embodiments are based on the same inventive concept, and the specific limitations of any of the above method embodiments are applicable to this embodiment, and can achieve the same technical effects as any of the above method embodiments.

[0058] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention 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 replace some or all of the technical features therein by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the technical solutions of the embodiments of the present invention.

Claims

1. A method for accurately identifying vehicles on upper and lower composite highways, characterized in that: include: Using the ETC antennas installed on the gantries at each level of the highway, the vehicle information passing through each level of the gantries is preliminarily identified; The radar and video equipment in the integrated radar and video machine installed on the upper gantry are used to perform secondary recognition on the vehicle targets passing through the upper gantry; specifically, the radar and video equipment in the integrated radar and video machine are used to perform secondary recognition on the vehicle targets passing through the upper gantry respectively, and the matching probability p of each first vehicle target and each second vehicle target is calculated according to the influence of the distance, speed and light intensity between each first vehicle target identified by the radar and each second vehicle target identified by the video equipment. ij : Among them, d ij and ij represents the distance and speed difference between the i-th first vehicle target and the j-th second vehicle target, d ik and ik represents the distance and speed difference between the i-th first vehicle target and the k-th second vehicle target, N V represents the total number of the second vehicles, l ij and l ik represents the influence of light intensity on matching accuracy, α, β and γ represent three weights respectively; the first vehicle target and the second vehicle target that meet the matching probability conditions are identified as the same vehicle target that has been successfully matched; based on the successfully matched vehicle target, the vehicle information that is secondarily identified by the integrated radar and vision device is generated; If a vehicle exists in the preliminary identification information of the upper gantry but does not exist in the secondary identification information of the upper gantry, it is determined that the vehicle is a suspected lower-level vehicle that was misidentified; Adding the misidentified suspected lower-level vehicle to the preliminary identification information of the lower-level gantry in the same latitude and longitude area as the upper-level gantry; Using a radar-visual integrated device installed on the lower gantry, secondary identification is performed on vehicle information passing through the lower gantry; If a vehicle exists in both the added preliminary identification information and the secondary identification information of the lower gantry, the vehicle is determined to be a lower-layer vehicle.

2. The method according to claim 1, characterized in that The radar and video equipment in the integrated radar and visual machine are used to perform secondary recognition on vehicle targets passing through the gantry where the integrated radar and visual machine is located, including: Using the radar in the integrated radar and visual device, the position of the vehicle target passing through the gantry where the integrated radar and visual device is located is identified; According to the relative position and angle between the radar and the road, the identified vehicle position is converted from the radar coordinate system to the road coordinate system as the vehicle target information.

3. The method according to claim 2, characterized in that The step of converting the identified vehicle position from the radar coordinate system to the road coordinate system according to the relative position and angle between the radar and the road comprises: Determining a translation matrix according to the relative position of the radar and the road; Determining a rotation matrix according to an angle between the radar and the road; The vehicle position in the radar coordinate system is translated and rotated according to the translation matrix and the rotation matrix to obtain the vehicle position in the road coordinate system.

4. The method according to claim 1, characterized in that: The radar and video equipment in the integrated radar and visual device are used to perform secondary recognition on vehicle targets passing through the gantry where the integrated radar and visual device is located, including: Using the video equipment in the integrated radar and video machine, take images of the covered road section on one side of the gantry where the integrated radar and video machine is located; The captured images are input into the neural network model to identify the vehicle location and license plate information in the image.

5. The method according to claim 4, characterized in that The method of inputting the captured image into a neural network model to identify the vehicle position and license plate information in the image includes: The captured image is input into the vehicle detection neural network model to identify the vehicle area in the image; Inputting the vehicle area into a license plate detection neural network model to identify the license plate area; The license plate area is input into the license plate information neural network model to identify specific license plate information.

6. A vehicle accurate identification system for upper and lower composite highways, characterized in that: It includes multiple gantries, multiple ETC antennas, multiple radar-visual integrated machines and industrial computers, wherein each layer of the expressway is respectively deployed with at least one gantry, and each gantry is respectively equipped with an ETC antenna and a radar-visual integrated machine; The industrial computer is used to perform the following operations: The radar and video equipment in the integrated radar and video machine installed on the upper gantry are used to perform secondary recognition on the vehicle targets passing through the upper gantry; specifically, the radar and video equipment in the integrated radar and video machine are used to perform secondary recognition on the vehicle targets passing through the upper gantry respectively, and the matching probability p of each first vehicle target and each second vehicle target is calculated according to the influence of the distance, speed and light intensity between each first vehicle target identified by the radar and each second vehicle target identified by the video equipment. ij : Among them, d ij and ij represents the distance and speed difference between the i-th first vehicle target and the j-th second vehicle target, d ik and ik represents the distance and speed difference between the i-th first vehicle target and the k-th second vehicle target, N V represents the total number of the second vehicles, l ij and l ik represents the influence of light intensity on matching accuracy, α, β and γ represent three weights respectively; the first vehicle target and the second vehicle target that meet the matching probability conditions are identified as the same vehicle target that has been successfully matched; based on the successfully matched vehicle target, the vehicle information that is secondarily identified by the integrated radar and vision device is generated; If a vehicle exists in the preliminary identification information of the upper gantry but does not exist in the secondary identification information of the upper gantry, it is determined that the vehicle is a suspected lower-level vehicle that was misidentified; Adding the misidentified suspected lower-level vehicle to the preliminary identification information of the lower-level gantry in the same latitude and longitude area as the upper-level gantry; Using a radar-visual integrated device installed on the lower gantry, secondary identification is performed on vehicle information passing through the lower gantry; If a vehicle exists in both the added preliminary identification information and the secondary identification information of the lower gantry, the vehicle is determined to be a lower-layer vehicle.

Images