CN114663773A - Multi-node cooperative vehicle identification method, device, system and storage medium - Google Patents

Abstract

A multi-node coordinated vehicle identification method, device, system, and storage medium, the multi-node coordinated vehicle identification method comprising: when an unlicensed vehicle enters a venue, acquiring an image of the vehicle to generate first feature information, and , obtain the wireless fidelity media access control information of the vehicle and surrounding areas to generate the first set of wireless fidelity media access control information, generate and save the entry record of the vehicle; two feature information and a second set of Wi-Fi media access control information, determine the entry record of the vehicle that is on the field according to the second feature information and the second set of Wi-Fi media access control information, and according to the set of The entry time of the entry record is used to charge the exiting vehicle. The solution provided by this embodiment can identify the vehicle through the feature information and WiFi MAC, with high traffic efficiency and good user experience.

Description

Multi-node collaborative vehicle identification method and device, system and storage medium

technical field

This paper relates to parking charging technology, especially a multi-node collaborative vehicle identification method and device, system and storage medium.

Background technique

In order to improve the efficiency of vehicle traffic, automatic license plate recognition systems are usually installed at the entrances and exits of large commercial supermarkets, residential and park parking lots. According to the license plate information and combined with the parking time, the system can generate the corresponding parking fee. However, in reality, there will always be situations where the license plate is defaced and the license plate is not hung. At this time, because the license plate information cannot be obtained, it is generally necessary for the driver and passenger to collect the ticket/register, and scan the QR code with the mobile phone to enter and exit the venue. Due to the addition of personnel self-service operations, the process is complex and uncontrollable, affecting traffic efficiency and user experience.

Another solution is to collect the driver's facial image and vehicle type information, associate the driver's facial image and vehicle type information, and start to calculate the parking time and cost. This solution involves the collection of biological information, which has legal risks; in addition, in order to ensure the automatic collection of faces in the car, the system should be equipped with flashing lights, which will cause light pollution to drivers and passengers; and, if the system needs to drive Passengers entering and exiting the venue actively cooperate (such as opening the window probe or getting off the bus), which will affect the non-inductive passing experience of the entrance and exit.

SUMMARY OF THE INVENTION

The embodiments of the present application provide a multi-node coordinated vehicle identification method, device, system, and storage medium, which can improve the efficiency of unlicensed vehicles.

The embodiment of the present application provides a multi-node collaborative vehicle identification method, including:

When an unlicensed vehicle enters the venue, acquiring an image of the vehicle to generate first feature information, and acquiring Wi-Fi media access control information of the vehicle and surrounding areas to generate a first Wi-Fi media access control information set , generate and save an admission record of the vehicle, where the admission record includes identification information assigned to the vehicle, the first feature information, the first Wi-Fi media access control information set, the the entry time of the vehicle;

When an unlicensed vehicle is on the field, acquiring the image of the vehicle that is on the field to generate second feature information, and acquiring the Wi-Fi media access control information of the vehicle and its surroundings to generate the second Wi-Fi media access A set of control information, determining the entry record of the vehicle that has entered the venue according to the second feature information and the second set of Wi-Fi media access control information, and determining the entry record of the exiting vehicle according to the entry time of the entry record. vehicles are billed.

In an exemplary embodiment, the acquiring the image of the vehicle that is on the field generates second feature information, and acquiring the Wi-Fi media access control information of the vehicle that is on the field and its surroundings generates the second Wi-Fi medium The access control information set includes:

Acquire the image of the vehicle that has appeared on the field to generate second feature information, and when the second feature information matches multiple entry records, acquire the wireless fidelity media access control information of the vehicle and surrounding areas that have appeared on the field to generate the a second set of wireless fidelity media access control information;

The method further includes: when the second feature information matches only one entry record, charging the exiting vehicle according to the entry time of the matched entry record.

In an exemplary embodiment, determining the entry record of the vehicle that is on the field according to the second feature information and the second set of Wi-Fi media access control information includes: when the second Wi-Fi media When the access control information set only has an intersection with the first Wi-Fi media access control information set in one of the matched admission records, the admission record with the intersection is the selected admission record. State the entry records of the vehicles that are on the field.

In an exemplary embodiment, the method further includes:

When the unlicensed vehicle is in the field, the third feature information is generated according to the image of the unlicensed vehicle collected by the camera passed by the unlicensed vehicle, and the wireless fidelity media of the vehicle and surrounding The access control information generates a third set of wireless fidelity media access control information, and generates and saves an on-site record of the vehicle, where the on-site record includes the identification information of the camera, the third feature information, the The third set of Wi-Fi media access control information, the acquisition time when the camera acquires the image of the unlicensed vehicle.

In an exemplary embodiment, the method further includes:

When there is an intersection between the second Wi-Fi media access control information set and the first Wi-Fi media access control information set of more than one admission record in the matched multiple admission records, the following is performed: Backtracking step: call the historical video, determine the time when the outgoing vehicle arrives at the current camera from the previous camera, there is no other vehicle similar to the outgoing vehicle in the historical video, and obtain the outgoing vehicle after passing the previous video. The in-field record of the previous camera within the time period of one camera is called the target in-field record, wherein the similarity of two vehicles means that the feature information of the two vehicles satisfies a preset matching condition;

When there is an intersection between the target in-field record and only one of the matched multiple admission records, the exiting vehicle is counted according to the admission time of the intersected admission record. fee.

In an exemplary embodiment, the method further includes:

When there is an intersection between the record in the target field and more than one admission record in the matched multiple admission records, the previous camera is used as the current camera, and the backtracking step is re-executed until the target field is found. A camera whose internal record only has an intersection with one of the matched multiple admission records is called a target camera, or until all cameras passed by the on-field vehicle in the field are found;

When the target camera is found, charging the exiting vehicle according to the entry time of the entry record with the intersection;

When the target camera is not found, it will prompt for manual billing.

In an exemplary embodiment, the first feature information and the second feature information are semi-structured feature vectors generated by a preset neural network model.

An embodiment of the present disclosure provides a multi-node cooperative vehicle identification device, including a memory and a processor, wherein the memory stores a program, and when the program is read and executed by the processor, the program implements any of the foregoing embodiments. A multi-node collaborative vehicle recognition method.

Embodiments of the present disclosure provide a computer-readable storage medium, where one or more programs are stored in the computer-readable storage medium, and the one or more programs can be executed by one or more processors to implement any of the foregoing The multi-node cooperative vehicle identification method described in the embodiment.

Embodiments of the present disclosure provide a multi-node cooperative vehicle identification system, including a camera located at an entrance, a camera located at an exit, and a background server, wherein,

The camera at the entrance is configured to, when an unlicensed vehicle enters the venue, acquire an image of the vehicle, or acquire an image of the vehicle to generate first feature information, and transmit the image or the first feature information and the entry time of the vehicle to the backend server; and acquiring the Wi-Fi media access control information of the vehicle and surrounding areas to generate a first set of Wi-Fi media access control information and send it to the backend server ;

The camera at the exit is configured to, when an unlicensed vehicle exits the venue, acquire an image of the vehicle exiting the venue, or acquire the exiting vehicle to generate second feature information, and send the image or the second characteristic information. sending the feature information to the backend server, and obtaining the Wi-Fi media access control information of the vehicle and its surroundings, generating a second set of Wi-Fi media access control information, and sending it to the backend server;

The background server is configured to receive the image sent by the camera at the entrance to generate the first feature information, or receive the first feature information sent by the camera at the entrance, and receive the first Wi-Fi sent by the camera at the entrance A set of media access control information to generate and save the entry record of the vehicle, where the entry record includes the identification information allocated to the vehicle, the first feature information, and the first Wi-Fi media access control information set, the entry time of the vehicle; and, receiving the image sent by the camera at the exit to generate the second feature information, or, receiving the second feature information sent by the camera at the exit, and receiving the first feature information sent by the camera at the exit. 2. A set of Wi-Fi media access control information, according to the second feature information and the second set of Wi-Fi media access control information to determine the entry record of the vehicle leaving the venue, according to the entry record of the entry record Time is billed for the vehicle that goes out.

Compared with the related art, the present application includes a multi-node coordinated vehicle identification method, device, system, and storage medium, wherein the multi-node coordinated vehicle identification method includes: when an unlicensed vehicle enters a venue, obtaining the vehicle Generate the first feature information from the image of the vehicle, and obtain the Wi-Fi media access control information of the vehicle and surrounding areas to generate the first Wi-Fi media access control information set, and generate and save the entry record of the vehicle. The admission record includes the identification information assigned to the vehicle, the first feature information, the first Wi-Fi media access control information set, and the entry time of the vehicle; when an unlicensed vehicle leaves the venue , acquiring the image of the vehicle that appears on the field to generate second feature information, and acquiring the Wi-Fi media access control information of the vehicle and its surrounding areas that are on the field to generate a second set of Wi-Fi media access control information, according to the The second feature information and the second set of Wi-Fi media access control information determine the entry record of the vehicle that appears, and the vehicle that appears is charged according to the entry time of the entry record. The solution provided by this embodiment can identify vehicles through feature information and WiFi MAC without manual participation (manual registration, code scanning, etc.), high traffic efficiency, good user experience, and can identify unlicensed vehicles of the same model. And there is no need to identify user biometric information, which can avoid legal risks.

Other features and advantages of the present application will be set forth in the description which follows, and in part will be apparent from the description, or may be learned by practice of the present application. Other advantages of the present application may be realized and attained by the approaches described in the specification and drawings.

Description of drawings

The accompanying drawings are used to provide an understanding of the technical solutions of the present application, and constitute a part of the specification. They are used to explain the technical solutions of the present application together with the embodiments of the present application, and do not constitute a limitation on the technical solutions of the present application.

FIG. 1 is a flowchart of a multi-node cooperative vehicle identification method provided by an exemplary embodiment;

FIG. 2 is a schematic diagram of a multi-node cooperative vehicle identification system provided by an exemplary embodiment;

3 is a schematic diagram of the distribution of intelligent IPCs in a multi-node cooperative vehicle identification system provided by an exemplary embodiment;

FIG. 4 is a block diagram of a multi-node cooperative vehicle identification device provided by an exemplary embodiment.

Detailed ways

This application describes a number of embodiments, but the description is exemplary rather than restrictive, and it will be apparent to those of ordinary skill in the art that within the scope of the embodiments described in this application can be There are many more examples and implementations. Although many possible combinations of features are shown in the drawings and discussed in the detailed description, many other combinations of the disclosed features are possible. Unless expressly limited, any feature or element of any embodiment may be used in combination with, or may be substituted for, any other feature or element of any other embodiment.

This application includes and contemplates combinations with features and elements known to those of ordinary skill in the art. The embodiments, features and elements that have been disclosed in this application can also be combined with any conventional features or elements to form unique inventive solutions as defined by the claims. Any features or elements of any embodiment may also be combined with features or elements from other inventive arrangements to form another unique inventive arrangement defined by the claims. Accordingly, it should be understood that any of the features shown and/or discussed in this application may be implemented alone or in any suitable combination. Accordingly, the embodiments are not to be limited except in accordance with the appended claims and their equivalents. Furthermore, various modifications and changes may be made within the scope of the appended claims.

Furthermore, in describing representative embodiments, the specification may have presented methods and/or processes as a particular sequence of steps. However, to the extent that the method or process does not depend on the specific order of steps described herein, the method or process should not be limited to the specific order of steps described. Other sequences of steps are possible, as will be understood by those of ordinary skill in the art. Therefore, the specific order of steps set forth in the specification should not be construed as limitations on the claims. Furthermore, the claims directed to the method and/or process should not be limited to performing their steps in the order written, as those skilled in the art will readily appreciate that these orders may be varied and still remain within the spirit and scope of the embodiments of the present application Inside.

The embodiment of the present disclosure proposes a multi-node collaborative vehicle identification method, which is based on vehicle identification and WiFi wireless sniffing to collaboratively identify vehicles, which can achieve the same traffic efficiency as licensed vehicles, and avoid possible legal risks faced by the face collection scheme. .

FIG. 1 is a flowchart of a multi-node cooperative vehicle identification method provided by an embodiment of the present disclosure. As shown in FIG. 1, the method for identifying unlicensed vehicles and parking fees provided by this embodiment includes:

Step 101: When an unlicensed vehicle enters the venue, acquire an image of the vehicle to generate first feature information, and acquire the wireless fidelity (Wireless Fidelity, WiFi) media access control information (Media Access Control) information of the vehicle and surrounding areas. , MAC) to generate the first WiFi MAC information set, generate and save the entry record of the vehicle, and the entry record includes the identification information assigned to the vehicle, the first feature information, the first WiFi MAC information set, the entry time of the vehicle;

Step 102, when an unlicensed vehicle is on the field, acquire the image of the vehicle that is on the field to generate second feature information, and acquire the WiFi MAC of the vehicle and its surrounding to generate a second set of WiFi MAC information, according to the first set of WiFi MAC information. The second feature information and the second set of WiFi MAC information determine the entry record of the vehicle going out, and the vehicle going out is charged according to the entry time of the entry record.

The solution provided in this embodiment uses feature information and WiFi MAC to identify vehicles without manual participation (manual registration, code scanning, etc.), high traffic efficiency, good user experience, and can identify unlicensed vehicles of the same model, and There is no need to identify user biometric information and legal risks can be avoided.

In this embodiment, the field is a parking lot, and cameras are provided at the entrance and exit of the parking lot for image acquisition; one or more cameras may be provided in the parking lot.

In an exemplary embodiment, the unlicensed vehicle may be detected by license plate location and recognition. When the license plate cannot be recognized, it is an unlicensed vehicle.

In an exemplary embodiment, the method further includes, when a vehicle with a license plate enters the venue, recognizing the license plate of the vehicle, and recording the license plate and the entry time of the vehicle; when the vehicle with a license plate exits the venue, identifying The license plate of the vehicle is charged according to the entry time corresponding to the license plate. The solution provided by this embodiment can directly charge a vehicle with a license plate according to the license plate, which is simple to implement and has high efficiency. However, the embodiment of the present disclosure is not limited to this, and the vehicle with a license plate can be identified according to the feature information and the WIFI MAC information.

In an exemplary embodiment, acquiring the image of the vehicle that has appeared on the scene to generate the second feature information, and acquiring the WiFi MAC of the vehicle that has appeared and the surrounding area to generate the second set of WiFi MAC information includes:

Obtaining the image of the vehicle that has appeared on the field to generate second feature information, and when the second feature information matches multiple entry records, acquiring the WiFi MAC of the vehicle and surrounding areas that have appeared on the field to generate the second WiFi MAC information set ;

The method further includes: when the second feature information matches only one entry record, charging the exiting vehicle according to the entry time of the matched entry record.

Wherein, the admission record matched by the second feature information refers to the admission record in which the first feature information is similar to the second feature information to be matched. set matching conditions. Taking the feature information as the feature vector as an example, when comparing the feature vectors, if the Euclidean distance or the cosine distance between the two is less than or equal to a preset threshold, the two feature vectors are similar. The two feature information is similar includes the case where the two feature information are the same.

In this embodiment, when only one admission record is matched according to the second feature information, identification and charging can be performed without acquiring WiFi MAC information, thereby simplifying charging.

In an exemplary embodiment, the first feature information and the second feature information may be semi-structured feature vectors generated by a preset neural network model. The preset neural network model may be a neural network model obtained through training, and a semi-structured feature vector can be obtained by inputting the acquired image into the preset neural network model. However, the embodiments of the present disclosure are not limited thereto, and the first feature information and the second feature information may be vehicle model information identified based on a neural network model. Using semi-structured feature vectors, in the face of new vehicle models, manufacturers no longer need to import new car data for supervised learning of neural network models, avoiding the trouble of periodically upgrading parking lot equipment, and obtaining semi-structured feature vectors through images Compared with obtaining model information, the calculation amount is lower, the running speed is high, the system hardware requirements are low, and the cost is low.

In an exemplary embodiment, the determining, according to the second feature information and the second set of WiFi MAC information, the entry record of the vehicle that leaves the venue includes: when the second set of WiFi MAC information is only related to the set of all WiFi MAC information; When there is an intersection of the first WiFi MAC information set of one of the matched admission records in one of the admission records, the admission record with the intersection is the admission record of the exiting vehicle. Usually, there are fewer unlicensed vehicles in the parking lot, and the probability of unlicensed vehicles with the same model entering the same parking lot is lower. Therefore, only one entry record is generally matched, so that the vehicle identification can be easily realized.

In an exemplary embodiment, the method further includes:

When the unlicensed vehicle is in the field, the third feature information is generated according to the image of the unlicensed vehicle collected by the camera passed by the unlicensed vehicle, and the WiFi MAC of the vehicle and its surroundings generates the third feature information. Three sets of WiFi MAC information to generate and save an on-site record of the vehicle. The on-site record includes the identification information of the camera, the third feature information, and the third set of WiFi MAC information. The camera collects The acquisition time of the image of the unlicensed vehicle; in this embodiment, when the vehicle is in the venue, the vehicle-related feature information and WiFi MAC information collection are recorded, so that the vehicle can be traced back when multiple entry records are matched when exiting the venue. .

In an exemplary embodiment, the method further includes:

When there is an intersection between the second WiFi MAC information set and the first WiFi MAC information set of more than one entry record among the matched multiple entry records, the following retrospective steps are performed: calling historical recording, determining the Within the time period from the last camera to the current camera of the exiting vehicle, and there is no other vehicle similar to the exiting vehicle in the historical video, obtain the last camera during the time period when the exiting vehicle passed the previous camera The in-field record of the two vehicles is called the target in-field record; wherein, the similarity of the two vehicles means that the feature information of the two vehicles satisfies the preset matching condition; that is, the characteristic information of the two vehicles is similar;

When there is an intersection between the target in-field record and only one of the matched multiple admission records, the exiting vehicle is counted according to the admission time of the intersected admission record. fee.

In an exemplary embodiment, the method further includes:

When there is an intersection between the record in the target field and more than one admission record in the matched multiple admission records, the previous camera is used as the current camera, and the backtracking step is re-executed until the target field is found. A camera whose internal record only has an intersection with one of the matched multiple admission records is called a target camera, or until all cameras passed by the on-field vehicle in the field are found;

When the target camera is found, charging the exiting vehicle according to the entry time of the entry record with the intersection;

When the target camera is not found, it will prompt for manual billing.

In the solution provided by the embodiments of the present disclosure, when there may be vehicles of the same model near the exiting vehicle, since the probability of the same vehicle reaching the exit through the same route is low, the exiting vehicle can be determined in a retrospective manner to realize the Identification and billing.

In an exemplary embodiment, one or a combination of the following may be used for prompting: voice prompt, display screen prompt.

In an exemplary embodiment, when there is an intersection between the second WiFi MAC information set and the first WiFi MAC information set of more than one entry record in the matched multiple entry records, prompt manual billing ; Or, after performing the backtracking step once, when the target in-field record has an intersection with multiple admission records, a manual billing is prompted.

The technical solutions of the embodiments of the present disclosure are further described below through an example.

In the embodiment of the present disclosure, when the vehicle enters the field, the camera collects the image of the vehicle, generates a semi-structured feature vector under the corresponding neural network model for the vehicle according to the collected image, and uses the WiFi sniffer (WiFiSniffer) built in the camera. The module collects the WiFi MAC information set around the moment, and sends the vehicle's feature vector, WiFi MAC information set and entry time to the backend server, and the background server saves the feature vector, WiFi MAC information set and entry time as the vehicle's entry record into the database; if two similar unlicensed vehicles (vehicles with similar feature information) are entered, the feature vectors and WiFi MAC information sets of different vehicles are stored separately.

When the vehicle leaves the venue, if an unlicensed vehicle is detected, it will cooperate with the background server to determine whether there are other vehicles similar to the vehicle. If it does not exist (that is, only one entry record is matched), it will be directly matched and charged. If it exists (that is, multiple entry records are matched), the WiFi MAC information set at the moment is collected. If the WiFi MAC information set only intersects with the WiFi MAC information set of one of the multiple matching entry records, then Vehicles can be identified and billed based on the vehicle's entry time.

If the collected WiFi MAC information set has an intersection with the WiFi MAC information set in the matched multiple entry records, it cannot be determined which vehicle among similar vehicles is. Whether there are other vehicles similar to the vehicle that appeared in the period from the camera to this camera, if not, send the feature vector of the appearance to the previous camera, and ask the previous camera to feed back the WiFiMAC information collected when the vehicle passed it. Collection; if the WiFi MAC information collection fed back by the last camera has an intersection with only one admission record, charging may be performed according to the admission time in the admission record with the intersection. If the WiFi MAC information set fed back by the last camera still intersects with multiple admission records, perform the same operation until the WiFi MAC information set fed back by a camera only intersects with one admission record The admission time in the field record is billed.

FIG. 2 is a schematic diagram of a multi-node cooperative vehicle identification system provided by an exemplary implementation. As shown in FIG. 2 , the multi-node coordinated vehicle identification system includes: a smart camera (or smart IPC), a display screen, and a background server. The display screen is connected to the intelligent IPC (which can be connected by wire or wirelessly), or the display screen and the intelligent IPC can be integrated, wherein:

The intelligent IPC can recognize license plate numbers and generate semi-structured feature vectors under the neural network model for unlicensed vehicles. A WiFi MAC collector (that is, WiFi Sniffer module) can be integrated in the intelligent IPC, which can collect information of wireless devices near the intelligent IPC, such as the MAC addresses of the mobile phones of car drivers and passengers. Intelligent IPCs can be installed at the entrances and exits of the parking lot and at each intersection inside.

The display screen can be placed at the entrance and exit of the parking lot. The display screen at the entrance can prompt the unlicensed car to turn on WiFi, so as to achieve the purpose of authorizing the WiFi Sniffer module to wirelessly collect the WiFiMAC of the driver's mobile phone, and can also display the recognized license plate number for the licensed car; the display screen at the exit can prompt The parking time of the vehicle and the amount charged. In another exemplary embodiment, a display screen may not be provided, and a voice prompt may be used.

The background server is configured to receive, store, and retrieve matching feature vectors uploaded by the intelligent IPC, and to support analyzing and filtering the WiFi MAC address provided by the WiFi Sniffer module of the intelligent IPC to identify the vehicle.

In another exemplary embodiment, the generation of the feature vector of the vehicle can be undertaken by the background server, and the intelligent IPC only needs to provide the image or video stream of the vehicle, which is not limited in the solution.

The multi-node coordinated vehicle identification method provided in this embodiment includes:

Step 301, after the vehicle enters the field, the intelligent IPC obtains the image generation feature information and obtains the MAC set;

When the vehicle enters the venue, the smart camera at the entrance generates the corresponding semi-structured feature vector under the neural network model for the vehicle, that is, the first feature vector, and collects the WiFi MAC around the smart camera through the WiFi Sniffer that comes with the smart camera. The first WiFi MAC information set is generated from the information, and the first feature vector of the vehicle and the first WiFi MAC information set are sent to the database of the background server for storage; if two similar unlicensed vehicles are entered, the vehicle feature vector and WiFi MAC information Collections and binding relationships are stored separately.

The intelligent IPC at the entrance obtains the vehicle's feature vector and WiFiMAC information set. After classifying and locating the objects of interest (such as cars/people) in the video image, the corresponding semi-structured feature vectors are extracted through the deep neural network. The feature vector can be a real number vector of 1*128 or 1*256. Different types of vehicle targets will obtain different feature vectors after being mapped by a deep neural network. While collecting video images, the intelligent IPC also obtains the surrounding WiFi MAC information through the WiFiSniffer module. The intelligent IPC reports feature information, Wifi MAC information set and admission time to the background server.

Assuming that an unlicensed vehicle A0 enters the park, the backend server generates a record of the vehicle captured by the intelligent IPC according to the information reported by the intelligent IPC at the entrance, and stores the vehicle's snapshot time (that is, the entry time), feature vector and Wifi MAC information set. called the entry record. As shown in Table 1 below, the admission record may include: vehicle identification information (that is, vehicle target ID) assigned to the vehicle by the background server, IPC number (that is, the number of the IPC that reports the feature information and Wifi MAC information set), feature vector, MAC information (that is, WiFi MAC information set), entry time.

Table 1 Admission Record Table 1

vehicle target ID IPC number Feature vector MAC information Admission time RecordID1 1 Eigenvector A a1, a2, c, d ET1

In Table 1, the actual MAC addresses of the mobile phones of the drivers and passengers in the unlicensed car A0 are a1 and a2. The other MAC addresses c and d belong to the mobile phones of other people nearby collected by the intelligent IPC at the entrance, and are interference addresses.

Other intelligent IPCs in the field will collect the image of the vehicle to generate the feature vector and the WiFi MAC when the vehicle is passing, and report the feature vector, the WiFi MAC when the vehicle is passing, and the collection time of collecting the image of the vehicle to the back-end server, and the back-end server will generate the field record. . As shown in Table 2, the field record may include: IPC number (that is, the number of the IPC that reports the feature information and the WiFi MAC information set), feature vector, MAC information (that is, the WiFi MAC information set), and collection time. In an exemplary embodiment, the vehicle identification information (that is, the vehicle target ID) may be included in the field record, wherein the vehicle identification information may be determined according to the vehicle identification information in the admission record matched by the feature vector and the WiFi MAC information set. , that is, after finding that there is only one entry record that is similar to the reported feature vector and has an intersection with the reported WiFi MAC information set, the vehicle identification information of the entry record is used as the identification information recorded in the venue; if If multiple entry records are found, the field can be left blank. Taking Table 2 as an example, the eigenvectors in Table 2 are the same as the eigenvectors in Table 1, and the MAC information in Table 2 and the MAC information in Table 1 overlap (both include MAC addresses a1, a2), then the table The vehicle object ID in 1 is used as the vehicle object ID in the field record in Table 2.

Table 2 In-field record sheet

vehicle target ID IPC number Feature vector MAC information Acquisition time RecordID1 5 Eigenvector A a1, a2, e, f ET5

Due to the matching of vehicle eigenvectors, there is no attribute retrieval method of finding the license plate of the vehicle, and the response is fast. In another exemplary embodiment, when the vehicle enters the venue, the intelligent IPC at the entrance locates and recognizes the license plate, and if there is a license plate, the parking time and fee are determined by the license plate location and identification of the vehicle; Then, the semi-structured feature vector under the preset neural network model is generated for the vehicle target. Considering that there are few unlicensed cars after all, and there are not many feature vectors, the real-time performance of matching retrieval can be guaranteed.

Step 302 , when the unlicensed car leaves the venue, if the characteristic information of the unlicensed car matches only one entry record, charging processing is performed according to the entry time in the matched entry record.

When the vehicle leaves the field, if an unlicensed vehicle is detected, it will cooperate with the backend server to determine whether there are other vehicles similar to the vehicle's feature vector.

When an unlicensed vehicle leaves the field, the intelligent IPC at the exit collects video images, and compares the collected feature vector of the vehicle with the feature vector of the unlicensed vehicle that has not yet left the field stored in the background server. If only one entry record is matched through the feature vector comparison, the parking fee is calculated directly according to the time difference between the exit time and the entry time, displayed on the display screen, and the driver is asked to pay the fee.

Step 303, when the unlicensed car exits, if the feature information of the unlicensed car matches multiple entry records, obtain the WiFi MAC information set of the unlicensed car and its surroundings at the current moment (that is, the WiFi MAC information set around the smart IPC at the exit). ), when the WiFi MAC information set only has an intersection with one of the matched multiple entry records, the current exit time is billed according to the entry time in the entry record with the intersection.

When two similar vehicles enter the venue, the intelligent IPC at the entrance will generate an entry record for each of them, as shown in Table 3 and Table 4, respectively. When the eigenvectors of the two vehicles are the same or similar, the vehicles are considered to be similar. In this embodiment, the feature vectors of the two are the same.

Table 3 Admission Record Table 2

vehicle target ID IPC number Feature vector MAC information Admission time RecordID1 1 Eigenvector A a1, a2, c, d ET1

Table 4 Admission Record Table 3

vehicle target ID IPC number Feature vector MAC information Admission time RecordID9 1 Eigenvector A a9, e, f ET9

When an unlicensed vehicle leaves the field, the export intelligent IPC collects video images, and compares the collected vehicle feature vector with the unlicensed vehicle feature vector stored in the background server that has not yet left the field. Through the feature vector comparison, if two entry records are matched, the WiFi MAC information set around the moment is collected. If the WiFi MAC information set only has an intersection with the WiFi MAC information set of one of the two matching admission records, the charging process is performed according to the admission time in the admission record with the intersection. For example, the collected WiFi MAC information set is {a1, g, h}, and only has an intersection with the admission records shown in Table 3, it means that the vehicle is the vehicle corresponding to the admission records shown in Table 3. The entry time and the current exit time will be billed.

Step 304, when the unlicensed car comes out, if the feature information of the unlicensed car matches multiple entry records, obtain the WiFi MAC information set of the unlicensed car and its surroundings at the current moment (that is, the WiFi MAC information set around the smart IPC at the exit). ), when there is an intersection between the WiFi MAC information set and more than one matching entry record, the entry record is determined by backtracking.

If the collected WiFi MAC information set intersects with multiple admission records, it cannot be determined which vehicle among the similar vehicles is. From the current moment back to the historical video, it is judged whether other similar vehicles have appeared during the period from the intelligent IPC at the last intersection of the intelligent IPC at the exit to the intelligent IPC at the exit. The feature vector collected by the intelligent IPC at the exit is sent to the intelligent IPC at the previous intersection, and the intelligent IPC at the previous intersection is asked to feed back the WiFi MAC information set collected by the intelligent IPC at the previous intersection when the vehicle passes through it. If the WiFi MAC information set collected by the intelligent IPC at the previous intersection only has an intersection with the WiFi MAC information set of one entry record among the matched multiple entry records, then the corresponding WiFi MAC information set according to the intersection is determined. Admission records are billed.

FIG. 3 is a schematic diagram of intelligent IPC distribution provided by an exemplary embodiment. The intelligent IPC collection at the exit found that the WiFi MAC information set at this time is {a1, a9, h, i}, which intersects with the WiFi MAC information sets of RecordID1 and RecordID9, so it is impossible to determine which vehicle the current vehicle is. The reason could be that another car was also parked nearby.

At this time, the intelligent IPCs at other intersections can make auxiliary judgments. The intelligent IPC at the exit obtains the time point time-A (or time period) of the current unlicensed vehicle passing the intelligent IPC at the previous intersection by calling the historical video and backtracking the historical video from the current moment, and judges the Whether there has been a similar vehicle during the period when the vehicle passed under the intelligent IPC at the previous intersection and arrived at this intelligent IPC, if not, send the vehicle's feature vector and time-A to the camera at the previous intersection ①, number 11 intelligent IPC.

Intelligent IPC No. 11 searches for the WiFi MAC information set collected near time-A. If the WiFi MAC information set collected by intelligent IPC No. 11 near time-A is only matched with one of the multiple admission records If there is an intersection between the records, identify the vehicle according to the entry record with the intersection, such as RecordID1, send the RecordID1 to the intelligent IPC at the exit, and charge the vehicle corresponding to the RecordID1.

The intelligent IPC at the exit determines that no other similar vehicles have passed the camera at the previous intersection to the camera through historical video before sending a query request, which can avoid judgment errors. If there are other vehicles that are similar to the vehicles that appear on the field, you can directly prompt manual billing.

If the previous intelligent IPC, that is, intelligent IPC No. 11, finds that the WiFi MAC information set of time-A still intersects with multiple admission records, the same backtracking operation is performed. Based on the feature vector of the unlicensed vehicle received from the intelligent IPC at the exit, the No. 11 intelligent IPC traces back the historical video from time-A to determine which intelligent IPC the unlicensed vehicle has passed through in the history, such as intelligent IPC12 , then find the time point time-B at which the unlicensed vehicle passes through the intelligent IPC12 through video playback; and from time-B to time-A, there are no other vehicles similar to the exiting vehicle in the historical video passing by, please intelligent IPC12 to find the time -A collection of WiFi MAC information collected near time B. If the intelligent IPC12 finds that the WiFi MAC information set collected near time-B only has an intersection with one of the multiple entry records matched with the exiting vehicle, it will identify the vehicle according to the intersection entry record, for example It is RecordID1, send RecordID1 to the intelligent IPC at the exit, and charge the vehicle corresponding to RecordID1.

If the intelligent IPC 12 still finds that the WiFi MAC information set collected by time-B has an intersection with the entry records of multiple vehicles corresponding to the eigenvectors, it will perform similar operations with reference to the intelligent IPC 11 .

It is extremely unlikely that multiple similar unlicensed vehicles will travel the same path at the same time. As long as this extreme situation does not occur, the above solution is feasible. If there is an extreme situation, that is, a camera finds that it cannot find a suitable camera at the previous intersection, it will feed back a message of "unable to assist" to the intelligent IPC at the exit, and the intelligent IPC at the exit will prompt "manual billing" on the display. .

The solution provided by this embodiment does not require manual registration/scanning code or face collection, and can cope with the problem of entering and exiting parking of multiple unlicensed vehicles of the same model in real life, so that unlicensed vehicles can be implemented the same as licensed vehicles. traffic efficiency.

As shown in FIG. 4 , an embodiment of the present disclosure provides a multi-node cooperative vehicle identification device 40 , including a memory 410 and a processor 420 , the memory 410 stores a program, and the program is read by the processor 420 When executed, the multi-node cooperative vehicle identification method described in any of the above embodiments is implemented.

The present disclosure is a computer-readable storage medium, where one or more programs are stored in the computer-readable storage medium, and the one or more programs can be executed by one or more processors, so as to implement any of the above-mentioned embodiments. The multi-node cooperative vehicle recognition method described above.

Embodiments of the present disclosure provide a multi-node cooperative vehicle identification system, including a camera located at an entrance, a camera located at an exit, and a background server, wherein,

The camera at the entrance is configured to, when an unlicensed vehicle enters the venue, acquire an image of the vehicle, or acquire an image of the vehicle to generate first feature information, and transmit the image or the first feature information and the entry time of the vehicle to the background server; and obtaining the WiFi MAC of the vehicle and surrounding areas to generate a first WiFiMAC information set and send it to the background server;

The camera at the exit is configured to, when an unlicensed vehicle exits the venue, acquire an image of the vehicle exiting the venue, or acquire the exiting vehicle to generate second feature information, and send the image or the second characteristic information. The feature information is sent to the background server, and the WiFi MAC of the vehicle and the surrounding area is obtained to generate a second WiFi MAC information set and send to the background server;

The background server is configured to receive the image sent by the camera at the entrance to generate the first feature information, or receive the first feature information sent by the camera at the entrance, and receive the first WiFi MAC information sent by the camera at the entrance collection, generate and save the entry record of the vehicle, the entry record includes the identification information allocated to the vehicle, the first feature information, the first WiFi MAC information set, and the entry time of the vehicle; and, receiving the image sent by the camera at the exit to generate the second feature information, or, receiving the second feature information sent by the camera at the exit, and receiving the second set of WiFi MAC information sent by the camera at the exit, according to the The second feature information and the second set of WiFi MAC information determine the entry record of the vehicle that leaves the venue, and the vehicle that exits the venue is charged according to the entry time of the entry record.

In an exemplary embodiment, the multi-node cooperative vehicle identification system may further include at least one camera located within the field. For the implementation of the camera located in the field, please refer to the foregoing embodiment, and details are not repeated here.

Those of ordinary skill in the art can understand that all or some of the steps in the methods disclosed above, functional modules/units in the systems, and devices can be implemented as software, firmware, hardware, and appropriate combinations thereof. In a hardware implementation, the division between functional modules/units mentioned in the above description does not necessarily correspond to the division of physical components; for example, one physical component may have multiple functions, or one function or step may be composed of several physical components Components execute cooperatively. Some or all of the components may be implemented as software executed by a processor, such as a digital signal processor or microprocessor, or as hardware, or as an integrated circuit, such as an application specific integrated circuit. Such software may be distributed on computer-readable media, which may include computer storage media (or non-transitory media) and communication media (or transitory media). As is known to those of ordinary skill in the art, the term computer storage media includes both volatile and nonvolatile implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data flexible, removable and non-removable media. Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disk (DVD) or other optical disk storage, magnetic cartridges, magnetic tape, magnetic disk storage or other magnetic storage devices, or may Any other medium used to store desired information and which can be accessed by a computer. In addition, communication media typically embodies computer readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and can include any information delivery media, as is well known to those of ordinary skill in the art .

Claims (10)

1. A multi-node cooperative vehicle identification method is characterized by comprising the following steps:

when a unlicensed vehicle enters, acquiring an image of the vehicle to generate first characteristic information, acquiring wireless fidelity media access control information of the vehicle and the periphery to generate a first wireless fidelity media access control information set, generating and storing an entry record of the vehicle, wherein the entry record comprises identification information distributed to the vehicle, the first characteristic information, the first wireless fidelity media access control information set and the entry time of the vehicle;

when an unlicensed vehicle leaves, acquiring an image of the vehicle leaving the vehicle to generate second characteristic information, acquiring wireless fidelity media access control information of the vehicle leaving the vehicle and the periphery to generate a second wireless fidelity media access control information set, determining an entrance record of the vehicle leaving the vehicle according to the second characteristic information and the second wireless fidelity media access control information set, and charging the vehicle leaving the vehicle according to the entrance time of the entrance record.

2. The multi-node cooperative vehicle identification method according to claim 1, wherein obtaining an image of the vehicle at the departure generates second feature information, and obtaining the wireless fidelity medium access control information of the vehicle and surroundings at the departure generates a second set of wireless fidelity medium access control information comprises:

acquiring an image of the vehicle which is on the scene to generate second characteristic information, and acquiring wireless fidelity media access control information of the vehicle which is on the scene and the periphery to generate a second wireless fidelity media access control information set when the second characteristic information is matched with a plurality of entrance records;

the method further comprises the following steps: and when the second characteristic information is only matched with one entry record, charging the vehicles for leaving according to the entry time of the matched entry record.

3. The multi-node coordinated vehicle identification method according to claim 2, wherein determining the entry record of the departing vehicle according to the second characteristic information and the second set of wireless fidelity medium access control information comprises: and when the second wireless fidelity medium access control information set only has intersection with the first wireless fidelity medium access control information set of one of the matched entry records, the entry record with the intersection is the entry record of the vehicle which is out of the field.

4. The multi-node coordinated vehicle identification method according to claim 2, further comprising:

when the unlicensed vehicle is in the field, generating third characteristic information according to the image of the unlicensed vehicle acquired by a camera passed by the unlicensed vehicle, generating a third wireless fidelity media access control information set according to the vehicle and the wireless fidelity media access control information around the vehicle, generating and storing an in-field record of the vehicle, wherein the in-field record comprises the identification information of the camera, the third characteristic information and the third wireless fidelity media access control information set, and the camera acquires the acquisition time of the image of the unlicensed vehicle.

5. The multi-node coordinated vehicle identification method according to claim 4, further comprising:

when the second wireless fidelity media access control information set is intersected with the first wireless fidelity media access control information set of more than one entry record in the matched entry records, executing the following backtracking steps: calling a history video, determining that other vehicles similar to the vehicle leaving the field do not exist in the history video within the time when the vehicle leaving the field arrives at the current camera from the last camera, and acquiring a field record of the vehicle leaving the field in the period when the vehicle passes through the last camera, wherein the field record is called a target field record, and the two vehicles are similar, namely the characteristic information of the two vehicles meets a preset matching condition;

and when the target in-field record only has intersection with one of the matched plurality of entrance records, charging the vehicles which come out according to the entrance time of the entrance records with the intersection.

6. The multi-node coordinated vehicle identification method according to claim 5, further comprising:

when the intersection exists between the target in-field record and more than one of the matched multiple entry records, taking the previous camera as the current camera, and re-executing the backtracking step until a camera, called a target camera, is found out, wherein the intersection exists between the target in-field record and only one of the matched multiple entry records, or all cameras, through which the vehicle leaving the field passes, in the field are found out;

when the target camera is found, charging the vehicles which are out of the field according to the entrance time of the entrance records with the intersection;

and when the target camera is not found, prompting to carry out manual charging.

7. The multi-node cooperative vehicle identification method according to any one of claims 1 to 6, wherein the first feature information and the second feature information are semi-structured feature vectors generated by a preset neural network model.

8. A multi-node cooperative vehicle identification apparatus comprising a memory and a processor, the memory storing a program which, when read and executed by the processor, implements the multi-node cooperative vehicle identification method according to any one of claims 1 to 7.

9. A computer-readable storage medium storing one or more programs, the one or more programs being executable by one or more processors to implement the multi-node cooperative vehicle identification method according to any one of claims 1 to 7.

10. A multi-node cooperative vehicle identification system is characterized by comprising a camera positioned at an entrance, a camera positioned at an exit and a background server, wherein,

the camera at the entrance is configured to acquire an image of the vehicle when an unlicensed vehicle enters the entrance, or acquire the image of the vehicle to generate first characteristic information, and send the image or the first characteristic information and the entrance time of the vehicle to the background server; acquiring wireless fidelity media access control information of the vehicle and the periphery, generating a first wireless fidelity media access control information set, and sending the first wireless fidelity media access control information set to the background server;

the camera at the exit is configured to acquire an image of a vehicle on the scene when the unlicensed vehicle is on the scene, or acquire second feature information generated by the vehicle on the scene, send the image or the second feature information to the background server, and acquire wireless fidelity media access control information of the vehicle on the scene and the periphery, generate a second wireless fidelity media access control information set and send the second wireless fidelity media access control information set to the background server;

the background server is configured to receive an image sent by a camera at an entrance to generate first characteristic information, or receive the first characteristic information sent by the camera at the entrance, receive a first wireless fidelity medium access control information set sent by the camera at the entrance, generate and store an entry record of a vehicle, wherein the entry record comprises identification information allocated to the vehicle, the first characteristic information, the first wireless fidelity medium access control information set and entry time of the vehicle; and receiving an image sent by a camera at an exit to generate second characteristic information, or receiving the second characteristic information sent by the camera at the exit, receiving a second wireless fidelity medium access control information set sent by the camera at the exit, determining an entry record of a vehicle which is out of the field according to the second characteristic information and the second wireless fidelity medium access control information set, and charging the vehicle which is out of the field according to the entry time of the entry record.

Images