Vehicle driving track analysis method based on license plate recognition

Abstract

The invention discloses a vehicle driving track analysis method based on license plate recognition, and relates to the technical field of intelligent traffic, and the method comprises the following steps: installing and configuring license plate recognition equipment at a preset road position in advance, and collecting a license plate image containing license plate information in real time; preprocessing the collected license plate image, wherein the preprocessing comprises image enhancement, denoising and tilt correction; inputting the preprocessed license plate image into a preset trained license plate character recognition model, and outputting recognized license plate characters; and performing recognition processing according to the recognized license plate characters. The license plate can be accurately recognized in a complex environment, the reliability of data acquisition is improved, a large amount of vehicle data can be efficiently processed, the vehicle driving track can be accurately calculated, errors are reduced, the vehicle driving track can be rapidly and accurately generated, and timely and effective support is provided for traffic management and security application.

Description

Technical Field

[0001] The present invention relates to the field of intelligent transportation technology, and in particular to a vehicle driving trajectory analysis method based on license plate recognition. Background Art

[0002] In today's intelligent transportation systems, accurate analysis of vehicle trajectories is crucial. With the increasing complexity of urban traffic and the growing demand for security, acquiring vehicle trajectories and extracting valuable information from them has become a key means of resolving traffic congestion and ensuring traffic safety.

[0003] Currently, traditional vehicle trajectory analysis methods have numerous limitations. Some rely on manual recording or simple sensing equipment, resulting in low efficiency and accuracy, making them unable to meet the demands of processing large amounts of vehicle data. Existing license plate recognition technologies also face challenges in recognizing vehicles accurately in complex environments, such as low lighting and damaged license plates. Furthermore, a lack of efficient data processing and analysis methods for the large amounts of collected vehicle data makes it difficult to quickly and accurately generate vehicle trajectories, hindering timely and effective support for traffic management and security applications. For example, in traffic management, traditional traffic flow monitoring methods rely on manual counting or simple sensing equipment, making it difficult to accurately capture vehicle trajectories and detailed information. This results in an inability to accurately and timely understand real-time traffic flow changes on different road sections, lacking a robust basis for developing traffic signal timing plans. For example, in some congested sections during morning and evening rush hours, the inability to obtain detailed vehicle trajectories and traffic flow patterns makes it difficult to adjust signal durations appropriately, exacerbating traffic congestion and reducing vehicle traffic efficiency. Furthermore, during road planning and construction, due to a lack of vehicle trajectory data, new road plans often fail to fully consider actual traffic demand. This results in low utilization rates for some newly built roads and continued congestion in congested sections. Furthermore, in the security field, when tracking suspected vehicles, existing monitoring technologies can only capture instantaneous information about the vehicle at a specific monitoring point, failing to consistently capture its trajectory and movement patterns. This makes it difficult for security departments to investigate cases and quickly determine the whereabouts and possible destinations of the suspected vehicle. For example, in some criminal cases, the suspected vehicle travels between multiple monitoring areas. Due to the lack of effective information correlation between monitoring points, it is impossible to accurately map its route, delaying the best opportunity to solve the case.

[0004] Currently, no effective solutions have been proposed for the problems in related technologies. Summary of the Invention

[0005] In response to the problems in the related art, the present invention proposes a vehicle driving trajectory analysis method based on license plate recognition to overcome the above technical problems existing in the existing related art.

[0006] The technical solution of the present invention is achieved as follows:

[0007] A vehicle driving trajectory analysis method based on license plate recognition includes the following steps:

[0008] Install license plate recognition equipment at preset road locations in advance to collect license plate images containing license plate information in real time;

[0009] Preprocess the collected license plate images, including image enhancement, denoising, and tilt correction;

[0010] Input the pre-processed license plate image into the preset trained license plate character recognition model to output the recognized license plate characters;

[0011] Performing recognition processing based on the recognized license plate characters, wherein the recognition processing at least includes: using a hash table duplication check algorithm to check for data duplication, performing legitimacy verification according to a compiled license plate number legitimacy verification program, and converting data collected from different devices into a unified JSON format;

[0012] The license plate characters processed through recognition are associated with trajectory generation, and the vehicle's driving trajectory is presented using the Geographic Information System (GIS) based on the vehicle's driving trajectory.

[0013] Furthermore, the data duplication check using a hash table duplication check algorithm includes the following steps:

[0014] For each piece of vehicle data collected, the license plate number is used as the hash key, its hash value is calculated and stored in the hash table;

[0015] When processing new data, first calculate the hash value of its license plate number and check whether data with the same hash value already exists in the hash table. If so, compare the license plate number and other key information to determine whether it is a duplicate record. If so, remove it.

[0016] Furthermore, the legitimacy verification is performed based on the compiled license plate number legitimacy verification program, including: presetting the encoding rules for the license plate number, and matching the license plate number through regular expressions to check whether it complies with the corresponding rules. Those that do not comply with the rules are removed as misidentification or abnormal data.

[0017] Furthermore, the data collected by different devices are converted into a unified JSON format, including: presetting a data conversion script, and converting the data fields into naming specifications and data types according to the preset JSON data structure.

[0018] Furthermore, the license plate characters processed by the recognition process are associated with the track generation, including the following steps:

[0019] Using license plate characters as key identifiers, we traverse all collected vehicle data records to obtain vehicle data belonging to the same vehicle at different times and locations.

[0020] For the vehicle data with the same license plate characters that have been filtered out, sort them in ascending order according to the collection time;

[0021] According to the sorted data, the time interval and driving distance between adjacent data points are calculated in sequence, and the driving direction is obtained;

[0022] The map data is abstracted into a directed graph data structure, and the shortest path calculation is performed. Based on the calculated shortest path, the shortest paths corresponding to each data point are connected to form the vehicle's driving trajectory.

[0023] Furthermore, the time interval between adjacent data points is calculated, including the following steps:

[0024] The acquisition time of two adjacent calibration data points is t1 and t2 respectively, and the time interval Δt is expressed as:

[0025] Δt=t2-t1;

[0026] Among them, t1 and t2 represent the timestamps of a specific moment respectively.

[0027] Furthermore, the travel distance between adjacent data points is calculated, including the following steps:

[0028] The longitude and latitude of the two data points are calibrated as (lat1, lon1) and (lat2, lon2), the radius of the earth is R, and the travel distance d is expressed as:

[0029]

[0030] Furthermore, obtaining the driving direction includes the following steps:

[0031] According to the calculated time interval and travel distance, the speed calculation formula is used to calculate the speed of the vehicle between adjacent data points, which is expressed as:

[0032]

[0033] Among them, v represents speed, Δs represents travel distance, and Δt represents time interval;

[0034] The vehicle's direction of travel is determined based on the positive or negative speed and the direction information on the map. On the map, north is at the top, south is at the bottom, west is on the left, and east is on the right. If the speed is positive, combined with the map direction, it is determined that the vehicle is moving forward; if the speed is negative, it indicates data abnormality.

[0035] Furthermore, the shortest path calculation is performed, including the following steps:

[0036] Let the current node be u, its adjacent node be v, the edge weight from u to v be w(u,v), the distance from the source node to u be d(u), and the distance from the source node to v after the update be d(v), which can be expressed as:

[0037] d(v)=min(d(v),d(u)+w(u,v).

[0038] Beneficial effects of the present invention:

[0039] 1. The present invention pre-installs and configures license plate recognition equipment at preset road locations to collect license plate images containing license plate information in real time; inputs the pre-processed license plate images into a preset trained license plate character recognition model to output the recognized license plate characters; performs recognition processing based on the recognized license plate characters, generates associated trajectories for the recognized license plate characters, and uses a geographic information system (GIS) to present the vehicle's driving trajectory based on the vehicle's driving trajectory, thereby realizing vehicle driving trajectory analysis based on license plate recognition. This method can not only accurately identify license plates in complex environments and improve the reliability of data collection, but also efficiently process large amounts of vehicle data, accurately calculate vehicle driving trajectories, reduce errors, and quickly and accurately generate vehicle driving trajectories, providing timely and effective support for traffic management and security applications.

[0040] 2. Through precise analysis of vehicle trajectories, the present invention can obtain real-time traffic flow data for different road sections and gain a deep understanding of how traffic flow changes over time. This provides a powerful basis for optimizing traffic signal timing, allowing traffic management departments to flexibly adjust signal light durations based on actual traffic flow. Based on a large amount of vehicle trajectory data, traffic planning departments can clearly understand road usage, including which sections have heavy traffic and which sections are prone to congestion. This data can help rationally plan new roads and optimize traffic rules on existing roads, such as setting up tidal lanes and adjusting lane divisions, making the layout of transportation facilities more scientific and reasonable, and better meeting the needs of urban transportation development.

[0041] At the same time, this method allows security departments to quickly track the travel routes and locations of suspected vehicles by simply entering their license plate numbers. Combined with other surveillance data, such as checkpoint monitoring and video surveillance, this method allows for a comprehensive analysis of the suspect vehicle's movements, accurately determining its likely destinations or crime zones, and providing key clues for solving cases. This significantly improves security department efficiency and case-solving success rates, safeguarding social security and stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0043] Figure 1 The figure is a flow chart of a vehicle driving trajectory analysis method based on license plate recognition according to an embodiment of the present invention. DETAILED DESCRIPTION

[0044] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention are within the scope of protection of the present invention.

[0045] According to an embodiment of the present invention, a vehicle driving trajectory analysis method based on license plate recognition is provided.

[0046] like Figure 1 As shown, the vehicle driving trajectory analysis method based on license plate recognition according to an embodiment of the present invention includes the following steps:

[0047] Install license plate recognition equipment at preset road locations in advance to collect license plate images containing license plate information in real time;

[0048] This technical solution involves installing license plate recognition equipment at key road locations, such as intersections, toll booths, and parking lot entrances and exits. Cameras must have a resolution of at least 5 megapixels, a noise level below 5%, and be equipped with starlight-level sensors and wide dynamic range technology. Through field measurements and simulated installations, the camera's position and angle are optimally adjusted to ensure that the license plate accounts for at least 30% of the image. Adaptive fill light technology is also installed to automatically adjust the brightness based on ambient light.

[0049] Preprocess the collected license plate images, including image enhancement, denoising, and tilt correction;

[0050] Image enhancement uses a histogram equalization algorithm. This algorithm calculates the image's grayscale distribution, evens out the grayscale histogram, expands the image's dynamic range, enhances image contrast, and highlights license plate features. For an 8-bit grayscale image, the algorithm calculates the number of pixels at each grayscale level to generate a grayscale histogram. The grayscale value of each pixel is then remapped using the cumulative distribution function to achieve image enhancement.

[0051] Denoising uses a combination of median and Gaussian filtering to remove noise. For each pixel, the median filter takes the median of the pixel values within its neighborhood, such as a 3×3 or 5×5 window, as the new value for that pixel, effectively removing salt and pepper noise. The Gaussian filter removes Gaussian noise by performing a weighted average of the neighboring pixels using a Gaussian function. For a noise-contaminated image, the median filter first removes obvious salt and pepper noise points, and then the Gaussian filter smoothes the image for a clearer image.

[0052] Tilt correction is based on a Hough transform algorithm. The Hough transform is a method for detecting straight lines in an image. For a license plate image, it is first converted into a binary image, and then straight lines are detected in Hough space. By calculating the parameters of the lines in Hough space, the line on which the license plate edge lies is found, and the tilt angle of the license plate is calculated. Finally, a rotation transformation is performed to correct the license plate to a horizontal position.

[0053] The preprocessed license plate image is input into the preset trained license plate character recognition model to output the recognized license plate characters.

[0054] This technical solution uses a license plate character recognition model based on a convolutional neural network (CNN) for license plate character recognition. The details are as follows:

[0055] The model is pre-trained using a large-scale license plate sample database, including the CCPD (Chinese City Parking Dataset). During training, the model extracts image features through convolutional layers, reduces their dimensionality through pooling layers, and maps them to a classification space through fully connected layers. A backpropagation algorithm is used to adjust model parameters, minimizing the loss function between the predicted results and the true labels, thereby continuously improving recognition accuracy. During recognition, the pre-processed license plate image is fed into the trained model, which then outputs the recognized license plate characters.

[0056] Performing recognition processing based on the recognized license plate characters, wherein the recognition processing at least includes: using a hash table duplication check algorithm to check for data duplication, performing legitimacy verification according to a compiled license plate number legitimacy verification program, and converting data collected from different devices into a unified JSON format;

[0057] The data duplication check uses a hash table algorithm. For each piece of collected vehicle data, the license plate number is used as a hash key, its hash value is calculated, and stored in a hash table. When processing new data, the hash value of the license plate number is first calculated to check whether data with the same hash value already exists in the hash table. If so, the license plate number and other key information are compared to determine whether it is a duplicate record. If so, the record is removed.

[0058] Among them, the legitimacy verification is carried out according to the written license plate number legitimacy verification procedure, as follows:

[0059] Based on the encoding rules for license plate numbers, for example, license plates are composed of Chinese characters, letters, and numbers, with different Chinese characters representing different provinces and certain rules for letter and number combinations, regular expressions are used to match license plate numbers and check whether they comply with the corresponding rules. Those that do not comply with the rules are treated as misidentified or removed as abnormal data.

[0060] The data collected by different devices is converted into a unified JSON format, including: pre-set data conversion scripts, naming standards and data type conversion of data fields according to the defined JSON data structure.

[0061] Specifically, the time fields are uniformly converted from different formats to the ISO8601 standard format, and the license plate number fields are uniformly converted to string types.

[0062] Generate associated tracks for the license plate characters that have been recognized;

[0063] Using license plate characters as key identifiers, we traverse all collected vehicle data records to obtain vehicle data belonging to the same vehicle at different times and locations.

[0064] This technical solution creates a database table containing information such as license plate numbers, collection time, and collection location during data storage. After reading the data from the database, a loop structure, using a Python for loop, checks the license plate numbers of each record one by one. When records with identical license plate numbers are found, these records are filtered out, indicating that they represent data collected from the same vehicle at different times and locations.

[0065] For the vehicle data with the same license plate characters that have been filtered out, sort them in ascending order according to the collection time;

[0066] This technical solution is implemented in Python using the built-in sorted() function combined with a custom sort key. Assuming the vehicle data is stored in a list, each element is a dictionary containing information such as the collection time. This can be sorted using data.sort(key = lambda x: x['collection time']) . The sorted data sequence reflects the vehicle's state changes at different points in time, providing the correct time sequence for subsequent calculations of driving direction and trajectory.

[0067] According to the sorted data, the time interval and travel distance between adjacent data points are calculated in sequence;

[0068] Among them, for the time interval, the collection time format of the calibration data point is a timestamp, such as a Unix timestamp, in seconds. The time interval (unit: second) can be obtained by subtracting the timestamp of the previous data point from the timestamp of the next data point.

[0069] Specifically, the acquisition times of two adjacent data points are t1 and t2 (the time format is unified as timestamp, in seconds), and the calculation formula of the time interval Δt is expressed as:

[0070] Δt=t2-t1;

[0071] Where t1 and t2 represent the timestamps of a specific moment. This time interval is used in subsequent speed calculations to reflect the travel time difference between adjacent vehicle locations.

[0072] The calculation of driving distance is based on the geographical location information of the collection location, such as longitude and latitude. The geographical distance calculation formula is as follows:

[0073] Given two data points with latitude and longitude (lat1, lon1) and (lat2, lon2), the radius of the Earth is R, which is approximately 6371000 meters, and the travel distance d, it can be expressed as:

[0074]

[0075] Obtaining driving directions involves the following steps:

[0076] According to the calculated time interval and travel distance, the speed calculation formula is used to calculate the speed of the vehicle between adjacent data points, which is expressed as:

[0077]

[0078] Among them, v represents speed, Δs represents travel distance, and Δt represents time interval;

[0079] In this technical solution, specifically, the speed unit of the calculation result depends on the units of distance and time. For example, if the distance is meter and the time is second, the speed unit is meter / second.

[0080] The vehicle's direction of travel is determined based on the positive or negative speed and the direction information on the map.

[0081] This technical solution marks the north at the top, the south at the bottom, the west on the left and the east on the right on the map.

[0082] Among them, if the speed is positive, combined with the map direction, it is determined that the vehicle is moving forward;

[0083] If the speed is negative, it indicates data anomalies, such as incorrect time sequence or special circumstances such as the vehicle traveling in the wrong direction. Further verification of the data accuracy is required. For example, if a vehicle is traveling from west to east on an east-west road, the speed is positive. If the speed is negative, the data needs to be reviewed.

[0084] Perform trajectory generation, abstract the map data into a directed graph data structure, perform shortest path calculation, and connect the shortest paths corresponding to each data point based on the calculated shortest path to form the vehicle's driving trajectory;

[0085] This technical solution abstracts map data into a directed graph data structure, including: road intersections in the map as nodes of the graph, road segments as directed edges connecting the nodes, and assigning a weight to each edge. The weight represents the length of the road, travel time, or other metrics related to travel costs.

[0086] Specifically, in Python, a dictionary can be used to represent the structure of a graph, where graph represents a simple directed graph, and graph['node 1']['node 2'] = 10 means that the edge weight from node 1 to node 2 is 10.

[0087] This technical solution performs the shortest path calculation, including: starting from the map node corresponding to the starting data point of the vehicle's travel, calculating the shortest path to the nodes corresponding to other data points, as follows:

[0088] Let the current node be u, its adjacent node be v, the edge weight from u to v be w(u,v), the distance from the source node to u be d(u), and the distance from the source node to v after the update be d(v), which can be expressed as:

[0089] d(v)=min(d(v),d(u)+w(u,v);

[0090] At the beginning of the algorithm, the distances from all nodes to the source node are set to infinity, and the distance from the source node to itself is set to 0. During the algorithm's execution, each time starting from the node u closest to the source node whose shortest path has not yet been determined, it traverses its adjacent nodes v and uses the above formula to update the distance from v to the source node. This iteration continues until the shortest distances of all nodes are determined, thus finding the shortest path from the source node to each node for fitting the vehicle's trajectory.

[0091] In addition, in practical applications, trajectory data can be stored as a series of node sequences, or converted into a format supported by Geographic Information System (GIS) (such as GeoJSON) for visualization on a map.

[0092] The vehicle's driving trajectory is presented using the Geographic Information System (GIS) based on the vehicle's driving trajectory.

[0093] This technical solution uses the JavaScript Leaflet library to draw vehicle trajectories on a map. It converts vehicle trajectory data into a format supported by GIS maps, such as GeoJSON. The vehicle's location points at different points in time are plotted on the map, and these points are connected by lines to show the vehicle's trajectory.

[0094] Specifically, in this application, traffic management departments use trajectory analysis results to calculate traffic flow on different road sections. By counting the trajectories of vehicles passing through a road section within a certain period, traffic flow data is generated. Changes in traffic flow over time are analyzed to identify periods of congestion. Based on traffic flow and congestion period data, traffic signal timing is optimized, such as increasing green light duration during periods of high traffic volume. Furthermore, this data provides data support for road construction and traffic diversion plans, enabling the planning of new roads or adjustments to traffic rules on existing roads.

[0095] Security departments also use trajectory analysis to track suspected vehicles. By entering the suspect's license plate number, they can query its travel routes and locations. Combined with other surveillance data, such as checkpoint monitoring and video surveillance, they analyze the suspect's movement patterns to assist in solving the case. For example, by analyzing the suspect's travel trajectory over different time periods, they can determine its likely destination or crime scene.

[0096] In summary, with the help of the above technical solution of the present invention, the following effects can be achieved:

[0097] 1. The present invention pre-installs and configures license plate recognition equipment at preset road locations to collect license plate images containing license plate information in real time; inputs the pre-processed license plate images into a preset trained license plate character recognition model to output the recognized license plate characters; performs recognition processing based on the recognized license plate characters, generates associated trajectories for the recognized license plate characters, and uses a geographic information system (GIS) to present the vehicle's driving trajectory based on the vehicle's driving trajectory, thereby realizing vehicle driving trajectory analysis based on license plate recognition. This method can not only accurately identify license plates in complex environments and improve the reliability of data collection, but also efficiently process large amounts of vehicle data, accurately calculate vehicle driving trajectories, reduce errors, and quickly and accurately generate vehicle driving trajectories, providing timely and effective support for traffic management and security applications.

[0098] 2. Through precise analysis of vehicle trajectories, the present invention can obtain real-time traffic flow data for different road sections and gain a deep understanding of how traffic flow changes over time. This provides a powerful basis for optimizing traffic signal timing, allowing traffic management departments to flexibly adjust signal light durations based on actual traffic flow. Based on a large amount of vehicle trajectory data, traffic planning departments can clearly understand road usage, including which sections have heavy traffic and which sections are prone to congestion. This data can help rationally plan new roads and optimize traffic rules on existing roads, such as setting up tidal lanes and adjusting lane divisions, making the layout of transportation facilities more scientific and reasonable, and better meeting the needs of urban transportation development.

[0099] At the same time, this method allows security departments to quickly track the travel routes and locations of suspected vehicles by simply entering their license plate numbers. Combined with other surveillance data, such as checkpoint monitoring and video surveillance, this method allows for a comprehensive analysis of the suspect vehicle's movements, accurately determining its likely destinations or crime zones, and providing key clues for solving cases. This significantly improves security department efficiency and case-solving success rates, safeguarding social security and stability.

[0100] The foregoing is merely a preferred embodiment of the present invention and is not intended to limit the present invention. A person skilled in the art will readily appreciate other embodiments of the present invention after considering the disclosure in the specification and examples. This application is intended to cover any variations, uses, or adaptations of the present invention, which follow the general principles of the present invention and include common knowledge or customary techniques in the art not disclosed herein. The description and examples are to be considered merely exemplary, and the true scope and spirit of the present invention are indicated by the claims.

[0101] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes can be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. A vehicle driving trajectory analysis method based on license plate recognition, characterized in that: The following steps are involved: Install license plate recognition equipment at preset road locations in advance to collect license plate images containing license plate information in real time; Preprocess the collected license plate images, including image enhancement, denoising, and tilt correction; Input the pre-processed license plate image into the preset trained license plate character recognition model to output the recognized license plate characters; Performing recognition processing based on the recognized license plate characters, wherein the recognition processing at least includes: using a hash table duplication check algorithm to check for data duplication, performing legitimacy verification according to a compiled license plate number legitimacy verification program, and converting data collected from different devices into a unified JSON format; The license plate characters processed through recognition are associated with trajectory generation, and the vehicle's driving trajectory is presented using the Geographic Information System (GIS) based on the vehicle's driving trajectory.

2. The vehicle driving trajectory analysis method based on license plate recognition according to claim 1 is characterized in that: The method of using a hash table duplication checking algorithm to check for duplicate data includes the following steps: For each piece of vehicle data collected, the license plate number is used as the hash key, its hash value is calculated and stored in the hash table; When processing new data, first calculate the hash value of its license plate number and check whether data with the same hash value already exists in the hash table. If so, compare the license plate number and other key information to determine whether it is a duplicate record. If so, remove it.

3. The vehicle driving trajectory analysis method based on license plate recognition according to claim 2 is characterized in that: The validity verification of the license plate number is performed according to the written license plate number validity verification program, including: presetting the encoding rules of the license plate number, and matching the license plate number through regular expressions to check whether it complies with the corresponding rules. Those that do not comply with the rules will be removed as misidentified or abnormal data.

4. The vehicle driving trajectory analysis method based on license plate recognition according to claim 3 is characterized in that: Convert data collected by different devices into a unified JSON format, including: pre-setting data conversion scripts, and converting data field naming specifications and data types according to the preset JSON data structure.

5. The vehicle driving trajectory analysis method based on license plate recognition according to claim 1 is characterized in that: Generate an associated trajectory for the license plate characters that have been recognized, including the following steps: Using license plate characters as key identifiers, we traverse all collected vehicle data records to obtain vehicle data belonging to the same vehicle at different times and locations. For the vehicle data with the same license plate characters that have been filtered out, sort them in ascending order according to the collection time; According to the sorted data, the time interval and driving distance between adjacent data points are calculated in sequence, and the driving direction is obtained; The map data is abstracted into a directed graph data structure, and the shortest path calculation is performed. Based on the calculated shortest path, the shortest paths corresponding to each data point are connected to form the vehicle's driving trajectory.

6. The vehicle driving trajectory analysis method based on license plate recognition according to claim 5 is characterized in that: Calculating the time interval between adjacent data points includes the following steps: The acquisition time of two adjacent calibration data points is t1 and t2 respectively, and the time interval Δt is expressed as: Δt=t2-t1; Among them, t1 and t2 represent the timestamps of a specific moment respectively.

7. The vehicle driving trajectory analysis method based on license plate recognition according to claim 5 is characterized in that: Calculating the driving distance between adjacent data points involves the following steps: The longitude and latitude of the two data points are calibrated as (lat1, lon1) and (lat2, lon2), the radius of the earth is R, and the travel distance d is expressed as:

8. The vehicle driving trajectory analysis method based on license plate recognition according to claim 1 is characterized in that: Obtaining driving directions involves the following steps: According to the calculated time interval and travel distance, the speed calculation formula is used to calculate the speed of the vehicle between adjacent data points, which is expressed as: Among them, v represents speed, Δs represents travel distance, and Δt represents time interval; The vehicle's direction of travel is determined based on the positive or negative speed and the direction information on the map. On the map, north is at the top, south is at the bottom, west is on the left, and east is on the right. If the speed is positive, combined with the map direction, it is determined that the vehicle is moving forward; if the speed is negative, it indicates data abnormality.

9. The vehicle driving trajectory analysis method based on license plate recognition according to claim 8 is characterized in that: The shortest path calculation includes the following steps: Let the current node be u, its adjacent node be v, the edge weight from u to v be w(u,v), the distance from the source node to u be d(u), and the distance from the source node to v after the update be d(v), which can be expressed as: d(v)=min(d(v),d(u)+w(u,v).

Images