Vehicle violation behavior detection method and device for traffic checkpoint

Abstract

The invention discloses a vehicle violation behavior detection method and device for a traffic gate, and relates to the technical field of traffic management, and the method comprises the steps: recognizing the identity of a vehicle based on a real-time vehicle passing image obtained by the traffic gate, and extracting a corresponding behavior constraint rule and a behavior index set from a behavior rule database; extracting real-time behavior characteristics in combination with the image data, identifying difference indexes and collecting additional information; and extracting additional behavior characteristics, combining the additional behavior characteristics with the real-time behavior characteristics, and carrying out illegal behavior detection based on the behavior index set. Therefore, the technical effects of improving the monitoring and detection accuracy and enhancing the adaptive capacity and the detection rate are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of traffic management, and in particular to a method and device for detecting vehicle violations at a traffic checkpoint. Background Art

[0002] Traditional traffic management models rely primarily on human patrols and simple electronic monitoring equipment, which struggles to effectively monitor complex traffic behaviors. To improve the efficiency and accuracy of traffic management, many checkpoints are equipped with video capture devices that capture real-time image data of passing vehicles. This provides the foundation for vehicle identification and behavior monitoring, and theoretically allows for the definition and classification of legal and illegal vehicle behavior.

[0003] However, existing technologies have many limitations in vehicle behavior monitoring. First, traditional methods often rely on a single data source for monitoring, such as vehicle identity recognition and simple behavior judgment based solely on passing vehicle image data, or violation comparison based solely on a behavioral rule database. This single-source data monitoring method cannot fully and accurately reflect the actual behavioral characteristics of vehicles, and is prone to misjudgment or omission. Secondly, the behavioral indicator sets in existing technologies are mostly static and cannot be flexibly adjusted according to different vehicle types and road scenarios, making it difficult to adapt to complex and changing traffic conditions. In addition, the monitoring of vehicle behavior lacks the ability to identify real-time differences and cannot capture potential violation risks at the moment when vehicle behavior deviates, affecting the timeliness and effectiveness of traffic management. Finally, existing violation detection algorithms are mostly based on single features or simple feature combinations, which make it difficult to identify hidden violations, and the reasoning efficiency and accuracy in high traffic volume scenarios need to be improved. Summary of the Invention

[0004] The present invention provides a method and device for detecting vehicle violations at traffic checkpoints to solve the technical problems in the existing technology, such as strong limitations of single-source data monitoring, which easily leads to misjudgment or missed judgment, static behavior indicator sets and lack of real-time difference recognition capabilities, difficulty in adapting to complex traffic conditions, and difficulty in identifying hidden violations by violation detection algorithms, thereby achieving the technical effects of improving monitoring and detection accuracy, enhancing adaptability and detection rate.

[0005] In a first aspect, the present invention provides a method for detecting vehicle violations at a traffic checkpoint, wherein the method comprises:

[0006] Vehicle identity recognition is performed based on real-time vehicle passing image data, wherein the real-time vehicle passing image is obtained by a video acquisition device at a traffic checkpoint.

[0007] According to the vehicle identity, the corresponding behavior constraint rules are extracted from the behavior rule database, and the corresponding behavior indicator set is defined.

[0008] Real-time behavior features are extracted based on the real-time vehicle passing image data, and a difference indicator set is identified in combination with the behavior indicator set, and additional information is collected based on the difference indicator set.

[0009] Additional behavior features of the vehicle are extracted according to the additional information collection result, and violation detection based on the behavior indicator set is performed by combining the additional behavior features with the real-time behavior features.

[0010] In a feasible implementation, vehicle identification is performed based on real-time vehicle image data, wherein the real-time vehicle image is acquired by a video acquisition device at a traffic checkpoint, including:

[0011] The real-time passing vehicle image data is obtained through a video acquisition device arranged at a traffic checkpoint.

[0012] An image recognition algorithm is used to extract the image feature identification information of the vehicle from the real-time passing vehicle image data, wherein the image feature identification information includes but is not limited to the license plate number, vehicle model outline, color characteristics and specific markers.

[0013] The image feature identification information is submitted to the background vehicle information database for identity matching to obtain the vehicle identity, wherein the vehicle identity includes vehicle category identity, unique identification identity and special task identity.

[0014] In a feasible implementation, the corresponding behavior constraint rules are extracted from the behavior rule database based on the vehicle identity, and a corresponding behavior indicator set is defined, including:

[0015] According to the vehicle identity, a corresponding behavior constraint information set is retrieved from a behavior rule database.

[0016] The behavior constraint information set is parsed based on a preset semantic recognition model to obtain discrete constraint fields.

[0017] Field rules are defined, and the discrete constraint fields are structured and processed based on the field rules, and the structured processing results are output as the behavior indicator set.

[0018] In a feasible implementation, the behavior indicator set includes at least one of a travel speed, a travel route, a time limit, a lane requirement, a stay time threshold, and a travel authority.

[0019] In a feasible implementation, extracting real-time behavior features based on the real-time vehicle passing image data, identifying a difference indicator set in combination with the behavior indicator set, and collecting additional information based on the difference indicator set includes:

[0020] Behavior detection is performed on the real-time vehicle passing image data to obtain the real-time behavior characteristics.

[0021] The real-time behavior characteristics are analyzed to extract a real-time behavior indicator set, and a difference item analysis is performed in combination with the behavior indicator set, and the complement of the real-time behavior indicator set is taken as the difference item analysis result.

[0022] Based on the results of the difference analysis, locate and interact with multiple information sources to collect additional information.

[0023] In a feasible implementation, additional behavior features of the vehicle are extracted based on the additional information collection results, and the additional behavior features are combined with the real-time behavior features to perform violation detection based on the behavior indicator set, including:

[0024] The additional information is comprehensively processed to extract additional baseline behavioral features.

[0025] According to the difference item analysis result, a multi-dimensional correlation relationship between the additional benchmark behavior feature and the real-time behavior feature is established, wherein the multi-dimensional correlation relationship includes a hierarchical correlation relationship and a temporal correlation relationship.

[0026] Based on the multi-dimensional association relationship, the additional behavior feature is calculated accordingly, and the additional behavior feature and the real-time behavior feature are fused and reconstructed according to the field rule to obtain a fused behavior feature.

[0027] The fused behavior features are compared with the behavior indicator set to detect illegal behaviors.

[0028] In a feasible implementation, comparing the fused behavior feature with the behavior indicator set to detect illegal behavior further includes:

[0029] Based on the field rule, the fused behavior feature and the behavior indicator set are traversed to perform feature value verification.

[0030] If any indicator fails to be verified, it is determined that a violation has occurred, and the violation detection result is output accordingly in combination with the behavior constraint rules, wherein the violation detection result at least includes vehicle identity, violation type, violation time, and violation location.

[0031] In a second aspect, the present invention further provides a vehicle violation detection device for a traffic checkpoint, wherein the vehicle violation detection device for a traffic checkpoint comprises:

[0032] The vehicle identification component is used to perform vehicle identification based on real-time vehicle passing image data, wherein the real-time vehicle passing image is obtained by a video acquisition device at a traffic checkpoint.

[0033] The behavior indicator set definition component is used to extract corresponding behavior constraint rules from the behavior rule database based on vehicle identity and define the corresponding behavior indicator set.

[0034] The difference recognition component is used to extract real-time behavior features based on the real-time vehicle passing image data, identify a difference indicator set in combination with the behavior indicator set, and collect additional information based on the difference indicator set.

[0035] The violation detection component is used to extract additional behavior features of the vehicle based on the additional information collection results, and combine the additional behavior features with the real-time behavior features to perform violation detection based on the behavior indicator set.

[0036] The present invention discloses a method and device for detecting vehicle violations at traffic checkpoints, comprising: performing vehicle identity recognition based on real-time vehicle passing image data, the real-time vehicle passing image being acquired by a video acquisition device at the traffic checkpoint; extracting corresponding behavior constraint rules from a behavior rule database according to the recognized vehicle identity, and defining a behavior indicator set accordingly; extracting real-time behavior features from the real-time vehicle passing image data, identifying a difference indicator set in combination with the behavior indicator set, and performing additional information collection based on the difference indicator set; extracting additional vehicle behavior features based on the additional information collection result, combining the additional behavior features with the real-time behavior features, and performing violation detection based on the behavior indicator set. The method and device for detecting vehicle violations at traffic checkpoints disclosed by the present invention solve the technical problems that single-source data monitoring has strong limitations, which easily leads to misjudgment or missed judgment, the behavior indicator set is static and lacks real-time difference recognition capability, making it difficult to adapt to complex traffic conditions, and the violation detection algorithm has difficulty in identifying hidden violations, thereby achieving the technical effects of improving monitoring and detection accuracy, and enhancing adaptability and detection rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 The figure is a flow chart of a method for detecting vehicle violations at a traffic checkpoint according to the present invention.

[0038] Figure 2 The figure is a schematic structural diagram of a vehicle violation detection device at a traffic checkpoint according to the present invention.

[0039] Description of the accompanying drawings: vehicle identification component 11, behavior indicator set definition component 12, difference identification component 13, violation detection component 14. DETAILED DESCRIPTION

[0040] The above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods of the specification to better understand the above technical solution. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments of the present invention. It should be understood that the present invention is not limited to the example embodiments used only to explain the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. In addition, it should be noted that, for the convenience of description, only the parts related to the present invention, rather than all, are shown in the drawings.

[0041] Example 1, as Figure 1 The flowchart of a method for detecting vehicle violations at a traffic checkpoint according to the present invention is shown below. The method comprises:

[0042] S100: Vehicle identity recognition is performed based on real-time vehicle passing image data, wherein the real-time vehicle passing image is acquired by a video acquisition device at a traffic checkpoint.

[0043] Specifically, vehicle identification based on real-time vehicle-passing image data refers to the use of vehicle-passing images acquired by video acquisition devices to determine the specific identity information of the vehicle through image processing and recognition technology. Real-time vehicle-passing images refer to image data captured instantly by video acquisition devices such as cameras when a vehicle passes through a traffic checkpoint. Vehicle identification involves extracting vehicle features from images, such as license plate numbers, vehicle model outlines, and colors, and comparing them with a vehicle information database to obtain detailed vehicle identity information, including vehicle category (such as small cars, large trucks, etc.), unique identification (such as license plate numbers), and special mission identities (such as police cars, ambulances, etc.).

[0044] Through the above process, real-time and accurate identification of vehicle identity is achieved, providing an accurate vehicle information basis for subsequent violation detection.

[0045] In some embodiments, vehicle identification is performed based on real-time vehicle image data, wherein the real-time vehicle image is acquired by a video acquisition device at a traffic checkpoint, including:

[0046] The real-time passing vehicle image data is obtained through a video acquisition device installed at a traffic checkpoint; an image recognition algorithm is used to extract the vehicle's image feature identification information from the real-time passing vehicle image data, wherein the image feature identification information includes but is not limited to the license plate number, vehicle model outline, color characteristics and specific markers; the image feature identification information is submitted to a background vehicle information database for identity matching to obtain the vehicle identity, wherein the vehicle identity includes vehicle category identity, unique identification identity and special task identity.

[0047] Specifically, this step of "vehicle identity recognition based on real-time vehicle passing image data" includes three key actions. First, real-time vehicle passing image data is obtained through a video acquisition device installed at the traffic checkpoint, that is, images of vehicles passing by are captured with the help of cameras and other equipment installed at the traffic checkpoint. Second, image recognition algorithms are used to extract the image feature identification information of the vehicle, which means using image processing technology to extract features that can represent the identity of the vehicle, such as license plate numbers, vehicle model outlines, etc., from the acquired images. Third, the image feature identification information is submitted to the background vehicle information database for identity matching to obtain the vehicle identity, that is, the extracted feature information is compared with the vehicle registration information in the background database to finally determine the identity of the vehicle, including vehicle category, unique identification code, etc.

[0048] Specifically, video acquisition devices are usually installed at appropriate locations at traffic checkpoints, and they can capture images in real time as vehicles pass by. For example, a camera installed at a checkpoint on a main road in a certain city can capture 30 frames of images per second with a resolution of 2048×1536 pixels, which can clearly record the appearance details of the vehicle. Next, the image recognition algorithm extracts feature identification information. Taking license plate recognition as an example, the image is first preprocessed, such as grayscale and binarization, and then the convolutional neural network (CNN) algorithm based on deep learning is used to identify the license plate number. Similarly, features such as the outline and color of the vehicle model can also be extracted. Finally, the extracted features are submitted to the background database for identity matching. The background vehicle information database stores a large amount of vehicle registration information, and can quickly find matching items in the database based on the extracted features.

[0049] Through this process, accurate and efficient identification of vehicle identities is achieved, laying a solid foundation for subsequent violation detection. On the one hand, real-time acquisition and identification of image data make traffic monitoring immediate and dynamic, and can quickly respond to changes in traffic conditions. On the other hand, the extraction and matching of multiple image features greatly improves accuracy and reliability compared to relying solely on single feature recognition. For example, when the license plate is partially obscured, the use of features such as the vehicle model outline and color to assist in judgment can reduce the misidentification rate by 80%. Furthermore, efficient matching with the background database ensures the authority and comprehensiveness of identity information, providing solid support for the subsequent extraction of behavioral constraint rules for different vehicle types and attributes.

[0050] S200: extracting corresponding behavior constraint rules from a behavior rule database based on the vehicle identity, and defining a corresponding behavior indicator set.

[0051] In some embodiments, corresponding behavior constraint rules are extracted from a behavior rule database based on the vehicle identity, and a corresponding behavior indicator set is defined, including:

[0052] Based on the vehicle identity, a corresponding behavior constraint information set is retrieved from a behavior rule database; the behavior constraint information set is parsed based on a preset semantic recognition model to obtain discrete constraint fields; field rules are defined, and the discrete constraint fields are structured and processed based on the field rules, and the structured processing results are output as the behavior indicator set.

[0053] Specifically, a vehicle identity is the unique identification information used to identify a vehicle, such as a license plate number or vehicle identification number. The behavior rule database stores the behavior constraint rules associated with different vehicle types or identities, and the behavior indicator set is the specific indicators extracted from these rules to guide vehicle behavior.

[0054] Specifically, the system first retrieves the corresponding behavioral constraint information from the behavioral rule database based on the vehicle's identity. For example, if a vehicle is identified as a "city bus," the system extracts the specific behavioral constraint rules for that type of vehicle from the database, such as "must drive in bus lanes," "stop at designated stops," and "speed limit of 50 km / h."

[0055] Next, the semantic recognition model in natural language processing (NLP) technology is used to parse this information and decompose it into discrete constraint fields, such as "bus lane", "designated stop", "speed limit 50", etc.

[0056] Then, field rules are defined, and the above discrete constraint fields are structured according to the field rules, and converted into quantifiable indicators, such as "lane requirements: bus lane", "stop position: within ±10 meters of the specified station", "traffic speed: no more than 50 kilometers per hour", etc., to finally form a set of behavioral indicators.

[0057] Through this process, we can effectively extract specific, actionable behavioral indicators from complex behavioral rules, providing precise guidance for vehicle behavior management. This structured processing not only improves the readability and operability of information, but also ensures that vehicles adhere to regulations while optimizing their traffic efficiency and safety.

[0058] In some embodiments, the behavior indicator set includes at least one of a traffic speed, a traffic route, a time limit, a lane requirement, a stop time threshold, and a traffic permission.

[0059] Specifically, the final output behavioral indicator set includes at least one or more of the following: travel speed (such as 60km / h), travel route (such as ABC), time limit (such as weekdays only), lane requirement (such as bus lanes only), stay time threshold (such as no more than 30 minutes) and access rights (such as only vehicles with specific licenses).

[0060] S300: extracting real-time behavior features based on the real-time vehicle passing image data, identifying a difference indicator set in combination with the behavior indicator set, and collecting additional information based on the difference indicator set.

[0061] Specifically, this step includes three key actions. First, "extracting real-time behavior features based on the real-time vehicle passing image data" refers to analyzing and extracting features that can reflect the current behavior of the vehicle from the real-time vehicle passing image data, such as the vehicle's driving speed, driving trajectory, and stay time. Second, "identifying a difference indicator set in combination with the behavior indicator set" refers to comparing the extracted real-time behavior features with the pre-defined behavior indicator set to find out the missing parts of the real-time behavior features compared to the indicator set requirements, and forming a difference indicator set. Third, "collecting additional information based on the difference indicator set" refers to further collecting additional information related to the abnormal or potential violations indicated by the difference indicator set, so as to provide more comprehensive data support for subsequent violation detection.

[0062] In some embodiments, extracting real-time behavior features based on the real-time vehicle passing image data, identifying a difference indicator set in combination with the behavior indicator set, and collecting additional information based on the difference indicator set includes:

[0063] Behavior detection is performed on the real-time vehicle passing image data to obtain the real-time behavior characteristics; the real-time behavior characteristics are analyzed to extract a real-time behavior indicator set, and a difference item analysis is performed in combination with the behavior indicator set, and the complement of the real-time behavior indicator set is taken as the difference item analysis result; based on the difference item analysis result, multiple information sources are located and interacted to collect additional information.

[0064] Specifically, real-time vehicle image data is real-time image information of vehicles acquired through devices such as cameras. Real-time behavior features are specific features of the vehicle's current behavior extracted from these image data.

[0065] Specifically, the real-time behavior indicator set refers to a set of indicators with structured semantics extracted from real-time behavior features, such as real-time traffic speed, current route, occupied lane type, length of stay, etc. The preset behavior indicator set is a set of target behavior specifications defined according to vehicle identity, traffic rules or scenario constraints. The difference indicator set between the two is the list of indicator categories that need to be supplemented for collection, that is, the part that the real-time behavior indicator set lacks compared to the preset behavior indicator set, that is, the behavioral indicator items that are not covered by the current image recognition but are required to be identified in the rules. These missing items are usually key indicators that need to be improved through additional information collection or supplementary identification methods.

[0066] Specifically, first, real-time passing vehicle image data is captured by a camera, and computer vision technology (such as YOLO, R-CNN, etc.) is used to detect behavior and extract the vehicle's real-time behavior features, such as current speed, driving route, lane, etc. These features are then parsed to extract a real-time behavior indicator set, that is, a set of indicator categories encompassed by the real-time behavior features. This real-time behavior indicator set is then compared with a preset behavior indicator set to identify difference items and form a difference indicator set. Based on the difference item analysis results, it is possible to locate and interact with multiple information sources, such as traffic management centers, weather information systems, etc., to collect additional information. This may include obtaining information on current road conditions, weather conditions, or other influencing factors to provide a more comprehensive understanding of vehicle behavior.

[0067] For example, for a specific vehicle, an image can identify its current speed as 65 km / h, its route as "XYZ," and its occupied lane as "Lane 2." However, due to occupancy or image quality issues, behavioral characteristics such as the number of passengers and whether the hazard lights are on cannot be identified. Simultaneously, the preset behavioral indicator set for this vehicle is retrieved, revealing that the required behavioral requirements for this type of vehicle on the current road section also include "Hazard lights must be on" and "No more than six passengers." Set operations (such as difference operations) can be used to identify the missing components of the real-time indicator set, namely, "Hazard lights on" and "Number of passengers," thereby forming a differential indicator set. These differential items prompt the system to conduct additional collection or inference using other information sources (such as the vehicle terminal, historical behavioral data, external sensors, and external data sources).

[0068] Through the above process, not only can the real-time behavior status of the current vehicle be identified, but the "information blind spots" in the current recognition results can also be intelligently discovered, and based on this, the supplementary information collection process can be actively triggered. This mechanism ensures the integrity and accuracy of behavior recognition, avoiding the lack of behavior judgment due to factors such as image occlusion and recognition accuracy limitations. For example, in traffic accident analysis or abnormal behavior warning, although certain key indicators (such as whether the hazard lights are turned on) cannot be directly obtained from the image, through the identification of the difference indicator set, the system can promptly call on-board data or historical behavior models to supplement it, providing more comprehensive background information to support decision-making, thereby improving the system's perception and response capabilities to complex behavior scenarios.

[0069] S400: extracting additional behavior features of the vehicle according to the additional information collection result, and combining the additional behavior features with the real-time behavior features to perform violation detection based on the behavior indicator set.

[0070] Specifically, additional information collection results refer to supplementary data obtained through interaction with external information sources after identifying the difference indicator set. Exemplary external information sources include, but are not limited to: on-board terminals (such as OBD, CAN bus), traffic management platforms (such as high-precision maps, traffic incident notifications), environmental perception systems (such as weather radar, roadside sensing units (RSUs), etc.

[0071] Specifically, the additional behavioral features extracted from these information sources are behavioral data that cannot be directly identified through images but are crucial for behavioral judgment. Examples include: lighting status (turn signals, hazard lights), number of occupants, brake status, wiper operation, navigation path deviation, etc. These additional features, together with real-time behavioral features (such as speed, lane, route, etc.), constitute a complete set of behavioral indicators.

[0072] For example, the additional behavior characteristics and the real-time behavior characteristics are shown in Table 1 below:

[0073] Table 1. Example of correspondence between additional behavior features and real-time behavior features

[0074] Indicator Final Value Data Source Real-time traffic speed 65km / h Image Recognition Current route X→Y→Z Image Recognition Occupied lane type Second lane Image Recognition Are the hazard warning lights on? yes CAN bus Number of passengers in the vehicle 8 Vehicle-mounted sensors

[0075] In some embodiments, extracting additional behavior features of the vehicle based on the additional information collection results, and combining the additional behavior features with the real-time behavior features to perform violation detection based on the behavior indicator set includes:

[0076] Comprehensively process the additional information to extract additional baseline behavior features; establish a multidimensional correlation relationship between the additional baseline behavior features and the real-time behavior features based on the difference item analysis results, wherein the multidimensional correlation relationship includes a hierarchical correlation relationship and a temporal correlation relationship; based on the multidimensional correlation relationship, calculate the additional behavior features accordingly, and fuse and reconstruct the additional behavior features and the real-time behavior features according to the field rules to obtain a fused behavior feature; compare the fused behavior feature with the behavior indicator set to detect violations.

[0077] Specifically, additional baseline behavioral features are structured features with behavioral semantics extracted by standardizing raw additional information (such as raw data from the CAN bus, V2X communications, onboard sensors, and navigation data). Examples include "number of occupants in vehicle is 8" and "navigation path deviation." These features often exist as low-level or primitive fields in the raw data (such as electrical signal states, Boolean variables, and sensor voltages), requiring parsing and mapping to obtain semantically meaningful baseline behavioral features.

[0078] Specifically, multidimensional associations are used to establish logical connections between additional baseline behavioral features and real-time behavioral features. These associations are divided into two dimensions: hierarchical associations (e.g., "turning on the turn signal" is a sub-behavior of "changing lanes") and temporal associations (e.g., "braking" occurs before "changing lanes"). Fusion behavioral features are an integrated, semantically complete, and data-consistent set of behavioral features that serve as the input for violation detection.

[0079] Specifically, the additional information is first processed comprehensively to extract additional baseline behavioral features. For example, the additional information may include the vehicle's evasive maneuvers after speeding or the use of turn signals when changing lanes. Through data cleaning and feature extraction, additional baseline behavioral features such as the number of emergency avoidance maneuvers and the duration of turn signal use are obtained.

[0080] Next, multi-dimensional associations are established based on the difference analysis results. For example, if the difference indicator set shows that the vehicle was speeding and not driving in the designated lane, the system will establish a hierarchical association between "speeding" and "lane departure" (indicating that the two violations may be caused by the same intention), as well as a temporal association between "speeding occurrence time" and "lane departure occurrence time" (analyzing the order in which the behaviors occurred).

[0081] Then, based on multidimensional associations, additional behavioral features are calculated, such as the "joint risk value of speeding and lane departure." These features are then fused and reconstructed with real-time behavioral features (such as "current speed" and "lane position") according to field rules to obtain the fused behavioral feature "comprehensive violation risk value." This fused behavioral feature is then compared with the behavioral indicator set for violation detection. For example, if the behavioral indicator set stipulates that a "comprehensive violation risk value" exceeding 100 indicates a violation, and the calculated value is 120, the vehicle is deemed to have violated the law, and the system immediately records and reports the violation.

[0082] Through this process, we achieve deep fusion of heterogeneous data from multiple sources, ensuring consistency in temporal, semantic, and formatting of behavioral features. Fusion of behavioral features is not only more complete and accurate, but also traceable (each field can be traced back to the original data), providing a reliable basis for violation detection.

[0083] In some implementations, comparing the fused behavior feature with the behavior indicator set to detect illegal behavior further includes:

[0084] Based on the field rules, the fused behavior features and the behavior indicator set are traversed to perform feature value verification; if any indicator verification fails, it is determined that a violation exists, and the violation detection result is output in combination with the behavior constraint rules, wherein the violation detection result at least includes vehicle identity, violation type, violation time, and violation location.

[0085] Specifically, first, based on the field rules, the fused behavior features and the behavior indicator set are traversed to perform feature value verification. For example, the field rules stipulate that the allowable error range of "traffic speed" is ±5 kilometers per hour, and the "stay time" must not exceed 1.2 times the specified value. Then, according to these rules, the "traffic speed" and "stay time" and other indicator values ​​in the fused behavior features are compared one by one with the corresponding standard values ​​in the behavior indicator set. Suppose the fused behavior features of a vehicle at a certain intersection show that its passing speed is 62 kilometers per hour, and the behavior indicator set stipulates that the speed limit of this section is 60 kilometers per hour, with an allowable error range of ±5 kilometers per hour, then the speed indicator verification result of the vehicle is qualified; but if the behavior indicator set stipulates a speed limit of 50 kilometers per hour, then the verification result is unqualified.

[0086] Specifically, if any indicator fails to verify during the traversal process, a violation is determined and, combined with the behavioral constraint rules, a violation detection result is output. For example, the system records the offending vehicle's identity information (such as license plate number), the violation type (such as speeding), the specific time of the violation (such as 2024-07-2014:30:25), and the location of the violation (such as a checkpoint at a certain intersection). This information is then integrated into the violation detection result and stored in the database. At the same time, it can trigger an audible and visual alarm or automatically push the information to the traffic management department's response platform.

[0087] In summary, the method for detecting vehicle violations at traffic checkpoints provided by the present invention has the following technical effects:

[0088] The method performs vehicle identity recognition based on real-time vehicle passing image data, and the real-time vehicle passing image is obtained by the video acquisition device of the traffic checkpoint; according to the recognized vehicle identity, the corresponding behavior constraint rules are extracted from the behavior rule database, and a behavior indicator set is defined accordingly; real-time behavior features are extracted from the real-time vehicle passing image data, and a difference indicator set is identified in combination with the behavior indicator set, and additional information is collected based on the difference indicator set; additional vehicle behavior features are extracted based on the additional information collection results, and the additional behavior features are combined with the real-time behavior features, and violation detection is implemented based on the behavior indicator set, thereby achieving the technical effect of improving monitoring and detection accuracy, enhancing adaptability and detection rate.

[0089] Example 2, as Figure 2 This is a schematic diagram of the structure of a vehicle violation detection device for a traffic checkpoint of the present invention. For example, Figure 1 The flow chart of the vehicle violation detection method at a traffic checkpoint of the present invention can be shown as follows: Figure 2 The structure shown is implemented.

[0090] Based on the same concept as the vehicle violation detection method at a traffic checkpoint in the above embodiment, the present invention further provides a vehicle violation detection device at a traffic checkpoint, comprising:

[0091] The vehicle identification component 11 is used to perform vehicle identification based on real-time vehicle passing image data, wherein the real-time vehicle passing image is obtained by a video acquisition device at a traffic checkpoint.

[0092] The behavior indicator set definition component 12 is used to extract corresponding behavior constraint rules from the behavior rule database according to the vehicle identity and define the behavior indicator set accordingly.

[0093] The difference recognition component 13 is used to extract real-time behavior features based on the real-time vehicle passing image data, identify a difference index set in combination with the behavior index set, and collect additional information based on the difference index set.

[0094] The violation detection component 14 is used to extract additional behavior features of the vehicle according to the additional information collection result, and perform violation detection based on the behavior indicator set by combining the additional behavior features with the real-time behavior features.

[0095] In some embodiments, the vehicle identification component 11 includes:

[0096] The real-time vehicle passing image data acquisition unit is used to acquire the real-time vehicle passing image data through a video acquisition device arranged at a traffic checkpoint.

[0097] The image feature identification information extraction unit is used to extract the image feature identification information of the vehicle from the real-time passing vehicle image data using an image recognition algorithm, wherein the image feature identification information includes but is not limited to the license plate number, vehicle model outline, color characteristics and specific markers.

[0098] The vehicle identity matching and acquisition unit is used to submit the image feature identification information to the background vehicle information database for identity matching and obtain the vehicle identity, wherein the vehicle identity includes vehicle category identity, unique identification identity and special task identity.

[0099] In some embodiments, the behavioral indicator set definition component 12 includes:

[0100] The behavior constraint information set retrieving unit is used to retrieve the corresponding behavior constraint information set from the behavior rule database according to the vehicle identity.

[0101] The discrete constraint field parsing unit is used to parse the behavior constraint information set based on a preset semantic recognition model to obtain discrete constraint fields.

[0102] The behavior indicator set generating unit is used to define field rules, and perform structured processing on the discrete constraint fields based on the field rules, and output the structured processing result as the behavior indicator set.

[0103] In some embodiments, the difference identification component 13 includes:

[0104] The real-time behavior feature acquisition unit is used to perform behavior detection on the real-time vehicle passing image data to acquire the real-time behavior feature.

[0105] The real-time behavior indicator set parsing and difference item analysis unit is used to parse the real-time behavior features to extract the real-time behavior indicator set, and perform difference item analysis on the behavior indicator set, taking the complement of the real-time behavior indicator set as the difference item analysis result.

[0106] The additional information collection unit is used to locate and interact with multiple information sources to collect additional information based on the difference item analysis results.

[0107] In some embodiments, the violation detection component 14 includes:

[0108] The additional benchmark behavior feature extraction unit is used to comprehensively process the additional information and extract the additional benchmark behavior feature.

[0109] The multi-dimensional correlation relationship establishing unit is used to establish a multi-dimensional correlation relationship between the additional baseline behavior feature and the real-time behavior feature according to the difference item analysis result, wherein the multi-dimensional correlation relationship includes a hierarchical correlation relationship and a temporal correlation relationship.

[0110] The fused behavior feature acquisition unit is used to calculate the additional behavior feature based on the multi-dimensional association relationship, and fuse and reconstruct the additional behavior feature with the real-time behavior feature according to the field rule to obtain the fused behavior feature.

[0111] The illegal behavior detection unit is used to compare the fused behavior feature with the behavior indicator set to perform illegal behavior detection.

[0112] In some embodiments, the violation detection component 14 further includes:

[0113] The feature value verification unit is used to traverse the fused behavior feature and the behavior indicator set to perform feature value verification based on field rules.

[0114] The violation determination and result output unit is used to determine that a violation exists if any indicator verification fails, and output the violation detection result in combination with the behavior constraint rules, wherein the violation detection result at least includes vehicle identity, violation type, violation time, and violation location.

[0115] In some implementations, the behavior indicator set includes at least one of a travel speed, a travel route, a time limit, a lane requirement, a stop time threshold, and a travel authority.

[0116] It should be understood that the embodiments mentioned in this specification focus on their differences from other embodiments. The specific embodiments in the aforementioned embodiment one are also applicable to the vehicle violation detection equipment at a traffic checkpoint described in embodiment two. For the sake of brevity of the specification, they will not be further elaborated here.

[0117] It should be understood that the embodiments disclosed in the present invention and the above description can enable those skilled in the art to use the present invention to implement the present invention. At the same time, the present invention is not limited to the embodiments mentioned above. It should be understood that those skilled in the art can still modify the technical solutions described in the above embodiments or replace some of the technical features therein with equivalents; and such modifications or replacements do not deviate from the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention and are all included in the scope of protection of the present invention.

Claims

1. A method for detecting vehicle violations at a traffic checkpoint, characterized in that: include: Vehicle identification is performed based on real-time vehicle passing image data, wherein the real-time vehicle passing image is obtained by a video acquisition device at a traffic checkpoint; Extract the corresponding behavior constraint rules from the behavior rule database based on the vehicle identity, and define the corresponding behavior indicator set; Extracting real-time behavior features based on the real-time vehicle passing image data, identifying a difference indicator set in combination with the behavior indicator set, and collecting additional information based on the difference indicator set; Additional behavior features of the vehicle are extracted according to the additional information collection result, and violation detection based on the behavior indicator set is performed by combining the additional behavior features with the real-time behavior features.

2. A method for detecting vehicle violations at a traffic checkpoint as claimed in claim 1, characterized in that: Vehicle identification is performed based on real-time vehicle passing image data, wherein the real-time vehicle passing image is obtained by a video acquisition device at a traffic checkpoint, including: Acquire the real-time vehicle passing image data through a video acquisition device provided at a traffic checkpoint; Using an image recognition algorithm to extract image feature identification information of the vehicle from the real-time vehicle passing image data, wherein the image feature identification information includes but is not limited to the license plate number, vehicle model outline, color characteristics and specific markers; The image feature identification information is submitted to the background vehicle information database for identity matching to obtain the vehicle identity, wherein the vehicle identity includes vehicle category identity, unique identification identity and special task identity.

3. A method for detecting vehicle violations at a traffic checkpoint as claimed in claim 2, characterized in that: Based on the vehicle identity, the corresponding behavior constraint rules are extracted from the behavior rule database, and the corresponding behavior indicator set is defined, including: Retrieving a corresponding behavior constraint information set from a behavior rule database based on the vehicle identity; Parsing the behavior constraint information set based on a preset semantic recognition model to obtain discrete constraint fields; Field rules are defined, and the discrete constraint fields are structured and processed based on the field rules, and the structured processing results are output as the behavior indicator set.

4. A method for detecting vehicle violations at a traffic checkpoint as claimed in claim 3, characterized in that: The behavior indicator set includes at least one of a travel speed, a travel route, a time limit, a lane requirement, a stay time threshold, and a travel authority.

5. The method for detecting vehicle violations at a traffic checkpoint according to claim 3, wherein: Extracting real-time behavior features based on the real-time vehicle passing image data, identifying a difference indicator set in combination with the behavior indicator set, and collecting additional information based on the difference indicator set, including: Performing behavior detection on the real-time vehicle passing image data to obtain the real-time behavior features; parsing the real-time behavior features to extract a real-time behavior indicator set, performing difference item analysis on the behavior indicator set, and taking the complement of the real-time behavior indicator set as the difference item analysis result; Based on the results of the difference analysis, locate and interact with multiple information sources to collect additional information.

6. A method for detecting vehicle violations at a traffic checkpoint as claimed in claim 5, characterized in that: Extracting additional behavior features of the vehicle based on the additional information collection results, and combining the additional behavior features with the real-time behavior features to perform violation detection based on the behavior indicator set, including: Comprehensively process the additional information to extract additional baseline behavioral features; Establishing a multi-dimensional correlation relationship between the additional baseline behavior feature and the real-time behavior feature based on the difference item analysis result, wherein the multi-dimensional correlation relationship includes a hierarchical correlation relationship and a temporal correlation relationship; Based on the multi-dimensional association relationship, the additional behavior feature is calculated accordingly, and the additional behavior feature is fused and reconstructed with the real-time behavior feature according to the field rule to obtain a fused behavior feature; The fused behavior features are compared with the behavior indicator set to detect illegal behaviors.

7. A method for detecting vehicle violations at a traffic checkpoint as claimed in claim 6, characterized in that: Comparing the fused behavior features with the behavior indicator set to detect illegal behaviors also includes: Based on the field rules, traverse the fused behavior features and the behavior indicator set to perform feature value verification; If any indicator fails to be verified, it is determined that a violation has occurred, and the violation detection result is output accordingly in combination with the behavior constraint rules, wherein the violation detection result at least includes vehicle identity, violation type, violation time, and violation location.

8. A vehicle violation detection device for a traffic checkpoint, characterized in that: A method for detecting vehicle violations at a traffic checkpoint according to any one of claims 1 to 7, comprising: A vehicle identification component is used to identify a vehicle based on real-time vehicle image data, wherein the real-time vehicle image is acquired by a video acquisition device at a traffic checkpoint; The behavior indicator set definition component is used to extract the corresponding behavior constraint rules from the behavior rule database based on the vehicle identity and define the corresponding behavior indicator set; a difference recognition component for extracting real-time behavior features based on the real-time vehicle passing image data, identifying a difference indicator set in combination with the behavior indicator set, and collecting additional information based on the difference indicator set; The violation detection component is used to extract additional behavior features of the vehicle based on the additional information collection results, and combine the additional behavior features with the real-time behavior features to perform violation detection based on the behavior indicator set.

Images