Abnormal person identification method for vehicle administration office

Abstract

The invention discloses an abnormal personnel identification method for a vehicle administration office, and the method comprises the following steps: 1, setting image collection equipment at an entrance of the vehicle administration office, carrying out the image collection of entering personnel and vehicles, and obtaining first image information; 2, analyzing the first image information to obtain a first analysis abnormal person; 3, arranging a plurality of image acquisition devices in the vehicle administration office, and performing image acquisition in the vehicle administration office to obtain second image information; 4, analyzing the second image information to obtain a second analysis abnormal person; and 5, an identity acquisition device and an image acquisition device are arranged at the entrance of the business hall of the vehicle administration office and are used for acquiring the identity information and the third image information of the personnel entering the business hall. According to the method and the system, the abnormal personnel of the vehicle administration office can be identified with higher intelligent degree and more accuracy, and the influence of the abnormal personnel on the handling progress of normal business handling personnel is reduced.

Description

Technical Field

[0001] The present invention relates to the field of identification methods, and in particular to an abnormal person identification method used in a vehicle management office. Background Art

[0002] The vehicle management office's comprehensive business hall handles many aspects of business, including replacement of motor vehicle licenses, replacement of driver's licenses, online number selection for motor vehicles, appointment for driver's license tests, and handling of traffic violations. The public's business needs are large and frequent, resulting in an overflow of abnormal agents, which adds a burden to citizens handling business and disrupts the normal business order. With the prosperity of the automobile market and the continuous growth of automobile ownership, the method of manually identifying queue-jumping, frequent entry and exit, and abnormal transaction behavior has become increasingly limited.

[0003] The existing vehicle management office mainly compares the face and ID card information for entry and exit. Only those who are correctly identified can pass, but it is unable to identify abnormal agents. It mainly relies on the staff to identify through experience in the business hall or "manually monitor the screen" in the video surveillance room. This method is time-consuming, labor-intensive and inefficient. This paper proposes a method for identifying abnormal personnel in the vehicle management office. Summary of the invention

[0004] The technical problem to be solved by the present invention is: how to solve the problem that the existing vehicle management office mainly compares the face and ID card information for entry and exit, and allows people to pass after correct identification, but cannot identify abnormal agents, and mainly relies on the staff to identify through experience in the business hall, or "manually stare at the screen" in the video monitoring room. Such methods are time-consuming and labor-intensive. A method for identifying abnormal personnel in the vehicle management office is provided.

[0005] The present invention solves the above technical problems through the following technical solutions, and the present invention comprises the following steps: Step 1: An image acquisition device is set up at the entrance of the vehicle management office to collect images of entering persons and vehicles and obtain first image information; Step 2: Analyze the first image information to obtain the first abnormal person; Step 3: multiple image acquisition devices are set up in the vehicle administration office to acquire images in the vehicle administration office and obtain second image information; Step 4: Analyze the second image information to obtain the second abnormal person; Step 5: An identity collection device and an image collection device are set up at the entrance of the service hall of the vehicle administration office to collect the identity information and third-party image information of the people entering the service hall; Step 6: Process the identity information and the third image information to obtain the third analysis abnormal person; Step 7: When any two of the first analysis abnormal persons, the second analysis abnormal persons and the third analysis abnormal persons appear in the persons entering the vehicle management office, a warning message is generated and sent to a preset receiving terminal.

[0006] Furthermore, the specific process of analyzing the first image information to obtain the first analysis abnormal person is as follows: Extracting the collected first image information, and extracting the image of the entering person from the first image information; Collect facial images of people entering the room, process the facial images, and obtain real-time facial parameters; Compare the real-time face parameters with other face parameters in the historical database to obtain the real-time face parameter appearance information; The real-time facial parameter information is processed to obtain personnel evaluation parameters. When the personnel evaluation parameters are abnormal, the first abnormal personnel analysis is generated.

[0007] Furthermore, the real-time facial parameter acquisition process is as follows: extracting the acquired facial image; Feature points are extracted from the face image, and the two outer corners of the eyes are set as feature points, which are marked as point q1 and point q2; Then set the two inner corners of the eyes as feature points and mark them as t1 and t2; Then set the two mouth corners as feature points and mark them as f1 and f2; Among them, q1, t1 and f1 are on the same side, and q2, t2 and f2 are on the other side; Connect q1 and q2 to get line segment L1, connect q2 and t2 to get line segment L2, connect t2 and t1 to get line segment L3, and connect q1 and t1 to get line segment L4; Line segments L1, L2, L3 and L4 enclose a first region, and the area of ​​the first region is measured to obtain a first parameter; Connect f1 and f2 to get line segment P1, connect f2 and t2 to get line segment P2, and connect f1 and t1 to get line segment P3; Line segments P1, P2, P3 and L3 enclose a second region, and the area of ​​the second region is measured to obtain a second parameter; The first parameter and the second parameter constitute the real-time face parameter.

[0008] The process of processing the real-time facial parameter appearance information to obtain the personnel evaluation parameters is as follows: extracting the first parameter and the second parameter from the real-time facial parameters, comparing them with other facial parameters in the historical image information of the historical database, and extracting the historical first parameter and the historical second parameter from the other facial parameters; Calculate the difference between the first parameter and the historical first parameter to obtain a first evaluation difference; Calculate the difference between the second parameter and the historical second parameter to obtain a second evaluation difference; Then, the number of times that the first evaluation difference and the second evaluation difference identified in the historical image information of the historical database are both smaller than the preset value is recorded, and the personnel evaluation parameter is obtained.

[0009] Furthermore, the process of obtaining the abnormal personnel in the second analysis is as follows: Extracting the second image information, and importing the real-time facial parameters of the first analyzed abnormal person into the second image information; Extracting the number of image acquisition devices corresponding to the image information whose difference with the real-time face parameter is less than a preset value from the second image information; When the number of image acquisition devices is greater than the preset value, the time information of the real-time facial parameters currently being in the vehicle management office is collected, that is, the evaluation time. When the evaluation time is also greater than the preset value, the second analysis abnormal person is generated; At the same time, the location information of the first abnormal person for analysis is collected, and within a preset time period, the distance between the location information of the first abnormal person for analysis and the location of other persons is less than a preset value and exceeds a preset time length, and the abnormal number of times information is obtained. When the abnormal number of times information is greater than the preset value, the second abnormal person for analysis is generated; Then extract the second image information, analyze all the face images appearing in the second image information, and obtain the number of appearances and total appearance duration within a preset time length; When the number of appearances of a person's face image within a preset time period is greater than a preset value, and the total appearance time is also greater than the preset value, a second abnormal person for analysis is generated.

[0010] Furthermore, the third process of analyzing abnormal personnel is as follows: Extract the collected identity information and third image information, where the identity information is the ID card information collected by the identity collection device, and the third image information is the face information of the ID card holder; After reading the ID card information, start face matching, perform 1V1 comparison between the photo on the ID card and the third image information, obtain the comparison result, and release the person if the comparison is passed. At the same time, perform entry number verification. When the comparison is passed, collect the number of entries of the person within the preset time period, and obtain the number of entries. Calculate the ratio of the number of entries to the preset time period, and obtain the entry evaluation parameter. When the entry evaluation parameter is greater than the preset value, a third analysis abnormal person is generated.

[0011] Furthermore, the process of 1V1 comparison is as follows: first, a face detection and face alignment method based on deep learning is adopted to achieve rapid detection and accurate alignment of face areas through three cascaded convolutional neural networks, namely P-Net, R-Net and O-Net; Among them, the P-Net network generates face candidate frames; Net screens and adjusts the face candidate frames to reduce the false detection rate; O-Net outputs the precise position of the face frame and facial key points, and further increases the image size to achieve face detection and alignment; Afterwards, the FaceNet model is used to complete face comparison; The image mapping method based on deep neural network and the loss function based on triplets are used to train the neural network. The network directly outputs a 128-dimensional vector space, which is the feature representation of the face. The feature representation of the face is the face features of the photo on the ID card and the face features in the third-party image information. For the input feature representation, the FaceNet model will extract its 128-dimensional feature vector and determine whether the two face images are of the same person by calculating the Euclidean distance between the two feature vectors, thus obtaining the comparison result.

[0012] Further, a vehicle entry image is extracted from the first image information, and the vehicle entry image is processed to obtain the license plate information; Analyze the license plate information, import the license plate information into the license plate information database of historically entered vehicles, and when the number of times the license plate information is recognized from the license plate information of historically entered vehicles is greater than a preset value, the driver of the vehicle is determined as the first analyzed abnormal person; Then, the movement trajectory and length of stay of the first analyzed abnormal person in the vehicle management office are tracked by image acquisition equipment; The movement trajectory and the length of stay of the first analyzed abnormal person in the vehicle management office are processed to obtain the length of stay of the first analyzed abnormal person in each area of ​​the vehicle management office. When the length of stay of the first analyzed abnormal person in each area of ​​the vehicle management office is greater than a preset value, a warning message is generated; When the sum of the number of times any license plate information is judged as an abnormal person in the first analysis, an abnormal person in the second analysis, and an abnormal person in the third analysis exceeds a preset value, a warning message is generated when the corresponding license plate information enters the vehicle management office.

[0013] Furthermore, when the warning information is generated, the personal information of the abnormal person is entered into the abnormal list. When the image acquisition equipment in the vehicle management office collects the information of the person in the abnormal list, the warning information is generated to issue a warning.

[0014] Furthermore, the personnel entering the entrance of the vehicle administration office and the personnel entering the service hall of the vehicle administration office are analyzed by frequency analysis, pattern analysis, and time interval analysis to identify abnormal personnel; The specific process of frequency analysis is as follows: Count the number of times a person enters within a preset period, which includes the number of times a person enters a single day and the cumulative number of times a person enters a multiple day period; The daily pass limit sets the maximum number of times each person is allowed to pass through the gate each day. If the number exceeds this limit, it will be marked as abnormal; The cumulative number of passes over multiple days sets the maximum number of passes within a period, and if it exceeds the maximum number, it will be marked as abnormal; The specific process of pattern recognition is as follows: Analyze traffic patterns, i.e., frequent entry and exit within a short period of time and concentrated appearance within a specific time period; If a person frequently enters and exits within a day and stays for a very short time, he / she will be marked as an abnormal person; If a person enters the preset time period for multiple consecutive days, he / she will be marked as an abnormal person; The specific process of time interval analysis is as follows: the shortest time interval between two passes is set. If the time interval of an entry person is less than the set threshold, he / she is marked as an abnormal person; When there are abnormal persons marked in the above process, the abnormal persons are directly imported into the abnormal list.

[0015] Compared with the prior art, the present invention has the following advantages: the abnormal person identification method used in the vehicle management office can enhance security, effectively identify and prevent potential security threats, such as the entry of abnormal persons, through image collection and identity verification in multiple links and multiple means, thereby protecting the normal order of the vehicle management office and the legal rights and interests of car owners, and improving the recognition accuracy. Not only is image collection set at the entrance, but also multiple images and identity information collection and analysis are performed inside the vehicle management office and at the entrance of the service hall. This multi-level recognition mechanism can significantly improve the recognition accuracy of abnormal persons and reduce false alarms and missed alarms. Enhanced deterrence, through clear warning information and instant notification of preset receiving terminals, a strong deterrent is formed for potential abnormal agents, forcing them to abandon abnormal activities, which helps to maintain the fairness, justice and transparency of the vehicle management office. The automated and intelligent identification system can greatly reduce manual intervention and improve management efficiency. Once an abnormal person is identified, the system can immediately generate and send a warning message, and the management personnel can respond quickly and take measures, reducing processing time and costs. By effectively identifying and blocking abnormal agents, it can ensure that car owners are not disturbed when handling business at the vehicle management office, improve efficiency and satisfaction, and thus enhance user experience, making this method more worthy of promotion and use. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1It is the overall flow chart of the present invention. DETAILED DESCRIPTION

[0017] The following is a detailed description of an embodiment of the present invention. This embodiment is implemented on the premise of the technical solution of the present invention, and a detailed implementation method and a specific operation process are given, but the protection scope of the present invention is not limited to the following embodiment.

[0018] like Figure 1 As shown, this embodiment provides a technical solution: a method for identifying abnormal persons in a vehicle management office, comprising the following steps: Step 1: An image acquisition device is set up at the entrance of the vehicle management office to collect images of entering persons and vehicles and obtain first image information; Step 2: Analyze the first image information to obtain the first abnormal person; Step 3: multiple image acquisition devices are set up in the vehicle administration office to acquire images in the vehicle administration office and obtain second image information; Step 4: Analyze the second image information to obtain the second abnormal person; Step 5: An identity collection device and an image collection device are set up at the entrance of the service hall of the vehicle administration office to collect the identity information and third-party image information of the people entering the service hall; Step 6: Process the identity information and the third image information to obtain the third analysis abnormal person; Step 7: When any two of the first analysis abnormal persons, the second analysis abnormal persons and the third analysis abnormal persons are found among the persons entering the vehicle management office, a warning message is generated and sent to a preset receiving terminal; The specific process of analyzing the first image information and obtaining the first abnormal person is as follows: Extracting the collected first image information, and extracting the image of the entering person from the first image information; Collect facial images of people entering the room, process the facial images, and obtain real-time facial parameters; Compare the real-time face parameters with other face parameters in the historical database to obtain the real-time face parameter appearance information; Process the real-time facial parameter information to obtain personnel evaluation parameters. When the personnel evaluation parameters are abnormal, the first abnormal personnel analysis is generated; The process of acquiring the real-time face parameters is as follows: extracting the acquired face image; Feature points are extracted from the face image, and the two outer corners of the eyes are set as feature points, which are marked as point q1 and point q2; Then set the two inner corners of the eyes as feature points and mark them as t1 and t2; Then set the two mouth corners as feature points and mark them as f1 and f2; Among them, q1, t1 and f1 are on the same side, and q2, t2 and f2 are on the other side; Connect q1 and q2 to get line segment L1, connect q2 and t2 to get line segment L2, connect t2 and t1 to get line segment L3, and connect q1 and t1 to get line segment L4; Line segments L1, L2, L3 and L4 enclose a first region, and the area of ​​the first region is measured to obtain a first parameter; Connect f1 and f2 to get line segment P1, connect f2 and t2 to get line segment P2, and connect f1 and t1 to get line segment P3; Line segments P1, P2, P3 and L3 enclose a second region, and the area of ​​the second region is measured to obtain a second parameter; The first parameter and the second parameter constitute the real-time face parameter.

[0019] The process of processing the real-time facial parameter appearance information to obtain the personnel evaluation parameters is as follows: extracting the first parameter and the second parameter from the real-time facial parameters, comparing them with other facial parameters in the historical image information of the historical database, and extracting the historical first parameter and the historical second parameter from the other facial parameters; Calculate the difference between the first parameter and the historical first parameter to obtain a first evaluation difference; Calculate the difference between the second parameter and the historical second parameter to obtain a second evaluation difference; Then, the number of times that the first evaluation difference and the second evaluation difference identified in the historical image information of the historical database are both smaller than the preset value is recorded, that is, the personnel evaluation parameter is obtained; Through the above process, high-precision face recognition is achieved. By extracting facial images and accurately extracting feature points (such as corners of the eyes and corners of the mouth), facial parameters (such as the first parameter and the second parameter) based on facial geometric features are constructed. This method can more accurately reflect subtle changes in the face, thereby improving the accuracy of face recognition.

[0020] Multi-dimensional comparison analysis not only compares the real-time facial parameters with the data in the historical database in a single dimension, but also further analyzes the changes in facial parameters by calculating the difference (first evaluation difference and second evaluation difference). This multi-dimensional comparison analysis can more comprehensively evaluate the similarities and differences of faces and reduce misjudgments and missed judgments.

[0021] The dynamic evaluation mechanism builds a dynamic personnel evaluation parameter by recording the number of times the evaluation difference identified in the historical image information in the historical database is less than the preset value. This mechanism can reflect the stability and change trend of individual facial parameters over time, which helps to identify abnormal personnel more accurately.

[0022] Enhanced security: Through strict facial recognition and comparison analysis, it can effectively identify and prevent potential security threats, such as the entry of abnormal agents, thereby protecting the normal order of the vehicle management office and the legitimate rights and interests of car owners.

[0023] Improved management efficiency, automated and intelligent identification system can greatly reduce manual intervention and improve management efficiency. Once an abnormal person is identified, the system can immediately generate a warning message, and the management staff can respond quickly and take measures.

[0024] The second analysis of the acquisition process of abnormal personnel is as follows: Extracting the second image information, and importing the real-time facial parameters of the first analyzed abnormal person into the second image information; Extracting the number of image acquisition devices corresponding to the image information whose difference with the real-time face parameter is less than a preset value from the second image information; When the number of image acquisition devices is greater than the preset value, the time information of the real-time facial parameters currently being in the vehicle management office is collected, that is, the evaluation time. When the evaluation time is also greater than the preset value, the second analysis abnormal person is generated; At the same time, the location information of the first abnormal person for analysis is collected, and within a preset time period, the distance between the location information of the first abnormal person for analysis and the location of other persons is less than a preset value and exceeds a preset time length, and the abnormal number of times information is obtained. When the abnormal number of times information is greater than the preset value, the second abnormal person for analysis is generated; Then extract the second image information, analyze all the face images appearing in the second image information, and obtain the number of appearances and total appearance duration within a preset time length; When the number of appearances of a person's face image within a preset time period is greater than a preset value, and the total appearance time is also greater than the preset value, a second abnormal person for analysis is generated; Through the above process, the continuity and accuracy of recognition are enhanced, and by importing the real-time facial parameters of the first abnormal person into the second image information, continuous monitoring from the entrance to the vehicle management office is achieved. This continuity not only improves the accuracy of recognition, but also ensures that the behavior of abnormal persons in the entire vehicle management office can be effectively tracked.

[0025] The process combines multiple dimensions of evaluation criteria, including the number of image acquisition devices, evaluation time, location information, and the number and total appearance time of facial images. This multi-dimensional evaluation can more comprehensively reflect the behavioral characteristics of abnormal personnel and reduce the possibility of misjudgment and missed judgment.

[0026] It also improves recognition efficiency. By setting preset values ​​and preset durations, the system can automatically filter out images and people that meet specific conditions, thereby improving recognition efficiency. Managers can quickly pay attention to potential abnormal people and take corresponding measures.

[0027] The third process of analyzing abnormal personnel is as follows: Extract the collected identity information and third image information, where the identity information is the ID card information collected by the identity collection device, and the third image information is the face information of the ID card holder; After reading the ID card information, start face matching, perform 1V1 comparison between the photo on the ID card and the third image information, obtain the comparison result, and release the person if the comparison is passed. At the same time, perform entry number verification. When the comparison is passed, collect the number of entries of the person within the preset time period, and obtain the number of entries. Calculate the ratio of the number of entries to the preset time period, and obtain the entry evaluation parameter. When the entry evaluation parameter is greater than the preset value, a third analysis abnormal person is generated.

[0028] The process of 1V1 comparison is as follows: First, a deep learning-based face detection and face alignment method is used to achieve rapid detection and accurate alignment of face regions through three cascaded convolutional neural networks, namely P-Net, R-Net and O-Net; Among them, the P-Net network generates face candidate frames; Net screens and adjusts the face candidate frames to reduce the false detection rate; O-Net outputs the precise position of the face frame and facial key points, including the eyes, nose, and left and right corners of the mouth, and further increases the image size to achieve face detection and alignment; Afterwards, the FaceNet model is used to complete face comparison; The image mapping method based on deep neural network and the loss function based on triplets are used to train the neural network. The network directly outputs a 128-dimensional vector space, which is the feature representation of the face. The feature representation of the face is the face features of the photo on the ID card and the face features in the third-party image information. For the input feature representation, the FaceNet model extracts its 128-dimensional feature vector and determines whether the two face images are of the same person by calculating the Euclidean distance between the two feature vectors, that is, obtaining the comparison result; The above process improves the accuracy of identifying abnormal agents. The setting of the double verification mechanism realizes the double verification of the authenticity and validity of the identity through the 1V1 comparison of the ID card information and the third image information. This mechanism can effectively identify abnormal agents who use forged or counterfeit ID documents and improve the accuracy of identifying abnormal agents.

[0029] High-precision face matching uses deep learning-based face detection and face alignment methods, as well as the FaceNet model for face matching, which can handle complex and changeable facial features and improve the accuracy and robustness of the matching. This helps to more accurately identify the true identity of abnormal agents.

[0030] Enhance the efficiency of abnormal agent management, quickly release legitimate personnel, and quickly release those who have passed the comparison and have entered the normal number of times, reducing waiting time and improving management efficiency. This helps to improve the service quality and user experience of the vehicle management office.

[0031] Intelligent monitoring of abnormal behavior. By calculating the ratio of the number of entries to the preset duration (i.e., the entry evaluation parameter), the system can intelligently evaluate the entry frequency of personnel. When the entry evaluation parameter exceeds the preset value, the system will automatically generate a third analysis of abnormal personnel, which helps managers to promptly discover and handle potential abnormal agency behaviors.

[0032] To promote the standardization of abnormal agent management, this process collects a large amount of identity information and facial image data, which can be used for further data analysis to help management understand the patterns and trends of abnormal agent activities and formulate more reasonable and scientific abnormal agent management strategies.

[0033] Through the clear 1V1 comparison process and the number of entries verification process, the management of abnormal agents is standardized and regularized. This helps to reduce the interference and misjudgment of human factors and improve the fairness and transparency of abnormal agent management.

[0034] It improves the overall security level of the vehicle management office and prevents abnormal agents from disrupting order. Through effective identification and management of abnormal agents, it can reduce the activities of abnormal agents in the vehicle management office and prevent them from taking advantage of information asymmetry and convenient conditions to engage in abnormal activities and disrupt the normal order of the vehicle management office.

[0035] To protect the legitimate rights and interests of car owners, the existence of abnormal agents is often accompanied by abnormal agency, fraud and other behaviors, which seriously damage the legitimate rights and interests of car owners. By strengthening the management of abnormal agents, the legitimate rights and interests of car owners when handling business at the vehicle management office can be protected from infringement.

[0036] The above-mentioned third process of analyzing abnormal personnel has significant benefits in the process of abnormal agent management at the vehicle management office, such as improving the accuracy of abnormal agent identification, enhancing the efficiency of abnormal agent management, promoting the standardization of abnormal agent management, and improving the overall safety level of the vehicle management office.

[0037] Then extract the vehicle entry image from the first image information, process the vehicle entry image, and obtain the license plate information; Analyze the license plate information, import the license plate information into the license plate information database of historically entered vehicles, and when the number of times the license plate information is recognized from the license plate information of historically entered vehicles is greater than a preset value, the driver of the vehicle is determined as the first analyzed abnormal person; Then, the movement trajectory and length of stay of the first analyzed abnormal person in the vehicle management office are tracked by image acquisition equipment; The movement trajectory and the length of stay of the first analyzed abnormal person in the vehicle management office are processed to obtain the length of stay of the first analyzed abnormal person in each area of ​​the vehicle management office. When the length of stay of the first analyzed abnormal person in each area of ​​the vehicle management office is greater than a preset value, a warning message is generated; When the sum of the number of times that any license plate information is determined as a first analysis abnormal person, a second analysis abnormal person, and a third analysis abnormal person exceeds a preset value, a warning message is generated when the corresponding license plate information enters the vehicle management office; The above process enhances the comprehensiveness and accuracy of abnormal agent identification. It not only identifies abnormal persons through face comparison and identity information verification, but also introduces the analysis of vehicle entry images, starting from license plate information, further enriching the means of identifying abnormal agents. This helps to find abnormal agents who may evade detection by changing their identities or forging documents; Import the license plate information into the historical database and judge the abnormality of the vehicle and driver by identifying the number of times the license plate information appears. This method is based on long-term data accumulation and can more accurately identify abnormal agents who appear frequently and behave abnormally.

[0038] Improve the efficiency and timeliness of abnormal agent management, and use image acquisition equipment to track the movement trajectory and length of stay of the first analysis abnormal personnel in the vehicle management office in real time, so as to ensure that the management personnel can understand the activities of abnormal agents in a timely manner. This helps to take appropriate measures to intervene before the abnormal agents start to carry out abnormal behavior.

[0039] When the length of time that the first analysis abnormal person stays in each area of ​​the vehicle management office exceeds the preset value, or the number of times that any license plate information is judged as the first, second, and third analysis abnormal person exceeds the preset value, the system will automatically generate a warning message. This greatly improves the efficiency and timeliness of abnormal agent management.

[0040] The whole process is based on the collection and analysis of a large amount of data, which provides rich data support for the management. Through the analysis of these data, the management can deeply understand the activity patterns and trends of abnormal agents and formulate more refined and scientific abnormal agent management strategies.

[0041] By analyzing the length of time that the first analysis of abnormal personnel stays in various areas of the vehicle management office, the key areas of abnormal agent activities can be identified. This helps the management to implement regional management measures and strengthen the supervision of key areas in a targeted manner.

[0042] When a warning message is generated, the personal information of the abnormal person is entered into the abnormal list. When the image acquisition equipment in the vehicle management office collects the information of the person in the abnormal list, a warning message is generated to issue a warning.

[0043] Frequency analysis, pattern analysis, and time interval analysis are used to identify abnormal persons when entering the entrance of the vehicle administration office and the service hall of the vehicle administration office; The specific process of frequency analysis is as follows: Count the number of times a person enters within a preset period, which includes the number of times a person enters a single day and the cumulative number of times a person enters a multiple day period; The daily pass limit sets the maximum number of times each person is allowed to pass through the gate each day. If the number exceeds this limit, it will be marked as abnormal; The cumulative number of passes over multiple days sets the maximum number of passes within a period, and if it exceeds the maximum number, it will be marked as abnormal; The specific process of pattern recognition is as follows: Analyze traffic patterns, i.e., frequent entry and exit within a short period of time and concentrated appearance within a specific time period; If a person frequently enters and exits within a day and stays for a very short time, he / she will be marked as an abnormal person; If a person enters the preset time period for multiple consecutive days, he / she will be marked as an abnormal person; The specific process of time interval analysis is as follows: the shortest time interval between two passes is set. If the time interval of an entry person is less than the set threshold, he / she is marked as an abnormal person; When there are abnormal persons marked in the above process, the abnormal persons are directly imported into the abnormal list; The above process comprehensively evaluates the personnel entering the vehicle management office through frequency analysis, pattern recognition and time interval analysis, which can more accurately identify abnormal personnel. This multi-dimensional analysis method is more comprehensive and reliable than a single indicator.

[0044] Parameters such as the number of passes within a preset period, the limit on the number of passes per day, the cumulative number of passes over multiple days, concentrated appearances within a specific time period, and the shortest time interval between two passes can all be customized according to actual conditions, improving the flexibility and accuracy of recognition.

[0045] The entire identification process is based on data analysis and algorithm models, which realizes automatic identification and marking of abnormal personnel, reduces manual intervention and subjective judgment, and improves the level of intelligent management.

[0046] Once an abnormal person is identified, the system will immediately import him into the abnormal list and notify the management personnel in real time to take corresponding measures, thus achieving rapid response and handling of abnormal behavior; Through methods such as frequency analysis, pattern recognition and time interval analysis, abnormal personnel among a large number of people can be quickly screened out, reducing the workload of managers and improving management efficiency.

[0047] By accurately identifying and handling abnormal personnel, their stay time in the vehicle management office can be reduced, thus avoiding unnecessary waiting and interference for other normal car owners.

[0048] By strengthening personnel management and identifying abnormal agents, the overall service quality of the vehicle management office can be improved and the satisfaction and trust of car owners can be enhanced.

[0049] It should be noted that in the technical solution disclosed in the present invention, the collection, collection, updating, processing, use, transmission and storage of personal user information involved are in compliance with the provisions of relevant laws and regulations, are used for legitimate functions, and do not violate public order and good morals. Necessary measures are taken for user personal information to prevent illegal access to user personal information data, thereby maintaining the security of user personal information, network security and national security, and complying with the provisions of relevant laws and regulations.

[0050] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In the description of the present invention, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0051] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0052] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. A method for identifying abnormal persons in a vehicle administration office, characterized in that: The following steps are involved: Step 1: An image acquisition device is set up at the entrance of the vehicle management office to collect images of entering persons and vehicles and obtain first image information; Step 2: Analyze the first image information to obtain the first abnormal person; Step 3: multiple image acquisition devices are set up in the vehicle administration office to acquire images in the vehicle administration office and obtain second image information; Step 4: Analyze the second image information to obtain the second abnormal person; Step 5: An identity collection device and an image collection device are set up at the entrance of the service hall of the vehicle administration office to collect the identity information and third-party image information of the people entering the service hall; Step 6: Process the identity information and the third image information to obtain the third analysis abnormal person; Step 7: When any two of the first analysis abnormal persons, the second analysis abnormal persons and the third analysis abnormal persons appear in the persons entering the vehicle management office, a warning message is generated and sent to a preset receiving terminal.

2. The abnormal person identification method for vehicle management office according to claim 1 is characterized by: The specific process of analyzing the first image information and obtaining the first abnormal person is as follows: Extracting the collected first image information, and extracting the image of the entering person from the first image information; Collect facial images of people entering the room, process the facial images, and obtain real-time facial parameters; Compare the real-time face parameters with other face parameters in the historical database to obtain the real-time face parameter appearance information; The real-time facial parameter information is processed to obtain personnel evaluation parameters. When the personnel evaluation parameters are abnormal, the first abnormal personnel analysis is generated.

3. The abnormal person identification method for vehicle management office according to claim 2 is characterized in that: The process of acquiring the real-time face parameters is as follows: extracting the acquired face image; Feature points are extracted from the face image, and the two outer corners of the eyes are set as feature points, which are marked as point q1 and point q2; Then set the two inner corners of the eyes as feature points and mark them as t1 and t2; Then set the two mouth corners as feature points and mark them as f1 and f2; Among them, q1, t1 and f1 are on the same side, and q2, t2 and f2 are on the other side; Connect q1 and q2 to get line segment L1, connect q2 and t2 to get line segment L2, connect t2 and t1 to get line segment L3, and connect q1 and t1 to get line segment L4; Line segments L1, L2, L3 and L4 enclose a first region, and the area of ​​the first region is measured to obtain a first parameter; Connect f1 and f2 to get line segment P1, connect f2 and t2 to get line segment P2, and connect f1 and t1 to get line segment P3; Line segments P1, P2, P3 and L3 enclose a second region, and the area of ​​the second region is measured to obtain a second parameter; The first parameter and the second parameter constitute the real-time face parameter; The process of processing the real-time facial parameter appearance information to obtain the personnel evaluation parameters is as follows: extracting the first parameter and the second parameter from the real-time facial parameters, comparing them with other facial parameters in the historical image information of the historical database, and extracting the historical first parameter and the historical second parameter from the other facial parameters; Calculate the difference between the first parameter and the historical first parameter to obtain a first evaluation difference; Calculate the difference between the second parameter and the historical second parameter to obtain a second evaluation difference; Then, the number of times that the first evaluation difference and the second evaluation difference identified in the historical image information of the historical database are both smaller than the preset value is recorded, and the personnel evaluation parameter is obtained.

4. The abnormal person identification method for vehicle management office according to claim 1 is characterized by: The second analysis of the acquisition process of abnormal personnel is as follows: Extracting the second image information, and importing the real-time facial parameters of the first analyzed abnormal person into the second image information; Extracting the number of image acquisition devices corresponding to the image information whose difference with the real-time face parameter is less than a preset value from the second image information; When the number of image acquisition devices is greater than the preset value, the time information of the real-time facial parameters currently being in the vehicle management office is collected, that is, the evaluation time. When the evaluation time is also greater than the preset value, the second analysis abnormal person is generated; At the same time, the location information of the first abnormal person for analysis is collected, and within a preset time period, the distance between the location information of the first abnormal person for analysis and the location of other persons is less than a preset value and exceeds a preset time length, and the abnormal number of times information is obtained. When the abnormal number of times information is greater than the preset value, the second abnormal person for analysis is generated; Then extract the second image information, analyze all the face images appearing in the second image information, and obtain the number of appearances and total appearance duration within a preset time length; When the number of appearances of a person's face image within a preset time period is greater than a preset value, and the total appearance time is also greater than the preset value, a second abnormal person for analysis is generated.

5. The abnormal person identification method for vehicle management office according to claim 1 is characterized by: The third process of analyzing abnormal personnel is as follows: Extract the collected identity information and third image information, where the identity information is the ID card information collected by the identity collection device, and the third image information is the face information of the ID card holder; After reading the ID card information, start face matching, perform 1V1 comparison between the photo on the ID card and the third image information, obtain the comparison result, and release the person if the comparison is passed. At the same time, perform entry number verification. When the comparison is passed, collect the number of entries of the person within the preset time period, and obtain the number of entries. Calculate the ratio of the number of entries to the preset time period, and obtain the entry evaluation parameter. When the entry evaluation parameter is greater than the preset value, a third analysis abnormal person is generated.

6. The abnormal person identification method for vehicle management office according to claim 5 is characterized by: The process of 1V1 comparison is as follows: First, a deep learning-based face detection and face alignment method is used to achieve rapid detection and accurate alignment of face regions through three cascaded convolutional neural networks, namely P-Net, R-Net and O-Net; Among them, the P-Net network generates face candidate frames; Net screens and adjusts the face candidate frames to reduce the false detection rate; O-Net outputs the precise position of the face frame and facial key points, and further increases the image size to achieve face detection and alignment; Afterwards, the FaceNet model is used to complete face comparison; The image mapping method based on deep neural network and the loss function based on triplets are used to train the neural network. The network directly outputs a 128-dimensional vector space, which is the feature representation of the face. The feature representation of the face is the face features of the photo on the ID card and the face features in the third-party image information. For the input feature representation, the FaceNet model will extract its 128-dimensional feature vector and determine whether the two face images are of the same person by calculating the Euclidean distance between the two feature vectors, thus obtaining the comparison result.

7. The abnormal person identification method for vehicle management office according to claim 2 is characterized by: Then extract the vehicle entry image from the first image information, process the vehicle entry image, and obtain the license plate information; Analyze the license plate information, import the license plate information into the license plate information database of historically entered vehicles, and when the number of times the license plate information is recognized from the license plate information of historically entered vehicles is greater than a preset value, the driver of the vehicle is determined as the first analyzed abnormal person; Then, the movement trajectory and length of stay of the first analyzed abnormal person in the vehicle management office are tracked by image acquisition equipment; The movement trajectory and the length of stay of the first analyzed abnormal person in the vehicle management office are processed to obtain the length of stay of the first analyzed abnormal person in each area of ​​the vehicle management office. When the length of stay of the first analyzed abnormal person in each area of ​​the vehicle management office is greater than a preset value, a warning message is generated; When the sum of the number of times any license plate information is judged as an abnormal person in the first analysis, an abnormal person in the second analysis, and an abnormal person in the third analysis exceeds a preset value, a warning message is generated when the corresponding license plate information enters the vehicle management office.

8. A method for identifying abnormal persons in a vehicle administration office according to any one of claims 1 to 7, characterized in that: When a warning message is generated, the personal information of the abnormal person is entered into the abnormal list. When the image acquisition equipment in the vehicle management office collects the information of the person in the abnormal list, a warning message is generated to issue a warning.

9. The abnormal person identification method for vehicle management office according to claim 1 is characterized by: Frequency analysis, pattern analysis, and time interval analysis are used to identify abnormal persons when entering the entrance of the vehicle administration office and the service hall of the vehicle administration office; The specific process of frequency analysis is as follows: Count the number of times a person enters within a preset period, which includes the number of times a person enters a single day and the cumulative number of times a person enters a multiple day period; The daily pass limit sets the maximum number of times each person is allowed to pass through the gate each day. If the number exceeds this limit, it will be marked as abnormal; The cumulative number of passes over multiple days sets the maximum number of passes within a period, and if it exceeds the maximum number, it will be marked as abnormal; The specific process of pattern recognition is as follows: Analyze traffic patterns, i.e., frequent entry and exit within a short period of time and concentrated appearance within a specific time period; If a person frequently enters and exits within a day and stays for a very short time, he / she will be marked as an abnormal person; If a person enters the preset time period for multiple consecutive days, he / she will be marked as an abnormal person; The specific process of time interval analysis is as follows: the shortest time interval between two passes is set. If the time interval of an entry person is less than the set threshold, he / she is marked as an abnormal person; When there are abnormal persons marked in the above process, the abnormal persons are directly imported into the abnormal list.

Images