Security system and method of configuring a security device

Abstract

Disclosed are a security inspection system and a method for configuring a security inspection device. In an embodiment, the security inspection system can include an identity information input device configured to input an identity of a person to be inspected, a parameter determination device configured to determine parameters for security inspection of the person to be inspected based on a security coefficient of the person to be inspected determined according to user data corresponding to the identity of the person to be inspected, and a security inspection device configured to perform security inspection on the person to be inspected based on the determined parameters. According to an embodiment of the present disclosure, by analyzing and mining user data in all aspects, user behavior can be accurately predicted and user danger or potential danger can be assessed, and a more accurate security inspection scheme can be provided.

Description

Technical Field

[0001] This invention relates to the field of security inspection technology, and in particular to a security inspection system and a method for configuring security inspection equipment that can implement customized security inspection strategies based on big data. Background Technology

[0002] Traditional body searches, employing metal detectors or manual pat-downs, are not only time-consuming but also infringe on personal privacy. With advancements in technology, people's demands for security checks are increasing, and traditional, inefficient manual methods can no longer meet the needs of the modern era. Body search equipment scans the body non-contactly, detecting objects concealed within clothing. It can detect not only metallic substances but also non-metallic materials such as ceramics, plastics, powders, liquids, and colloids. This non-contact method does not reveal any physical characteristics, fully protecting the privacy of those being searched. Furthermore, this method is more efficient and can operate continuously, far exceeding the efficiency of traditional manual checks. The automated interpretation system accurately locates suspicious objects, effectively reducing the impact of human error and lessening the workload of security personnel.

[0003] Body scanners are widely used in the security field due to their unique advantages. However, existing equipment relies on a relatively simplistic approach to assessing potential threats from users, analyzing scanned images solely through concealed object detection algorithms to determine the potential threat posed by the inspected individual. Furthermore, current concealed object detection algorithms in body scanners use a generic framework, failing to consider the user's identity or the differences between different groups. They analyze and process all inspected individuals using the same algorithm framework, all based on the assumption that all users have an equal probability of carrying dangerous items. Therefore, the detection algorithms frequently experience false alarms and missed detections, impacting the user experience. Summary of the Invention

[0004] In view of the above problems, the purpose of this disclosure is at least in part to provide a security inspection system and a method for configuring security inspection equipment that can customize security inspection strategies for different users based on user data.

[0005] According to one aspect of this disclosure, a security inspection system is provided, comprising: an identity information input device for inputting the identity identifier of a person to be inspected; a parameter determination device for determining parameters for performing a security inspection on the person to be inspected based on a security factor of the person to be inspected determined according to user data corresponding to the identity identifier of the person to be inspected; and a security inspection device for performing a security inspection on the person to be inspected based on the determined parameters.

[0006] According to embodiments of this disclosure, the security inspection system may further include: a security factor determination device, used to acquire user data related to the inspected person based on the inspected person's identity identifier, and to determine the inspected person's security factor based on the acquired user data. For example, the security factor determination device can determine the inspected person's security factor by substituting the inspected person's user data into a relationship model between user data and security factors, wherein, in the relationship model, the user data is categorized into several classes, and each class is assigned a different security factor.

[0007] According to embodiments of this disclosure, the identity information entry device can also be used to obtain a registered photograph of the person being inspected.

[0008] According to embodiments of this disclosure, the security inspection system may further include: a video device for capturing images of the person being inspected in real time.

[0009] According to embodiments of this disclosure, the security inspection system may further include: an identity verification device for extracting a facial image of the person being inspected from an image captured by a video device, and verifying the identity of the person being inspected by comparing the extracted facial image with facial images in a registered photograph and / or a database of untrusted individuals.

[0010] According to embodiments of this disclosure, the security inspection system may further include: an abnormal behavior determination device, used to determine whether the inspected person exhibits abnormal behavior based on an image of the inspected person captured by a video device.

[0011] According to embodiments of this disclosure, user data may include one or more of the subject's personal data, credit, social relationships, and historical behavior.

[0012] According to embodiments of this disclosure, the parameter determination device can determine the parameters used in the concealed object identification algorithm employed by the security inspection equipment based on the security factor of the inspected person.

[0013] According to embodiments of this disclosure, the security inspection system may further include: a display device for displaying the corresponding security level of the person being inspected during the security inspection process.

[0014] According to another aspect of this disclosure, a method for configuring security inspection equipment in the aforementioned security inspection system is provided, comprising: obtaining the identity identifier of the person being inspected through an identity information input device; obtaining user data related to the person being inspected based on the identity identifier of the person being inspected, and determining the security factor of the person being inspected based on the obtained user data; determining parameters for security inspection of the person being inspected based on the security factor of the person being inspected through a parameter determination device; and configuring the security inspection equipment based on the determined parameters.

[0015] According to embodiments of this disclosure, the parameters may include parameters used in the concealed object identification algorithm employed by the security inspection equipment. For example, the parameters may include one or more of the following: the type of classifier, the parameters of the classifier, and the alarm threshold.

[0016] According to embodiments of this disclosure, by means of analysis such as data mining, high-value information that can represent user behavior characteristics can be extracted from massive data with low value density characteristics. A model of the user's personal security coefficient can be established from this data and combined with the security inspection parameter settings of the security inspection equipment. Unique security inspection strategies can be customized for different users, making security inspection methods more humane and personalized, and achieving the goal of precise security inspection. Attached Figure Description

[0017] The above and other objects, features and advantages of this disclosure will become clearer from the following description of embodiments with reference to the accompanying drawings, in which:

[0018] Figure 1 This is a schematic diagram illustrating a security inspection system according to an embodiment of the present disclosure;

[0019] Figure 2 This is a diagram illustrating various user data;

[0020] Figure 3 This is a schematic diagram illustrating the operation flow of a security inspection system according to an embodiment of the present disclosure;

[0021] Figure 4 This is a schematic diagram illustrating an example clustering of user data. Detailed Implementation

[0022] Embodiments of the present disclosure will now be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the disclosure. Furthermore, descriptions of well-known structures and technologies are omitted in the following description to avoid unnecessarily obscuring the concepts of the present disclosure.

[0023] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit this disclosure. The terms “a,” “an,” and “the,” as used herein, should also include the meanings of “a plurality” and “multiple,” unless the context clearly indicates otherwise. Furthermore, the terms “comprising,” “including,” etc., as used herein indicate the presence of the stated features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0024] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein are to be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0025] The accompanying drawings show some block diagrams and / or flowcharts. It should be understood that some blocks or combinations thereof in the block diagrams and / or flowcharts can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when executed by the processor, these instructions can create means for implementing the functions / operations described in these block diagrams and / or flowcharts.

[0026] Therefore, the technology disclosed herein can be implemented in hardware and / or software (including firmware, microcode, etc.). Additionally, the technology disclosed herein can take the form of a computer program product stored on a computer-readable medium, which can be used by or in conjunction with an instruction execution system. In the context of this disclosure, a computer-readable medium can be any medium capable of containing, storing, transmitting, propagating, or transmitting instructions. For example, a computer-readable medium can include, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, apparatuses, or propagation media. Specific examples of computer-readable media include: magnetic storage devices, such as magnetic tape or hard disk drives (HDDs); optical storage devices, such as optical discs (CD-ROMs); memories, such as random access memory (RAM) or flash memory; and / or wired / wireless communication links.

[0027] Figure 1 This is a schematic diagram illustrating a security inspection system according to an embodiment of the present disclosure.

[0028] like Figure 1 As shown, the security inspection system 100 according to this embodiment includes security inspection equipment (103, 105) installed on the security inspection channel 101.

[0029] Security checkpoint 101 can be located at the entrance of any location requiring secure access, such as airports, train stations, stadiums, museums, etc. Security checkpoint 101 can be a physical passageway enclosed by barriers such as fencing, or a virtual passageway defined by, for example, video equipment. For instance, a person being inspected can move within the monitoring range of the video equipment, which can be considered a "passageway."

[0030] The person being inspected passes through security checkpoint 101 in the direction indicated by the arrow in the diagram and undergoes a security check by the security equipment. If the person passes the security check, they are allowed to enter the location they wish to visit.

[0031] Security inspection equipment may include a scanning device 103 and a control device 105. For human body security inspection, the scanning device 103 may be non-contact. For example, under the control of the control device 105, the scanning device 103 may irradiate the subject with rays (e.g., millimeter waves) and collect rays from the subject (e.g., through scattering). The data received by the scanning device 103 may be transmitted to the control device 105 via a wired link (e.g., cable) or a wireless link (e.g., WiFi), where a reconstruction algorithm may be used to reconstruct an image of the subject's (body surface). In addition, the control device 105 may also use a concealment detection algorithm to identify items that the subject may be hiding in their clothing. The imaging results and recognition results may be output via an output device such as a display, which will be further described below. The control device 105 may include various computing devices, such as computers, servers, etc.

[0032] According to embodiments of this disclosure, the concealment detection algorithm for human body scan images can include two parts: classifier training and online recognition. Both parts are based on partitioned images. The human body scan image can be divided into several regions based on the location of key body points, such as arms, torso, and legs. In one example, classifier training can be performed as follows: 1) Establish a database of positive and negative samples for each partitioned image; 2) Feature extraction: Use dense scale-invariant feature transform (SIFT) features to describe the features of each partitioned image; 3) Train an overcomplete dictionary of dense SIFT features using sparse coding principles; 4) Project the feature descriptors from step 2) into the dictionary to obtain encoded vectors; 5) Train the classifier using a support vector machine (SVM). Online recognition can be performed as follows: 1) First, extract dense SIFT features from the partitioned image to be recognized; 2) Calculate the projection vector of the aforementioned features into the dictionary; 3) Input the projection vector into the trained SVM classifier for category classification.

[0033] Of course, in addition to SVM classifiers, other classifiers such as Linear Spatial Pyramid Matching (LSPM), Locality-Constrained Linear Coding (LLC), or combinations thereof can be used. Many other concealment detection algorithms exist in this field, which will not be described in detail here.

[0034] In this example, the scanning device 103 is shown as a gate-type structure, but this disclosure is not limited thereto. The scanning device 103 can take different forms, such as a rotary scanning type. Various security inspection devices exist in the art, and currently commonly used human body security inspection devices are, for example, based on millimeter-wave holographic imaging technology.

[0035] Additionally, in this example, only the scanning device 103 for the human body is shown. According to embodiments of this disclosure, scanning devices (not shown) for items such as luggage carried by the person being inspected, such as X-ray scanning devices (which may share the same control unit as the human body scanning device 103 or have a separate control unit), may also be included. For example, when the person being inspected enters the security checkpoint 101, they may place their luggage on a conveyor belt to pass through the X-ray scanning device while they themselves walk through the scanning device 103.

[0036] As mentioned above, in conventional technology, security screening equipment employs the same screening strategy for different individuals, specifically using the same parameters (including those used in scanning device 103 and control device 105). However, in actual security screening processes, the likelihood of different groups of people carrying dangerous goods varies greatly. If the principle of "equality for all" is still followed during the security screening of individuals, the accuracy of the screening results will be affected.

[0037] According to embodiments of this disclosure, different security screening strategies can be adopted for different individuals being screened. Specifically, a security screening model specific to each individual can be established based on their identity and behavior. This can significantly improve security screening efficiency, reduce false alarms from security equipment for safe individuals, and strengthen security screening of suspicious persons, enabling timely detection and identification of potential illegal activities during the screening process, thus prompting law enforcement personnel to take appropriate measures.

[0038] To this end, system 100 may include an identity information entry device 107 for entering the identity identifier of the person being inspected. The identity information entry device 107 may include various information input devices. For example, the identity information entry device 107 may be a keyboard and / or a touchscreen. Security personnel can verify the person's identification documents, such as an ID card or passport, and manually enter the person's identity identifier through the identity information entry device 107. This identifier can uniquely identify the person, such as the person's identification document number, such as an ID card number or passport number (generally, the name is not used because there is a high possibility of name duplication). Alternatively, the identity information entry device 107 may be an optical or electronic reading device. For example, the identity information entry device 107 can read a one-dimensional barcode (e.g., a barcode) or a two-dimensional barcode present on the identification document, or it can wirelessly read information on the identification document, for example, based on radio frequency identification (RFID) or near field communication (NFC) technology.

[0039] Of course, it doesn't necessarily have to be an ID card; other documents associated with the user's identity can also be used. For example, at an airport, a boarding pass can be used. By scanning the barcode on the boarding pass and querying the airline's database, the user's identity and corresponding identification, such as an ID card number or passport number, can be obtained. Similarly, at a train station, if a ticket is purchased using a real-name system, the user's identity and corresponding identification can be obtained by scanning the QR code on the ticket and querying the train station's ticketing database.

[0040] Of course, the identity information entry device 107 is not limited to entering the examinee's identity identifier; it can also input other identity information. For example, as described in the following embodiments, the identity information entry device 107 can also acquire the examinee's registration photograph. Here, the term "registration photograph" refers to a photograph that clearly reflects the examinee's appearance when registering their identity with a trusted institution. For example, such a registration photograph could be the photograph used by the examinee when applying for an ID card or passport at a public security agency.

[0041] The identity information entry device 107 can obtain the aforementioned registration photo directly from a photo ID document (e.g., through image recognition), or it can obtain the aforementioned registration photo from a relevant trusted database (e.g., a database of a public security authority) based on the entered identity identifier.

[0042] An identity information entry device 107 can be installed at the entrance of security checkpoint 101. When a person being inspected enters security checkpoint 101, their identity can be entered by the identity information entry device 107 and transmitted to control device 105 via a wired or wireless link. Control device 105 can obtain user data related to the person being inspected identified by the identity from databases 111-1, 111-2, ..., 111-n via network 109. Here, user data can include various data that can reflect user characteristics, such as data related to the identity and behavior of the person being inspected, including but not limited to one or more of the person's personal data, credit, social relationships, and historical behavior.

[0043] With the rapid development of the internet, the Internet of Things, and the mobile internet, various types of data have seen a significant increase in recent years, and research indicates that data volume will grow exponentially in the coming years. In the era of big data, user data is widely distributed across various fields. Figure 2This diagram illustrates various types of user data. For example, the Ministry of Public Security's personal information query system can obtain a user's basic personal information, including name, gender, age, former name, place of birth, contact information, workplace, residential address, marital history, and kinship. Telecom operators can obtain a user's call records, communication records, and internet browsing history. Analyzing this data can reveal a user's contact information, call duration, call frequency, time and location of each internet access, and the IP addresses of visited websites. Banks can provide information such as bank card details, transaction information related to daily life (clothing, food, housing, transportation), credit history, and loan information. E-commerce and B2C companies can provide details of a user's personal consumption, product browsing categories, payment habits, website transaction details, and contact information. The internet can track every trace a user leaves online: frequently visited websites and their themes and content, browsing logs, website browsing history, chat information, shared information, and comments on social media platforms including WeChat and Weibo.

[0044] User data holds immense value, with interconnected data points existing to varying degrees. Analyzing the various aspects of this data can reveal a wide range of information about a user, including interests, personality traits, thought patterns, emotional fluctuations, family relationships, social networks, work experience, economic status, credit history, borrowing history, lifestyle habits, behavioral habits, consumption habits, and travel patterns. By analyzing data on family relationships, work status, interpersonal relationships, and recent significant changes in a user's family or surroundings, potential dangers can be predicted. User consumption data reveals recent spending records; analyzing these records can reveal whether a user has purchased prohibited or dangerous goods, or materials that could be used to create dangerous items. Furthermore, analyzing the consistency between recent spending records and past habits can predict the likelihood of a user engaging in socially harmful or abnormal behavior. By examining frequently visited websites, forwarded messages, and publicly expressed opinions, the potential for antisocial behavior can be identified. By analyzing a user's activity data on social networks, a relationship graph can be drawn centered on that user. By comprehensively analyzing data such as the social activities of other users who frequently interact with the user, the user's level of security can be indirectly predicted based on the security of the user's frequently used contacts.

[0045] According to embodiments of this disclosure, system 100 may include a security factor determination device for acquiring user data related to the inspected person based on the inspected person's identity, and determining the inspected person's security factor based on the acquired user data. In this example, the security factor determination device may be implemented, for example, by a control device 105 running program instructions. However, this disclosure is not limited thereto. The security factor determination device may also be implemented by a separate entity, such as a computing device. For example, control device 105 may extract data information related to a person's identity and behavior from massive amounts of data, for example, through data mining techniques, and classify the population accordingly, applying different levels of security inspection strategies to different categories of people.

[0046] For example, a corresponding scoring mechanism can be established to characterize the safety level of an inspected person using a safety factor. A higher safety factor indicates that the inspected person's various indicators are more normal. The corresponding safety factor can be used as a guide for the parameters used in security equipment when performing security checks on that person (e.g., parameters used in concealed object detection algorithms), achieving differentiated and personalized security checks for different inspected persons. In reality, some inspected persons may have missing or incomplete data. To prevent the impact of missed detection of dangerous items during security checks, these inspected persons are assigned a lower safety factor value, i.e., security checks are strengthened for this group. The determination of the safety factor will be described in further detail below.

[0047] In addition, user data can also include historical security check data, including past departure and destination locations, and details of previous security checks (including body checks of the person being checked and their companions, and baggage checks). When building a user security coefficient model, the weighting of data containing information such as specific locations and prior criminal records should be increased, thus increasing the impact of this data on the security coefficient.

[0048] According to embodiments of this disclosure, system 100 may include a parameter determination device for determining parameters for security checks on the person being checked based on a determined security factor. In this example, the parameter determination device may be implemented, for example, by a control device 105 running program instructions. However, this disclosure is not limited thereto. The parameter determination device may also be implemented by a separate entity, such as a computing device. The basic principle of parameter determination is: for people being checked with relatively high security factors, relatively lenient parameters can be set to appropriately reduce the intensity of security checks on this group, improve the user experience of the security equipment for these relatively safe individuals, and speed up the pass rate for this group; for users with relatively low security factors, relatively strict parameters can be set to strengthen the intensity of security checks on these users.

[0049] These parameters can be any parameters related to the security check intensity in the system, such as hardware parameters (e.g., scanning parameters in scanning device 103) and / or software parameters (e.g., various algorithms used in the system). Of course, it is inconvenient to continuously change hardware parameters for the person being checked, so it is preferable to change the software parameters. For example, these parameters can be parameters applied in the concealed object identification algorithm used by the security equipment, including one or more of the classifier type, classifier parameters, and alarm threshold. For example, an SVM classifier can be used. In this case, the type of classifier can be controlled by adjusting the kernel function. Classifier parameters can include penalty coefficients, etc. The alarm threshold refers to the distance between the classification surface corresponding to the classifier and the optimal segmentation hyperplane. The closer the classification surface is to the suspected object category, the smaller its alarm threshold. For inspectors with a high safety factor, a linear classifier with strong generalization ability and an SVM classifier with a small penalty coefficient can be used, while increasing the alarm threshold. For inspectors with a low safety factor, a nonlinear classifier and an SVM classifier with a large penalty coefficient can be used. Increasing the penalty coefficient can improve the detection rate of dangerous goods and lower the alarm threshold, thus avoiding safety losses caused by missed detection of dangerous goods.

[0050] According to embodiments of this disclosure, system 100 may further include a video device 113, such as a camera, for capturing images of the person being inspected in real time. In this example, the video device 113 is shown as being mounted on the scanning device 103, but this disclosure is not limited thereto. The video device 113 may be mounted at any other location in the security checkpoint, and the number of such devices is not limited to one, but may be more. The video device 113 may be configured such that its or their capturing or monitoring range covers the security checkpoint 101, thereby enabling real-time monitoring of the person being inspected without blind spots as they pass through the security checkpoint 101. As mentioned above, the security checkpoint can also be defined by the monitoring range of the video device 113.

[0051] With video device 113 installed, system 100 may further include an abnormal behavior determination device for determining whether the subject exhibits abnormal behavior based on images of the subject captured by video device 113. In this example, the abnormal behavior determination device may be implemented, for example, by running program instructions through control device 105. In this case, video device 113 may be connected to control device 105 via wired or wireless means. However, this disclosure is not limited thereto. The abnormal behavior determination device may also be implemented by a separate entity, such as a computing device. Control device 105 may utilize computer vision technology to extract the subject's contour image from the video images captured by video device 113 and perform dynamic tracking, using a pre-trained crowd modeling model employing machine learning methods to monitor abnormal behavior, and promptly notify law enforcement personnel once abnormal behavior is detected in the subject.

[0052] Additionally, system 100 may include an authentication device for extracting a facial image of the person being inspected from images captured by video device 113 and verifying the person's identity by comparing the extracted facial image with the person's registered photograph. In this example, the authentication device may be implemented, for example, by running program instructions via control device 105. However, this disclosure is not limited thereto. The authentication device may also be implemented by a separate entity, such as a computing device. Here, the person's registered photograph may be, for example, the registered photograph entered by identity information entry device 107, or a registered photograph obtained from a relevant trusted database as described above. If there is a discrepancy between the person and the identification document, the system may issue an alarm to prompt security personnel to conduct a manual verification of the person being inspected, and may also increase the intensity of baggage inspection.

[0053] On the other hand, the identity verification device can also compare the extracted facial images with facial images in an untrusted population database. For example, the untrusted population database may include facial image databases of fugitive criminal suspects provided by the Ministry of Public Security, or facial image databases of restricted individuals restricted from entering or leaving a certain location published by the courts. If a high degree of matching is found, relevant law enforcement personnel can be notified to conduct a security review.

[0054] The control device 105 can configure the security inspection equipment using parameters determined by the parameter determination device, such as configuring the parameters used in the concealed object detection algorithm employed by the security inspection equipment (configuring the classifier type, classifier parameters, and / or alarm thresholds). By processing the image of the person being inspected obtained through the scanning device 103 using such a configured algorithm, it is possible to identify whether the person is concealing items. The inspection results can then be output to security personnel.

[0055] According to embodiments of this disclosure, a complementary display method combining remote and device-side methods can be employed. Specifically, at the security screening equipment (e.g., at control device 105), only a mannequin reflecting the human outline, corresponding alarm information, and the identity information of the person being inspected can be displayed. Alternatively, a remote control terminal can be set up in a closed monitoring room at a certain distance from the security screening equipment. The actual scanned image is only displayed on the remote control terminal, and the face is blurred in the displayed scanned image. The security personnel at the remote control terminal can view and process the scanned image of the current person being inspected, but are unaware of any personal information related to the identity of the current person being inspected. At the remote control terminal, after completing a scan and corresponding inspection, the scanned image of the person being inspected can be deleted immediately. Furthermore, since the mannequin displayed at the security screening equipment does not contain any surface information of the scanned person, the personal privacy of the person being inspected can be fully protected.

[0056] In other words, security check results can be displayed in a complementary and independent manner at both the remote and device ends. This protects the privacy of the person being checked while allowing for a clear view of the inspection results.

[0057] In the above embodiments, a centralized control device 105 is shown; however, this disclosure is not limited thereto, and a distributed control device may also be used. For example, one or more components in system 100 may have their own control devices, while other components may have other control devices.

[0058] In the above example, the security factor determination device is integrated into the security inspection system 100. However, this disclosure is not limited to this. For example, multiple security inspection systems can share the same security factor determination device. For instance, at an airport, security inspection systems located at different security checkpoints can share a server used as the security factor determination device. Alternatively, the security factor determination device can be provided by a professional data analytics company. The security inspection system 100 can send the entered identity information of the person being inspected to the server of the data analytics company, which then retrieves relevant user information and performs data analysis to obtain the security factor of the person being inspected. Subsequently, the security factor can be sent to the security system 100, and the parameter determination device can then determine the security parameters for the person being inspected based on the received security factor.

[0059] The following will refer to Figure 3 Describe the operation process of the aforementioned security inspection system.

[0060] First, as shown in 310, model training can be performed to establish a relationship model between user data and security factors. This can be done in the aforementioned security inspection system 100 (e.g., control device 105), or it can be done outside the security inspection system 100, for example, by the aforementioned professional data analysis company.

[0061] Specifically, as shown in 311, multiple databases can be utilized (e.g., such as...). Figure 1The data (111-1, 111-2, ..., 111-n) shown in the diagram receive massive amounts of shared data from third-party organizations with which they have established cooperative relationships. This data includes various user-related information such as name, age, place of origin, residence, identification number, interpersonal relationships, communication data, transaction details, credit data, social activities, and social networks. After cleaning and preprocessing, this massive amount of data is imported into a large distributed database. User data can be hierarchically stored based on key information such as age, region, and occupation, facilitating later data retrieval. For example, user-related data such as text, images, audio, and video can be obtained from departments with rich and reliable data sources, such as the Ministry of Public Security, telecom operators, banks, e-commerce companies, B2C enterprises, and internet companies. This massive amount of data is then formatted and imported into the large distributed database for storage.

[0062] Subsequently, as shown in 313, the massive amounts of data storing user information can be analyzed and processed using methods such as data mining, machine learning, and statistical analysis to establish a relationship model between user data and security coefficients. According to embodiments of this disclosure, user data can be clustered into several categories, and each category can be assigned a different security coefficient value. For example, such a model can be established as follows.

[0063] Suppose we extract a massive amount of user data from the database. ,in, , This indicates the total number of user data entries. Indicates the first Data for each user Indicates the first Data dimensions for each user Indicates the first The first user's Dimensional data can be structured data such as numbers, or unstructured data such as text, images, audio, video, and web pages. See also Figure 4 These user data can be represented as points in the data space.

[0064] User data can be clustered into several categories. For example, the K-means clustering algorithm can be used to divide the input user data into categories based on their characteristics. Each category is centered using [a specific method / mechanism]. express. Figure 4 The classification results are schematically illustrated using dashed circles. According to embodiments of this disclosure, different security coefficient values ​​are assigned to each category based on the security level exhibited by the users corresponding to each category. For example... Figure 4 As shown, class 0 has the highest security, class 1 has the second highest security, ..., class (K-1) has the lowest security. Here, class centers are used... Indicates the first There are two security levels. That is, users of the same type have the same security level. ).for .

[0065] Then, a relationship model between user data and security factors is established. For example, user data... The relationship between it and its safety factor can be used It means that, among them ,in , Belongs to the Categories (based on) With each category center The Euclidean distance between them is determined, for example It belongs to the category center closest to it. (representing the category) This represents the Euclidean distance.

[0066] Different users have different security coefficients due to differences in their data, with the coefficient ranging from 0 to 1. A higher security coefficient indicates a higher level of user security. This is determined by the relationship between user data and the security coefficient. It can be seen that for data belonging to the same category, the farther the user data is from its category center, the higher its corresponding security level. The larger the value, the better. However, this is just an example. Other models can be built, for example, assigning substantially the same safety factor to data of the same category.

[0067] The established model can be stored in the security factor determination device (as mentioned above, it can be inside or outside the security inspection system 100), or it can be stored outside the security factor determination device (in which case the security factor determination device can request the model from the device that stores the model).

[0068] In the actual security check process 320, the user's identity is first entered by the identity information entry device 107 (optionally, other identity information such as the registration photo mentioned above is also entered). If the registration photo is entered, identity verification can be performed at 321. For example, as described above, identity verification can be performed by comparing the registration photo with the facial image captured by the video device 113.

[0069] The security factor determination module (e.g., control device 105) can acquire user data based on the entered user identity identifier and determine the user's security factor using the model established as described above, as shown in 323. Alternatively, control device 105 can send the entered user identity identifier to an external security factor determination module and receive the determined security factor from that external module. After determining the user's security factor, the security inspection equipment can be configured accordingly. For example, as shown in 325, parameters for the security inspection equipment can be generated based on the security factor, for example, using a self-learning algorithm. These parameters might include those used in the concealed object detection algorithm, such as classifier type, classifier parameters, and alarm thresholds. These parameters can be represented in parameter vector form.

[0070] Specifically, it is possible to calculate current user data. and The relationship between category centers determines the user's category center. Then user data Substitute into the above model This allows for the calculation of the safety factor of the currently inspected person. It also allows for the assignment of the user's corresponding safety level. The output is sent to the device and remote end to inform security personnel of the user's security level and prompt them to take appropriate preparatory measures, which can improve the device's pass rate and reduce the incidence of dangerous incidents.

[0071] By analyzing users' security levels, security check methods and intensity can be adjusted accordingly. As mentioned above, for users with higher security levels (e.g., ...), ... For users with a relatively low security level, the intensity of security checks can be appropriately reduced to improve their experience with the security equipment and increase their pass rate; while for users with a lower security level (e.g., ...), the security check intensity can be appropriately reduced to improve their experience with the security equipment and increase their pass rate. For users who are identified as high-risk, security checks can be intensified. The system can also provide staff with information on the security level of these users for early warning purposes, prompting them to take appropriate protective measures.

[0072] Security screening equipment can receive security parameters, for example, transmitted in the form of parameter vectors, configure itself using the received parameters, and perform security checks on users. For example, it can emit millimeter waves to illuminate the user, image the user based on the millimeter waves, and process the image using a concealed object detection algorithm, as shown in 327. The inspection results can be displayed, as shown in 329. As mentioned above, this display can employ a complementary display method between the remote end and the device end.

[0073] According to embodiments of this disclosure, the data can also be updated, as shown in 330.

[0074] After each security check, the data of the person being checked can be updated. For example, various information involved in the security check process can be collected in real time and stored in the corresponding database, as shown in 331. The data update includes two parts: information about the security check process and the personal information of the person being checked. For example, the security check process information includes the current security equipment number and location, the information of the security personnel on duty, etc., while the personal information of the person being checked includes identification information, travel location, facial image, video data and imaging data recorded in the security check lane, and the results of the security check.

[0075] Furthermore, in the era of big data, user data is constantly being updated, and the user's security level changes accordingly. Therefore, as shown in 333, large-scale data provided by third-party platforms can be captured and updated in real time to ensure the real-time performance and reliability of security checks.

[0076] According to embodiments of this disclosure, by analyzing and mining comprehensive user data, user behavior can be accurately predicted and the user's risk or potential risk can be assessed, providing a more accurate security check solution.

[0077] The embodiments of this disclosure have been described above. However, these embodiments are for illustrative purposes only and are not intended to limit the scope of this disclosure. Although various embodiments have been described above, this does not mean that the measures in the various embodiments cannot be used advantageously in combination. The scope of this disclosure is defined by the appended claims and their equivalents. Various substitutions and modifications can be made by those skilled in the art without departing from the scope of this disclosure, and all such substitutions and modifications should fall within the scope of this disclosure.

Claims

1. A security inspection system, comprising: Identity information entry device, used to enter the identity information of the examinee; A security factor determination device is used to acquire user data related to the examinee based on the examinee's identity identifier, and determine the examinee's security factor by substituting the examinee's user data into a relationship model between user data and security factors. In the relationship model, user data is classified into several categories, and each category is assigned a different security factor. The category to which the examinee belongs is determined based on the Euclidean distance between the user data and the category centers of each category. Parameter determination equipment is used to determine the parameters for security checks on a person based on a determined safety factor for that person; and Security screening equipment, based on predetermined parameters and using a trained concealed object detection algorithm, processes scanned images of individuals to conduct security checks and identify whether they are concealing items. The parameters include one or more of the classifier type, classifier parameters, and alarm threshold used in the concealed object identification algorithm. The security check of the person being checked includes: extracting the features of the person being checked, calculating the projection vector of the features in a dictionary, and inputting the projection vector into a classifier corresponding to the determined parameters for classification. This includes updating the user data of the person being checked at the end of each security check. The safety factor includes a first safety factor and a second safety factor lower than the first safety factor. The parameters include a first parameter determined based on the first safety factor and a second parameter determined based on the second safety factor. The first parameter includes a linear classifier, a first penalty coefficient of the classifier, and a first alarm threshold. The second parameter includes a nonlinear classifier, a second penalty coefficient of the classifier, and a second alarm threshold. The second penalty coefficient is higher than the first penalty coefficient, and the second alarm threshold is lower than the first alarm threshold. This includes updating the user data of the person being checked at the end of each security check, which involves: collecting information during the security check process in real time and storing it in the database. The user data is related to the identity and behavior of the person being tested. The relationship model between user data and security coefficient is as follows: ， in, Represents user data, Represents user data The corresponding security factor, where k is the category index and K is the total number of categories. , Belongs to the Categories Indicates the first Category centers of each category This represents the Euclidean distance.

2. The security inspection system according to claim 1, wherein, The identity information entry device is also used to obtain the registration photo of the person being inspected.

3. The security inspection system according to claim 2 further includes: Video equipment used to capture images of the examinee in real time.

4. The security inspection system according to claim 3 further includes: Abnormal behavior determination equipment is used to determine whether an inspected person exhibits abnormal behavior based on images captured by video equipment.

5. The security inspection system according to claim 3 further includes: An identity verification device is used to extract a facial image of a person being examined from images captured by a video device and to verify the person's identity by comparing the extracted facial image with facial images in a database of registered photos and / or untrusted individuals.

6. The security inspection system according to claim 1, wherein, User data includes one or more of the subject's personal data, credit, social relationships, and historical behavior.

7. The security inspection system according to claim 1, further comprising: Display devices are used to show the corresponding security level of the person being inspected during the security check process.

8. A method for configuring security inspection equipment in any one of the security inspection systems of claims 1-7, comprising: The identity information of the examinee is obtained through the identity information entry device; Based on the examinee's identity, user data related to the examinee is obtained, and the examinee's security coefficient is determined by substituting the obtained user data into the relationship model between user data and security coefficient. In the relationship model, user data is classified into several categories, and each category is assigned a different security coefficient. The category to which the examinee belongs is determined based on the Euclidean distance between the user data and the category centers of each category. The parameters are determined by the equipment, and based on the determined safety factor of the person being inspected, the parameters for security checks on that person are determined; and Based on the determined parameters, security inspection equipment is configured, and a trained concealed object detection algorithm is used to process the scanned image of the person being inspected to conduct a security check and identify whether the person is concealing any items. The parameters include one or more of the classifier type, classifier parameters, and alarm threshold used in the concealed object identification algorithm. The security check of the person being checked includes: extracting the features of the person being checked, calculating the projection vector of the features in a dictionary, and inputting the projection vector into a classifier corresponding to the determined parameters for classification. This includes updating the user data of the person being checked at the end of each security check. The safety factor includes a first safety factor and a second safety factor lower than the first safety factor. The parameters include a first parameter determined based on the first safety factor and a second parameter determined based on the second safety factor. The first parameter includes a linear classifier, a first penalty coefficient of the classifier, and a first alarm threshold. The second parameter includes a nonlinear classifier, a second penalty coefficient of the classifier, and a second alarm threshold. The second penalty coefficient is higher than the first penalty coefficient, and the second alarm threshold is lower than the first alarm threshold. This includes updating the user data of the person being checked at the end of each security check, which involves: collecting information during the security check process in real time and storing it in the database. The user data is related to the identity and behavior of the person being tested. The relationship model between user data and security coefficient is as follows: ， in, Represents user data, Represents user data The corresponding security factor, where k is the category index and K is the total number of categories. , Belongs to the Categories Indicates the first Category centers of each category This represents the Euclidean distance.

Images