Intelligent access control method and system

Abstract

The invention relates to an intelligent access control method and system.The intelligent access control method comprises the steps that a user image is collected, a face part image and a trunk part image are obtained from the user image, the face part image is compared with a pre-stored face image, and the trunk part image is obtained; screening out a pre-stored face image with the highest face similarity with the face part image, then obtaining a hit trunk image, then respectively calculating a trunk index of a trunk part image of the user image and a trunk index of the hit trunk part image, setting a trunk similarity threshold value, and based on the trunk similarity threshold value, obtaining a hit trunk part image of the user image; according to the intelligent access control method disclosed by the invention, secondary inspection can be carried out on the entering personnel through the face and trunk characteristics, so that the passing accuracy of the access control system is improved.

Description

Technical Field

[0001] The present application relates to the field of access control technology, and in particular to an intelligent access control method and system. Background Art

[0002] Access control system is in the field of intelligent buildings, which means Access Control System, or ACS for short. It refers to the prohibition of "doors" and the precautions against "doors". The "doors" here, in a broad sense, include various passages that can be passed, including doors for people to pass through, doors for vehicles to pass through, etc. Therefore, access control includes vehicle access control. In parking lot management applications, vehicle access control is an important means of vehicle management. It is not for the purpose of collecting parking fees, but mainly for managing vehicle access rights.

[0003] The entrance and exit security management system is a new modern security management system, which integrates microcomputer automatic identification technology and modern security management measures. It involves many new technologies such as electronics, machinery, optics, computer technology, communication technology, biotechnology, etc. It is an effective measure to solve the security management of the entrances and exits of important departments. It is suitable for various confidential departments, such as banks, hotels, parking lot management, computer rooms, armories, confidential rooms, offices, intelligent communities, factories, etc.

[0004] Existing access control systems generally use facial recognition to determine whether a user is allowed to pass. However, relying solely on facial recognition may sometimes result in misidentification, making it difficult to ensure the security of the access control system. Summary of the invention

[0005] Based on this, the present application provides an intelligent access control method and system to address the problem that misidentification may sometimes occur when relying solely on face recognition, which makes it difficult to ensure the security of the access control system.

[0006] On the one hand, the present application provides an intelligent access control method, the intelligent access control method comprising:

[0007] Collecting a user image of the user, obtaining a partial face image and a partial torso image from the user image, comparing the partial face image with a pre-stored face image, and selecting a pre-stored face image having the highest face similarity to the partial face image;

[0008] Obtaining a torso image corresponding to a pre-stored face image having the highest face similarity to the face image, and defining the torso image as a hit torso image;

[0009] respectively calculating a torso index of a torso image of the user image and a torso index of a hit torso image;

[0010] A torso similarity threshold is set, and based on the torso index of the torso image of the user image and the torso index and torso similarity threshold of the hit torso image, it is determined whether the torso similarity between the torso image of the user image and the hit torso image is greater than or equal to the similarity threshold; if the torso similarity between the torso image of the user image and the hit torso image is greater than or equal to the similarity threshold, the user is defined as a passerby.

[0011] On the other hand, the present application also provides an intelligent access control system, the intelligent access control system comprising:

[0012] Access control components; used to control the entry and exit of pedestrians;

[0013] A collection component, which is used to collect images of the user and the companions;

[0014] The processing component, the access control component and the collection component are both communicatively connected to the processing component, and the processing component executes any one of the aforementioned intelligent access control methods according to the user image and the companion image input by the collection component.

[0015] The present application relates to an intelligent access control method and system, wherein the intelligent access control method collects user images, obtains a facial partial image and a torso partial image from the user image, compares the facial partial image with a pre-stored facial image, screens out the pre-stored facial image with the highest facial similarity to the facial partial image, obtains a hit torso image, and then respectively calculates the torso index of the torso partial image of the user image and the torso index of the hit torso partial image, and sets a torso similarity threshold. Based on the torso similarity threshold, it is determined whether the user is a passerby. The intelligent access control method described in the present application can perform a secondary check on the entrants through facial and torso features, thereby improving the pass accuracy of the access control system. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A flowchart of an intelligent access control method provided for an implementation of the present application.

[0017] Figure 2 A schematic diagram of the structure of an intelligent access control system provided for an implementation of this application.

[0018] Reference numerals

[0019] 100, access control component; 200, data collection component; 300, processing component; 301, storage unit. DETAILED DESCRIPTION

[0020] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0021] The present application provides an intelligent access control method and system.

[0022] like Figure 1 As shown, the present application provides an intelligent access control method, the intelligent access control method comprising:

[0023] S100, collecting a user image of a user, obtaining a partial face image and a partial torso image from the user image, comparing the partial face image with a pre-stored face image, and selecting a pre-stored face image having the highest face similarity to the partial face image.

[0024] S200, obtaining a torso image corresponding to a pre-stored face image having the highest face similarity to a face image, and defining the torso image as a hit torso image.

[0025] Specifically, before obtaining the torso image corresponding to the pre-stored face image having the highest face similarity to the face image, the method further includes:

[0026] S001, create a torso integrity recognition model and training database.

[0027] S002, obtaining multiple complete torso images, and putting all the complete torso images into a training database.

[0028] S003, randomly selecting a complete torso image from the training database, intercepting a partial torso image of the complete torso image, and marking the proportion of the partial torso image to the complete torso image. The intercepted partial torso image is also put into the training database.

[0029] The complete torso images, partial torso images, and the ratio of partial torso images to complete torso images are shown in Table-1 training database table.

[0030] Table-1

[0031]

[0032] S004, returning to randomly selecting a complete torso image from the training database until all complete torso images in the training database are selected, thereby obtaining a plurality of partial torso images.

[0033] Specifically, a complete torso image may be cut out into multiple partial torso images, and the proportion of each partial torso image may be different.

[0034] S005, taking a complete torso image, a partial torso image of the complete torso image, and the ratio of the partial torso image to the complete torso image as a test pair, and dividing all data in the training database into a training set and a test set in a random proportion.

[0035] S006, inputting the training set into the trunk integrity recognition model to train the trunk integrity recognition model, and obtaining a trained trunk integrity recognition model.

[0036] Specifically, when training the torso integrity recognition model, each training session uses a test pair as the basic training unit.

[0037] S007, randomly selecting a partial torso image from the test set and inputting it into the trained torso integrity recognition model, the trained torso integrity recognition model outputs the ratio of the partial torso image to the complete torso image. The ratio of the partial torso image to the complete torso image output by the trained torso integrity recognition model is recorded as the predicted ratio.

[0038] S008, determining whether the predicted proportion is equal to the proportion of the partial torso image to the complete torso image.

[0039] Optionally, a model qualification rate can be set to improve the model pass rate.

[0040] For example, if the model qualification rate is set to 80%, then for all images in the test set, as long as more than 80% of the test pairs in the test set are qualified, the trained torso integrity recognition model can be considered qualified.

[0041] S009, if the predicted proportion is equal to the proportion of the partial torso image to the complete torso image, the trained torso integrity recognition model is considered qualified.

[0042] S010, if the predicted proportion and the proportion of the partial torso image to the complete torso image are not equal, return to step S003 until the predicted proportion and the proportion of the partial torso image to the complete torso image are equal.

[0043] In this embodiment, by creating a torso integrity recognition model in advance, when the user performs torso similarity matching, even if the torso image of the collected user image is incomplete, matching can still be performed, thereby improving the applicability of the present application.

[0044] S300, respectively calculating a torso index of a torso image of a user image and a torso index of a hit torso image.

[0045] S400, setting a torso similarity threshold, based on the torso index of the torso image of the user image and the torso index and torso similarity threshold of the hit torso image, determining whether the torso similarity between the torso image of the user image and the hit torso image is greater than or equal to the similarity threshold, if the torso similarity between the torso image of the user image and the hit torso image is greater than or equal to the similarity threshold, the user is defined as a passerby.

[0046] Specifically, after the user is defined as a passerby, the access control system can be opened, for example, the access control lock can be opened, so that the passerby can pass through.

[0047] In this embodiment, by collecting user images, obtaining facial partial images and torso partial images from the user images, comparing the facial partial images with pre-stored facial images, screening out the pre-stored facial images with the highest facial similarity to the facial partial images, and then obtaining the hit torso images. Thereafter, the torso index of the torso partial image of the user image and the torso index of the hit torso partial image are calculated respectively, and a torso similarity threshold is set. Based on the torso similarity threshold, it is determined whether the user is a passerby. The intelligent access control method described in the present application can perform secondary checks on entrants through facial and torso features, thereby improving the pass accuracy of the access control system.

[0048] In one embodiment of the present application, the S100 includes:

[0049] S110, collecting user images.

[0050] Specifically, the collected user image can be an image captured by the user using a user terminal, such as a camera, at a certain distance and angle, and the image includes a complete and clear face and torso part.

[0051] When collecting images, the standards of all users are consistent, so the collected user images can reflect some torso features to a certain extent.

[0052] For example, if the collection standard is that the collection distance is 2 meters and the collection angle is 30 degrees in elevation, user A is 180 cm tall and weighs 75 kg, and user B is 160 cm tall and weighs 55 kg, the difference in torso features between user A and user B can be clearly seen if the collection standard is agreed upon.

[0053] In order to improve the consistency between the captured image and the pre-stored face image, when capturing the pre-stored image, the same capture standard as that for capturing the user image should be used.

[0054] S120, dividing the user image into a face part image and a torso part image by using a YOLO image segmentation algorithm.

[0055] Specifically, YOLO solves object detection as a regression problem. Based on a separate end-to-end network, it completes the output from the input of the original image to the object position and category.

[0056] S130, extracting facial features from the partial face image, and matching the pre-stored facial image in the database with the most similarity to the partial face image according to the facial features in the partial face image.

[0057] Specifically, facial features include information such as facial features and facial contours of the face.

[0058] Optionally, the number of matches can be set to multiple, for example, 3, so that 3 results can be obtained when matching the closest pre-stored image in the database according to the facial features in the partial facial image.

[0059] The reason for setting multiple matching numbers is that although the detection accuracy of the initial detection is reduced, the detection speed is effectively improved. Since there is a second step of torso detection later, the detection efficiency of the access control system can be improved by setting multiple matching numbers during peak traffic hours.

[0060] In this embodiment, before determining whether to allow a user to enter through the access control system, an image of the user is first collected. The image is a full-body photo of the user, and the image should include the user's head and torso.

[0061] After collecting the user image, YOLO is used to divide the face and torso parts in the image for subsequent comparison.

[0062] In one embodiment of the present application, the S200 includes:

[0063] S210, loading a partial face image.

[0064] This is accomplished through the loadTrainingData() function.

[0065] S220, projecting the partial face image into the principal component analysis subspace.

[0066] Specifically, through cvEigenDecomposite(), this step is to observe the ratio of those eigenfaces that best represent the image.

[0067] S230, searching for a pre-stored image with the most similar ratio from a database based on the feature value, and outputting the person corresponding to the pre-stored image.

[0068] Specifically, it is implemented through findNearestNeighbor(), using Euclidean distance as the similarity metric.

[0069] Optionally, the Mahalanobis distance can also be used, USE_MAHALANOBIS_DISTANCE.

[0070] In this embodiment, since the user image has been divided into the face part and the torso part by YOLO in the previous step, a preliminary detection is performed first, that is, the image of the face part is compared with the face images pre-stored in the database, so as to screen out the pre-stored image that is closest to the face part of the user image and the person corresponding to the pre-stored image.

[0071] In one embodiment of the present application, the S300 includes:

[0072] S310, obtaining an image of the hit torso part.

[0073] S320, obtaining the volume, maximum width and maximum height of the torso part in the torso part image of the user image in the entire image, and obtaining the volume, width and height of the torso part in the hit torso part image in the entire image.

[0074] S330, calculating the torso index of the torso image of the user image and the torso index of the hit torso image respectively by using Formula 1.

[0075]

[0076] Among them, Q is the torso index, V is the volume of the torso part in the entire image, h is the maximum height of the torso part, and w is the maximum width of the torso part.

[0077] In this embodiment, by performing the second step detection on the pre-stored images that have been screened, the two-step detection process can effectively avoid the occurrence of detection errors, thereby improving the detection accuracy.

[0078] In one embodiment of the present application, the S400 includes:

[0079] S410, setting a torso similarity threshold.

[0080] S420, calculating the torso similarity between the torso image of the user image and the hit torso image using Formula 2.

[0081]

[0082] Among them, Q1 is the torso index of the torso image of the user image, Q2 is the torso index of the hit torso image, and P is the torso similarity between the torso image of the user image and the hit torso image.

[0083] S430, determining whether the torso similarity between the torso image of the user image and the hit torso image is greater than or equal to a torso similarity threshold.

[0084] S440: If the torso similarity between the torso image of the user image and the hit torso image is greater than or equal to a torso similarity threshold, the user is defined as a passerby.

[0085] Specifically, if a person is defined as a passer-by, the person can pass through the access control.

[0086] S450: If the torso similarity between the torso image of the user image and the hit torso image is less than a torso similarity threshold, the user is defined as a prohibited person.

[0087] Specifically, if a person is defined as a prohibited person, the person is prohibited from passing through the access control.

[0088] In this embodiment, whether to allow the user to pass is determined by determining the similarity between the user's torso and the hit torso image.

[0089] Since people's body shapes are different, after detecting the image of the face part, the user's body shape can also be detected.

[0090] If there are twins A and B, but user A is 170cm tall and weighs 60kg, while user B is 180cm tall and weighs 65kg, then if only the face part of the two is detected, they will be judged as the same person. Therefore, a second step of detection is required for user A and user B to avoid false detection.

[0091] In one embodiment of the present application, the intelligent access control method further includes:

[0092] S500, determining whether the traveler has any companions.

[0093] S510: If the passerby has a companion, collect images of the companion.

[0094] S520: storing the images of the companions of the passerby in a database, and storing the user image and the images of the companions in correspondence with each other.

[0095] In this embodiment, if the access control system described in the present application is used for the company's access control system, due to the situation where customers visit, if the customer has not pre-stored the image in the pre-stored image database, the customer will not be able to pass through the access control system, so it is set to add the pre-stored user's companions to the pre-stored database so that the customer can also pass through the access control system.

[0096] In one embodiment of the present application, before S520, the method further includes:

[0097] K100 sends a query command to the user end to enter the image of the companion into the database, and determines whether it has received the consent instruction returned by the user end:

[0098] K110, if the consent to save command returned by the user terminal is received, a companion information entry window pops up. The companion information entry window is used to input the relationship data between the user and the companion.

[0099] K120, receiving the relationship data between the user and the companions, generating a relationship tag based on the relationship data between the user and the companions, and assigning the relationship tag to the companion image.

[0100] In this embodiment, before adding the companion to the pre-stored database, it is necessary to first confirm whether the user agrees to enter the companion's image into the pre-stored database to avoid mistakenly entering non-companion into the pre-stored database.

[0101] In one embodiment of the present application, the intelligent access control method further includes:

[0102] S540, obtaining the reservation personnel information and reservation personnel image added by the user, and storing the reservation personnel information and reservation personnel image added by the user in the reservation personnel information unit in the database correspondingly.

[0103] S550, setting a reservation time period.

[0104] S560, collecting the information of the person to be confirmed in the reservation time period, and determining whether the information of the person to be confirmed is the same as any information of the person in the reservation information unit.

[0105] S570: If the reservation person information to be confirmed is the same as any reservation person information in the reservation person information unit, then obtain the reservation person image corresponding to the reservation person information and record it as the hit reservation person image.

[0106] S580, collecting images of persons whose reservations are to be confirmed.

[0107] S590, setting a reservation similarity threshold.

[0108] S591, comparing the image of the person to be confirmed with the image of the person to be hit, and determining whether the similarity between the image of the person to be confirmed and the image of the person to be hit is greater than or equal to a reservation similarity threshold.

[0109] Specifically, when dealing with the person who has made an appointment, the image of the person who has made an appointment should also be divided into an image of the face part of the person who has made an appointment and an image of the body part of the person who has made an appointment, and a similarity threshold comparison should be performed on the image of the face part of the person who has made an appointment and the image of the body part of the person who has made an appointment respectively. When both items are qualified, the person who has made an appointment is considered to be a passer-by.

[0110] Optionally, weights may be set for the facial image and the torso image of the appointment person respectively, and the similarity of the appointment person may be calculated using Formula 3.

[0111] Z=M*C+N*(1-C) Formula 3.

[0112] Among them, Z is the similarity of the appointment person, M is the torso similarity between the appointment torso image of the appointment person and the hit torso image of the hit appointment person, N is the face similarity between the appointment face image of the appointment person and the hit face image of the hit appointment person, and C is the weight of the torso similarity of the appointment person.

[0113] Then, it is determined whether the similarity of the appointment persons is greater than or equal to the appointment similarity threshold. When the similarity of the appointment persons is greater than or equal to the appointment similarity threshold, the appointment person to be confirmed can be defined as a pass person.

[0114] In this embodiment, the reservation person similarity of the reservation person is calculated by combining the reservation person's face similarity and the reservation person's torso similarity, and the reservation person similarity of the reservation person is compared with the reservation similarity threshold, thereby improving the judgment accuracy of the reservation person.

[0115] S592: If the similarity between the image of the person to be confirmed and the image of the person who has made an appointment is greater than or equal to the appointment similarity threshold, the person to be confirmed is defined as a passerby.

[0116] In this embodiment, by setting up appointment personnel, the appointment personnel can pass through the access control system within a specified time period, so as to avoid the phenomenon that the appointment personnel cannot pass through.

[0117] As shown in the figure, in one embodiment of the present application, the S591 further includes:

[0118] S593: If the similarity between the image of the person to be confirmed and the image of the person to be confirmed is less than the reservation similarity threshold, the image of the person to be confirmed is sent to the user terminal to inquire the user's opinion.

[0119] S594, receiving a user consent message fed back by the user terminal, and determining whether the user agrees to travel with the user who is about to make a reservation according to the user consent message; if the user agrees to travel with the user who is about to make a reservation, the person who is about to make a reservation is defined as a passerby.

[0120] S595: If the user does not agree to the prospective reservation user to travel with him / her, the person to be confirmed will be defined as a prohibited person.

[0121] In this embodiment, since there may be problems with the collected image of the person making the reservation, it is necessary to collect the user's opinion at this time. If the user allows the person making the reservation to pass, the person making the reservation is allowed to pass.

[0122] like Figure 2 As shown, in one embodiment of the present application, an intelligent access control system is provided, and the intelligent access control system includes an access control component, a collection component and a processing component.

[0123] The intelligent access control system and the user terminal interact to implement the intelligent access control method mentioned in the above embodiment. The user terminal can be a smart terminal worn by the user, including but not limited to a smart bracelet, a smart watch, and a mobile phone.

[0124] The access control component 100 is used to control the entry and exit of pedestrians. The acquisition component 200 is used to acquire user images and images of fellow travelers. The access control component 100 and the acquisition component 200 are both in communication connection with the processing component 300, and the processing component 300 executes the intelligent access control method as described in the above embodiment according to the user images and images of fellow travelers input by the acquisition component 200.

[0125] Specifically, in order to make the text concise, only the embodiments of the intelligent access control system are numbered.

[0126] The acquisition component 200 may be a camera.

[0127] The processing component 300 includes a storage unit, in which a complete torso image is stored. The complete torso image includes a face part image and a torso part image.

[0128] In this embodiment, before the user needs to pass through the access control component 100, the user's user image is first collected by the collection component 200, and the user image includes a face part and a torso part. The collected user image is then input into the processing component 300. The processing component 300 executes the intelligent access control method as described in the above embodiment according to the input user image, and determines whether the user can pass through the access control component 100.

[0129] The technical features of the above-described embodiments may be arbitrarily combined, and the execution order of the method steps is not limited. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0130] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. An intelligent access control method, characterized in that: The intelligent access control method comprises: Collecting a user image of the user, obtaining a partial face image and a partial torso image from the user image, comparing the partial face image with a pre-stored face image, and selecting a pre-stored face image having the highest face similarity to the partial face image; Obtaining a torso image corresponding to a pre-stored face image having the highest face similarity to the face image, and defining the torso image as a hit torso image; respectively calculating a torso index of a torso image of the user image and a torso index of a hit torso image; A torso similarity threshold is set, and based on the torso index of the torso image of the user image and the torso index and torso similarity threshold of the hit torso image, it is determined whether the torso similarity between the torso image of the user image and the hit torso image is greater than or equal to the similarity threshold; if the torso similarity between the torso image of the user image and the hit torso image is greater than or equal to the similarity threshold, the user is defined as a passerby.

2. The intelligent access control method according to claim 1, characterized in that: The collecting of the user image of the user, obtaining a partial face image and a partial torso image from the user image, comparing the partial face image with a pre-stored face image, and finding a corresponding person that is closest to the pre-stored face image, includes: Collect user images; The user image is divided into a face image and a torso image using the YOLO image segmentation algorithm; The facial features in the partial face image are extracted, and the pre-stored face image in the database with the highest similarity to the partial face image is matched according to the facial features in the partial face image.

3. The intelligent access control method according to claim 2, characterized in that: The step of obtaining a torso image corresponding to a pre-stored face image having the highest face similarity to the face image, and defining the torso image as a hit torso image, includes: Load the partial face image and project the partial face image into the principal component analysis subspace; Based on the feature value, a pre-stored face image with the most similar ratio is searched from the database, and the corresponding person of the pre-stored image is output.

4. The intelligent access control method according to claim 3, characterized in that: The step of respectively calculating the torso index of the torso image of the user image and the torso index of the hit torso image comprises: Acquire an image of the hit torso part; Obtaining the volume, maximum width and maximum height of the torso in the torso image of the user image in the entire image, and obtaining the volume, width and height of the torso in the torso image of the hit torso; The torso index of the torso image of the user image and the torso index of the hit torso image are calculated respectively by formula 1; Among them, Q is the torso index, V is the volume of the torso part in the entire image, h is the maximum height of the torso part, and w is the maximum width of the torso part.

5. The intelligent access control method according to claim 4, characterized in that: The torso similarity threshold is set to determine whether the torso similarity between the torso image of the user image and the torso image of the corresponding person is greater than or equal to the similarity threshold, and when the torso similarity between the torso image of the user image and the torso image of the corresponding person is greater than or equal to the similarity threshold, the user is allowed to pass, including: Set the torso similarity threshold; The torso similarity between the torso image of the user image and the torso image of the hit is calculated by formula 2; Wherein, Q1 is the torso index of the torso image of the user image, Q2 is the torso index of the hit torso image, and P is the torso similarity between the torso image of the user image and the hit torso image; Determine whether the torso similarity between the torso image of the user image and the hit torso image is greater than or equal to a torso similarity threshold; If the torso similarity between the torso image of the user image and the torso image of the hit is greater than or equal to the torso similarity threshold, the user is defined as a passerby; If the torso similarity between the torso image of the user image and the torso image of the hit is less than the torso similarity threshold, the user is defined as a prohibited person.

6. The intelligent access control method according to claim 5, characterized in that: After defining the user as a passer-by, the intelligent access control method further includes: Determine whether the passerby has any companions; If the passerby is accompanied by a companion, the image of the companion is collected; The images of the companions of the passerby are stored in a database, and the user image and the images of the companions are stored in correspondence with each other.

7. The intelligent access control method according to claim 6, characterized in that: Before storing the user image and the companion image correspondingly, the intelligent access control method further includes: Sending a query instruction to the user terminal to enter the image of the companion into the database, and determining whether a consent instruction to store is received from the user terminal; If the user receives a consent instruction to save returned by the user, a companion information entry window pops up; the companion information entry window is used to input the relationship data between the user and the companion; Relationship data between the user and the companions is received, and relationship tags are generated based on the relationship data between the user and the companions, and the relationship tags are assigned to the companion images.

8. The intelligent access control system according to claim 7, characterized in that: The intelligent access control method further includes: Acquire the reservation personnel information and reservation personnel image added by the user, and store the reservation personnel information and reservation personnel image added by the user in the reservation personnel information unit in the database correspondingly; Set appointment time slot; Collect the information of the person to be confirmed in the reservation time period, and determine whether the information of the person to be confirmed is the same as any information of the person in the reservation information unit; If the reservation person information to be confirmed is the same as any reservation person information in the reservation person information unit, then the reservation person image corresponding to the reservation person information is obtained and recorded as the hit reservation person image; Collect images of people whose appointments are to be confirmed; Set appointment similarity threshold; Compare the image of the person to be confirmed with the image of the person who has made the reservation, and determine whether the similarity between the image of the person to be confirmed and the image of the person who has made the reservation is greater than or equal to the reservation similarity threshold; If the similarity between the image of the person to be confirmed and the image of the person with the reservation is greater than or equal to the reservation similarity threshold, the person with the reservation to be confirmed is defined as a passerby.

9. The intelligent access control system according to claim 8, characterized in that: If the information of the person who is about to make an appointment is the same as the information of the person who has made an appointment, collecting the image of the person who is about to make an appointment also includes: If the similarity between the image of the person to be confirmed and the image of the person to be confirmed is less than the reservation similarity threshold, the image of the person to be confirmed is sent to the user terminal to ask the user for his opinion; Receive a user consent message fed back by the user terminal, and determine whether the user agrees to travel with the user who is about to make an appointment according to the user consent message. If the user agrees to travel with the user who is about to make an appointment, define the person who is about to make an appointment as a passerby; If the user does not agree to allow the prospective reservation user to travel with him / her, the person whose reservation is to be confirmed will be defined as a prohibited person.

10. An intelligent access control system, applied to the intelligent access control method according to claim 1 to claim 9, characterized in that: The intelligent access control system includes: Access control components; used to control the entry and exit of pedestrians; A collection component, which is used to collect images of the user and the companions; A processing component, wherein the access control component and the collection component are both communicatively connected to the processing component, and the processing component executes the intelligent access control method as described in any one of claims 1 to 9 based on the user image and the companion image input by the collection component.

Images