A multi-device integrated smart park access control face recognition device based on the Internet of Things

Abstract

The present invention discloses a multi-device fusion smart park access control face recognition device based on the Internet of Things, and relates to the field of face recognition technology. The system discloses a door access control face recognition module, a feature recognition and determination module, and a multi-device intervention module. The door access control face recognition module and the feature recognition and determination module are set, and a reasonable and efficient face feature recognition sequence can be customized for the park access control. When performing face recognition, it is no longer necessary to perform complex fusion calculations, and it is no longer necessary to wait for all features to be extracted before performing fusion processing. Each face feature is analyzed independently, and the recognition accuracy and efficiency of the face feature are guaranteed. The multi-device intervention module is set, and the face recognition device of the park access control can be periodically comprehensively analyzed, and the security equipment that needs to be intervened can be accurately determined according to the analysis results, so as to provide effective and reasonable security protection for the park.

Description

Technical Field

[0001] The present invention relates to the field of face recognition technology, and more specifically, to a multi-device fusion smart park access control face recognition device based on the Internet of Things. Background Art

[0002] Smart parks are paying more and more attention to security management. At present, smart parks will install facial recognition access control at key entrances and exits. However, the current facial recognition access control requires complex fusion calculations during facial recognition, resulting in low efficiency of facial recognition. At the same time, the corresponding security equipment in the park cannot be effectively associated and integrated with the facial recognition equipment. Therefore, how to intelligently associate and integrate facial recognition equipment with the security equipment in the park is an urgent problem to be solved. Summary of the invention

[0003] In view of the shortcomings of the prior art, the purpose of the present invention is to provide a multi-device fusion smart park access control face recognition device based on the Internet of Things.

[0004] To achieve the above object, the present invention provides the following technical solutions:

[0005] A multi-device integrated smart park access control face recognition device based on the Internet of Things, including an access control face recognition module, a feature recognition and determination module, and a multi-device intervention module;

[0006] The access control face recognition module is used to set the facial feature recognition order of the park. When the face recognition device recognizes the face of a person, a video of the face is shot, and the shot video is framed in real time. When a face frame is obtained by frame extraction, the facial features in the face frame are recognized and processed according to the facial feature recognition order.

[0007] The feature recognition determination module obtains the facial feature value each time a feature recognition record is generated, and determines whether the facial recognition is passed based on the comparison result between the facial feature value and the threshold value;

[0008] The multi-device intervention module obtains the security notification value based on a fixed period, determines the security device intervention level based on the comparison result between the security notification value and its threshold value, and intervenes in the corresponding security device according to the security device intervention level.

[0009] Furthermore, the facial feature recognition order of the park is set in the following way:

[0010] Get all feature recognition records before the current time of the system, get the facial feature types of feature recognition records, mark feature recognition records of the same facial feature types as same feature recognition records, sum up the feature recognition values ​​of all same feature recognition records and take the average to get the same feature recognition mean TSH, sort all same feature recognition records in the order of feature recognition time, compare the character numbers of two adjacent same feature recognition records after sorting, when the character numbers of two adjacent same feature recognition records are the same, increase the number of repeated recognitions by one, when the character numbers of two adjacent same feature recognition records are different, do not process, sum up all repeated recognitions to get the total number of repeated recognitions ASE, use the formula The recognition sequence value SZw of the facial feature type is obtained, wherein all facial feature types are sorted in descending order according to the numerical values ​​of the recognition sequence values, thereby generating the facial feature recognition sequence of the park.

[0011] Furthermore, when a facial feature in each pair of facial frames is identified, a feature recognition record of the facial feature is generated, and the feature recognition record is uploaded to the IoT storage end through the IoT. The feature recognition record includes the character number, facial frame number, facial feature type, feature recognition value, and feature recognition time.

[0012] Furthermore, the feature recognition value of the feature recognition record is obtained in the following manner: obtaining the facial feature type of the feature recognition record, performing feature extraction on the facial feature type of the facial frame to obtain facial feature data of the facial frame, obtaining a facial feature library of the facial feature type in the park in the system, matching the facial feature data of the facial frame with the facial feature library to generate a feature recognition value; when the facial feature data of the facial frame matches the facial feature library, the feature recognition value is 1; when the facial feature data of the facial frame does not match the facial feature library, the feature recognition value is -1.

[0013] Furthermore, each time a feature recognition record is generated, the facial feature value of the person is obtained, and the upper and lower limits of the facial features are set;

[0014] When the facial feature value of the person is greater than or equal to the upper limit value of the facial feature, the person passes the facial recognition;

[0015] When the facial feature value of the person is less than or equal to the lower limit of the facial feature, the person fails the facial recognition, and a recognition failure record is generated synchronously. The recognition failure record is uploaded to the IoT storage terminal through the IoT. The recognition failure record includes the person number, the number of facial frames, and the time of recognition failure.

[0016] When the facial feature value of the person is between the upper limit value and the lower limit value of the facial feature, continue to identify and process the subsequent facial features in the facial frame in the facial feature recognition order. When the recognition and processing of all facial features in a face frame is completed, the facial features of the next face frame are identified and processed in the facial feature recognition order.

[0017] Furthermore, the facial feature values ​​of the person are obtained in the following manner: obtain all feature recognition records of the person, obtain the feature recognition values ​​of the feature recognition records, sum the feature recognition values ​​of all feature recognition records to obtain the feature recognition sum value Bs, sort all feature recognition records in the order of feature recognition time, compare the feature recognition values ​​of two adjacent feature recognition records after sorting, when the feature recognition values ​​of two adjacent feature recognition records are the same, no processing is performed, when the feature recognition values ​​of two adjacent feature recognition records are different, the number of recognition fluctuations is increased once, all the number of recognition fluctuations is summed to obtain the total number of recognition fluctuations Pm, the face frame numbers of two adjacent feature recognition records after sorting are calculated to differ, and the face frame number variation value is obtained, and all the face frame number variation values ​​are summed to obtain the face frame number and variation value SL, and the formula is used. The facial feature value TZG of the person is obtained, where a1 is the feature recognition and value coefficient, a2 is the total number of recognition fluctuation coefficients, and a3 is the facial frame number and variable value coefficient.

[0018] Furthermore, based on the fixed cycle, at each periodic node of the fixed cycle, the security notification value is obtained, and the security notification threshold is set. When the security notification value is greater than or equal to the security notification threshold, the difference between the security notification value and the security notification threshold is calculated to obtain the security equipment intervention value Ki. The security equipment intervention value obtains the security equipment intervention level, and the corresponding security equipment is intervened according to the security equipment intervention level.

[0019] When the security notification value is less than the security notification threshold, no processing is performed.

[0020] Furthermore, the device notification value is obtained in the following manner: obtain all the recognition failure records generated in the previous cycle before the cycle node, mark the total number of recognition failure records as SEJ, sort all the recognition failure records in the order of recognition failure time, calculate the time difference between the recognition failure times of two adjacent recognition failure records after sorting, and obtain the recognition failure interval, sum up all the recognition failure intervals and take the average, and obtain the recognition failure average interval gB, calculate the difference between the number of face frames of two adjacent recognition failure records after sorting, and obtain the difference in the number of failed frames, set the threshold difference in the number of failed frames, when the difference in the number of failed frames is greater than or equal to the threshold difference in the number of failed frames, increase the number of failed frame fluctuations by one, when the difference in the number of failed frames is less than the threshold difference in the number of failed frames, do not process it, sum up all the number of failed frame fluctuations, and obtain the total number of failed frame fluctuations RST, using the formula The security notification value ST is obtained, wherein c1 is the coefficient of the total number of records that failed to be identified, c2 is the coefficient of the average interval of records that failed to be identified, and c3 is the coefficient of the total number of fluctuations of failed frames.

[0021] Compared with the prior art, the present invention has the following beneficial effects:

[0022] 1. Setting up the access control face recognition module and feature recognition and judgment module can customize a reasonable and efficient facial feature recognition sequence for the park access control. When performing face recognition, there is no need to perform complex fusion calculations, and there is no need to wait for all features to be extracted before fusion processing. Each facial feature is analyzed independently, and the recognition accuracy and efficiency of facial features are guaranteed;

[0023] 2. Setting up a multi-device intervention module can periodically conduct a comprehensive analysis of the facial recognition equipment of the park access control, accurately determine the security equipment that needs to be intervened based on the analysis results, and conduct effective and reasonable security management of the park. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 It is a flowchart of the present invention;

[0025] Figure 2 A flowchart for intervening in corresponding security equipment according to the level of intervention of security equipment;

[0026] Figure 3 It is a module block diagram of the present invention. DETAILED DESCRIPTION

[0027] Reference Figures 1 to 3 A multi-device fusion smart park access control face recognition device based on the Internet of Things, including an access control face recognition module, a feature recognition and judgment module, and a multi-device intervention module.

[0028] Access control face recognition module: set the facial feature recognition order of the park (face recognition needs to recognize various facial features of the face, including but not limited to bone structure features, mouth features, muscle distribution features, skin color features, forehead contour features, and the facial feature recognition order is to sort and recognize the facial features in sequence. This embodiment lists a facial feature recognition order: 1. Bone structure features; 2. Muscle distribution features; 3. Mouth features; 4. Forehead contour features; 5. Skin color features, and the facial features are recognized in sequence according to the above order).

[0029] The facial feature recognition order of the park is set in the following way: obtain all feature recognition records before the current time of the system, obtain the facial feature type of the feature recognition record, mark the feature recognition records of the same facial feature type as the same feature recognition record, sum the feature recognition values ​​of all the same feature recognition records and take the average to obtain the same feature recognition mean TSH, sort all the same feature recognition records in the order of feature recognition time, compare the character numbers of the two adjacent same feature recognition records after sorting, when the character numbers of the two adjacent same feature recognition records are the same, increase the number of repeated recognitions by one, when the character numbers of the two adjacent same feature recognition records are different, do not process them, sum all the repeated recognition times to obtain the total number of repeated recognitions ASE, and use the formula The recognition order value SZw of the facial feature type is obtained, where b1 is the total number of repeated recognition coefficients, and the value of b1 is 1.25. All facial feature types are sorted in descending order according to the numerical value of the recognition order value, and then the facial feature recognition order of the park is generated.

[0030] When the face recognition device recognizes the face of a person, a video of the face is shot, and frame extraction processing is performed on the shot video in real time (frame extraction of the shot video can obtain face frames). When a face frame is obtained by extracting each frame, the face features in the face frame are identified in the order of face feature recognition. When a face feature in each face frame is recognized, a feature recognition record of the face feature is generated, and the feature recognition record is uploaded to the storage end of the Internet of Things through the Internet of Things. The feature recognition record includes a person number (the face recognition device numbers each person each time it recognizes a person, and each person corresponds to an independent and unique number, that is, when there is no change in the person in the face recognition device, the person number remains unchanged, and the person number is only the number assigned to the face recognition, that is, when the same person performs face recognition once in the morning and once in the evening, the person numbers are different twice), a face frame number (each face frame corresponds to an independent and unique number, and the numbers of adjacent face frames are continuous, such as the face frame number is i, i=1, 2, ..., N, and N is a positive integer), a face feature type, a feature recognition value, and a feature recognition time.

[0031] The feature recognition value of the feature recognition record is obtained in the following manner: obtain the facial feature type of the feature recognition record, perform feature extraction on the facial frame of the facial feature type (if the facial feature type is a bone structure feature, extract the bone structure feature in the facial frame), obtain facial feature data of the facial frame, obtain the facial feature library of the facial feature type in the park in the system (the Internet of Things storage end in the system stores facial feature libraries of various facial feature types, including but not limited to a facial feature library of bone structure features and a facial feature library of mouth features. The facial feature library of bone structure features stores the bone structure feature data of everyone in the park, and the facial feature library of mouth features stores the mouth feature data of everyone in the park), match the facial feature data of the facial frame with the facial feature library, and generate a feature recognition value. When the facial feature data of the facial frame matches the facial feature library, the feature recognition value is 1. When the facial feature data of the facial frame does not match the facial feature library, the feature recognition value is -1.

[0032] Feature recognition and determination module: each time a feature recognition record is generated, the facial feature value of the person is obtained, and the upper limit value and the lower limit value of the facial feature are set, wherein the upper limit value of the facial feature is greater than the lower limit value of the facial feature, and the upper limit value and the lower limit value of the facial feature are both system-set thresholds, which can be modified according to actual needs;

[0033] When the facial feature value of the person is greater than or equal to the upper limit value of the facial feature, the person passes the facial recognition;

[0034] When the facial feature value of the person is less than or equal to the facial feature lower limit, the person fails to pass the facial recognition, and a recognition failure record is generated synchronously, and the recognition failure record is uploaded to the IoT storage terminal through the IoT. The recognition failure record includes the person number, the number of facial frames (i.e., the total number of facial frames for facial feature recognition of the person with the number. If two facial frames of the person with the number are recognized, the number of facial frames is 2), and the recognition failure time (i.e., the corresponding time of generating the recognition failure record);

[0035] When the facial feature value of the person is between the upper limit value and the lower limit value of the facial feature, continue to identify and process the subsequent facial features in the facial frame in the facial feature recognition order. When the recognition and processing of all facial features in a face frame is completed, the facial features of the next face frame are identified and processed in the facial feature recognition order.

[0036] The facial feature values ​​of a person are obtained in the following manner: obtain all feature recognition records of the person, obtain the feature recognition values ​​of the feature recognition records, sum the feature recognition values ​​of all feature recognition records to obtain the feature recognition sum value Bs, sort all feature recognition records in the order of feature recognition time, compare the feature recognition values ​​of two adjacent feature recognition records after sorting, when the feature recognition values ​​of two adjacent feature recognition records are the same, no processing is performed, when the feature recognition values ​​of two adjacent feature recognition records are different, the number of recognition fluctuations is increased once, all the number of recognition fluctuations is summed to obtain the total number of recognition fluctuations Pm, the face frame numbers of two adjacent feature recognition records after sorting are calculated by difference, and the face frame number variation value is obtained, all the face frame number variation values ​​are summed to obtain the face frame number and variation value SL, and the formula is used. The facial feature value TZG of the character is obtained, where a1 is the feature recognition and value coefficient, a2 is the total recognition fluctuation coefficient, a3 is the face frame number and variable value coefficient, the value of a1 is 1.89, the value of a2 is 1.05, and the value of a3 is 3.59.

[0037] By setting up the access control face recognition module and the feature recognition and judgment module, a reasonable and efficient face feature recognition sequence can be customized for the park access control. When performing face recognition, there is no need to perform complex fusion calculations, and there is no need to wait for all features to be extracted before fusion processing. Each facial feature can be analyzed independently to ensure the recognition accuracy and efficiency of the facial features.

[0038] Multi-device intervention module: Based on a fixed cycle, at each cycle node of the fixed cycle, the security notification value is obtained, and the security notification threshold is set (the security notification threshold is the system setting threshold, which can be modified according to actual needs). When the security notification value is greater than or equal to the security notification threshold, the difference between the security notification value and the security notification threshold is calculated to obtain the security device intervention value Ki. The security device intervention value obtains the security device intervention level, and intervenes in the corresponding security device according to the security device intervention level. The range of each security device intervention value corresponds to a security device intervention level. The range of the security device intervention value is [0, K1], (K1, K2], …, (Ki-1, Ki], security device intervention level 1, security device intervention level 2, …, security The level of device intervention is i-1, and the level of security device intervention is i. The higher the level of security device intervention is, the more security devices are involved in the park. For example, when the level of security device intervention is 1, the video surveillance device (the video surveillance device can continuously monitor the park access control) is involved. When the level of security device intervention is 2, the video surveillance device and the intrusion detection device (the intrusion detection device can detect the park boundary for human movement detection to prevent relevant personnel from entering the park from the park boundary without passing the park access control) are involved. When the level of security device intervention is 3, the video surveillance device, the intrusion detection device and the electronic patrol device (the electronic patrol device can arrange patrol personnel to patrol at the park access control location or other locations) are involved. The above is just an example of this embodiment.

[0039] When the security notification value is less than the security notification threshold, no processing is performed.

[0040] The device notification value is obtained in the following way: obtain all the recognition failure records generated in the previous cycle before the cycle node, mark the total number of recognition failure records as SEJ, sort all the recognition failure records in the order of recognition failure time, calculate the time difference between the recognition failure times of the two adjacent recognition failure records after sorting, and obtain the recognition failure interval, sum up all the recognition failure intervals and take the average, and obtain the recognition failure average interval gB, calculate the difference between the number of face frames of the two adjacent recognition failure records after sorting, and obtain the difference in the number of failed frames, set the threshold difference of the number of failed frames (the threshold difference of the number of failed frames is the threshold set by the system and can be modified according to actual needs), when the difference in the number of failed frames is greater than or equal to the threshold difference of the number of failed frames, increase the number of failed frame fluctuations by one, when the difference in the number of failed frames is less than the threshold difference of the number of failed frames, do not process it, sum up all the number of failed frame fluctuations, and obtain the total number of failed frame fluctuations RST, using the formula The security notification value ST is obtained, where c1 is the coefficient of the total number of records that were not identified, c2 is the coefficient of the average interval of records that were not identified, and c3 is the coefficient of the total number of frame fluctuations that were not identified. The value of c1 is 0.72, the value of c2 is 0.97, and the value of c3 is 0.82.

[0041] By setting up a multi-device intervention module, you can periodically conduct a comprehensive analysis of the facial recognition equipment of the park access control system, accurately determine the security equipment that needs to be intervened based on the analysis results, and provide effective and reasonable security protection for the park.

[0042] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0043] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website site, computer, server or data center to another website site, computer, server or data center by wired or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0044] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0045] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0046] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0047] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0048] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage media include: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM), random access memories (RAM), magnetic disks or optical disks.

[0049] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A multi-device fusion smart park access control face recognition device based on the Internet of Things, characterized by: Including access control face recognition module, feature recognition and determination module, and multi-device intervention module; The access control face recognition module is used to set the facial feature recognition order of the park. When the face recognition device recognizes the face of a person, a video of the face is shot, and the shot video is framed in real time. When a face frame is obtained by frame extraction, the facial features in the face frame are recognized and processed according to the facial feature recognition order. The feature recognition determination module obtains the facial feature value each time a feature recognition record is generated, and determines whether the facial recognition is passed based on the comparison result between the facial feature value and its threshold value; The multi-device intervention module obtains the security notification value based on a fixed period, determines the security device intervention level based on the comparison result between the security notification value and the threshold value, and intervenes in the corresponding security device according to the security device intervention level; The facial feature values ​​of a person are obtained in the following manner: obtain all feature recognition records of the person, obtain the feature recognition values ​​of the feature recognition records, sum the feature recognition values ​​of all feature recognition records to obtain the feature recognition sum value Bs, sort all feature recognition records in the order of feature recognition time, compare the feature recognition values ​​of two adjacent feature recognition records after sorting, when the feature recognition values ​​of two adjacent feature recognition records are the same, no processing is performed, when the feature recognition values ​​of two adjacent feature recognition records are different, the number of recognition fluctuations is increased once, all the number of recognition fluctuations is summed to obtain the total number of recognition fluctuations Pm, the face frame numbers of two adjacent feature recognition records after sorting are calculated by difference, and the face frame number variation value is obtained, all the face frame number variation values ​​are summed to obtain the face frame number and variation value SL, and the formula is used. Get the facial feature value TZG of the person, where a1 is the feature recognition and value coefficient, a2 is the total recognition fluctuation coefficient, and a3 is the face frame number and variable value coefficient; The security notification value is obtained in the following way: obtain all the recognition failure records generated in the previous cycle before the cycle node, mark the total number of recognition failure records as SEJ, sort all the recognition failure records in the order of recognition failure time, calculate the time difference between the recognition failure times of the two adjacent recognition failure records after sorting, and obtain the recognition failure interval, sum up all the recognition failure intervals and take the average, and obtain the recognition failure average interval gB, calculate the difference between the number of face frames of the two adjacent recognition failure records after sorting, and obtain the difference in the number of failed frames, set the threshold difference in the number of failed frames, when the difference in the number of failed frames is greater than or equal to the threshold difference in the number of failed frames, increase the number of failed frame fluctuations by one, when the difference in the number of failed frames is less than the threshold difference in the number of failed frames, do not process it, sum up all the number of failed frame fluctuations, and obtain the total number of failed frame fluctuations RST, using the formula The security notification value ST is obtained, wherein c1 is the coefficient of the total number of records that failed to be identified, c2 is the coefficient of the average interval of records that failed to be identified, and c3 is the coefficient of the total number of fluctuations of failed frames.

2. According to the multi-device fusion smart park access control face recognition device based on the Internet of Things in claim 1, it is characterized in that: The facial feature recognition order of the park is set in the following way: Get all feature recognition records before the current time of the system, get the facial feature types of feature recognition records, mark feature recognition records of the same facial feature types as same feature recognition records, sum up the feature recognition values ​​of all same feature recognition records and take the average to get the same feature recognition mean TSH, sort all same feature recognition records in the order of feature recognition time, compare the character numbers of two adjacent same feature recognition records after sorting, when the character numbers of two adjacent same feature recognition records are the same, increase the number of repeated recognitions by one, when the character numbers of two adjacent same feature recognition records are different, do not process, sum up all repeated recognitions to get the total number of repeated recognitions ASE, use the formula The recognition sequence value SZw of the facial feature type is obtained, wherein all facial feature types are sorted in descending order according to the numerical values ​​of the recognition sequence values, thereby generating the facial feature recognition sequence of the park.

3. According to the multi-device fusion smart park access control face recognition device based on the Internet of Things in claim 1, it is characterized in that: Each time a facial feature in a facial frame is recognized, a feature recognition record of the facial feature is generated and uploaded to the IoT storage terminal through the IoT. The feature recognition record includes the person number, facial frame number, facial feature type, feature recognition value, and feature recognition time.

4. According to claim 3, a multi-device fusion smart park access control face recognition device based on the Internet of Things is characterized in that: The feature recognition value of the feature recognition record is obtained in the following manner: obtain the facial feature type of the feature recognition record, perform feature extraction on the facial feature type of the facial frame, obtain facial feature data of the facial frame, obtain the facial feature library of the facial feature type in the park in the system, match the facial feature data of the facial frame with the facial feature library, and generate a feature recognition value. When the facial feature data of the facial frame matches the facial feature library, the feature recognition value is 1. When the facial feature data of the facial frame does not match the facial feature library, the feature recognition value is -1.

5. According to the multi-device fusion smart park access control face recognition device based on the Internet of Things in claim 1, it is characterized in that: Each time a feature recognition record is generated, the facial feature value of the person is obtained, and the upper and lower limits of the facial features are set; When the facial feature value of the person is greater than or equal to the upper limit value of the facial feature, the person passes the facial recognition; When the facial feature value of the person is less than or equal to the lower limit of the facial feature, the person fails the facial recognition, and a recognition failure record is generated synchronously. The recognition failure record is uploaded to the IoT storage terminal through the IoT. The recognition failure record includes the person number, the number of facial frames, and the time of recognition failure. When the facial feature value of the person is between the upper limit value and the lower limit value of the facial feature, continue to identify and process the subsequent facial features in the facial frame in the facial feature recognition order. When the recognition and processing of all facial features in a face frame is completed, the facial features of the next face frame are identified and processed in the facial feature recognition order.

6. According to the multi-device fusion smart park access control face recognition device based on the Internet of Things in claim 1, it is characterized in that: Based on a fixed cycle, at each periodic node of the fixed cycle, the security notification value is obtained, and the security notification threshold is set. When the security notification value is greater than or equal to the security notification threshold, the difference between the security notification value and the security notification threshold is calculated to obtain the security equipment intervention value Ki. The security equipment intervention value obtains the security equipment intervention level, and the corresponding security equipment is intervened according to the security equipment intervention level; When the security notification value is less than the security notification threshold, no processing is performed.

Images