Face recognition method and device based on security system and storage medium

Abstract

The application discloses a face recognition method and device based on a security system and a storage medium, when a person approaches a security area, the security system automatically starts a face recognition function, collects face information data, uploads the collected face information data to the cloud after preprocessing, performs face recognition based on an intelligent face database, and outputs a recognition result; when the recognition result is 1, the face identity authentication is passed, and the person can enter the security area; when the recognition result is 0, the recognition result is displayed on a terminal and a corresponding warning is sent. The face recognition method of the security system has the characteristics of artificial intelligence, that is, no additional personnel is arranged for operation, the security system automatically judges and handles the conditions according to the set recognition result; the face recognition method of the security system is accurate and reliable by using corresponding algorithms, and can correctly distinguish between the personnel who can enter the security area and strangers.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of face recognition, and particularly relates to a face recognition method and device based on a security system and a storage medium. BACKGROUND

[0002] In today's society, people have increasingly high requirements for security monitoring systems in some actual scenarios, such as the deck on a ship in industrial production, an indoor processing workshop of a factory, a residential community, and the like. At present, it is no longer satisfied to monitor indoor and outdoor scenes by manpower. This traditional security video monitoring method has high cost, needs to employ and assign designated personnel to supervise video pictures all day long, consumes a large amount of manpower and material resources, and the acuteness of cognitive perception ability of the human eye also decreases after a long time of staring at the screen pictures of electronic equipment, which is very easy to cause misjudgment and missed judgment. This situation is extremely dangerous in some scenes involving high confidentiality and high risk, which not only may cause huge property losses, but also may threaten the safety of people's lives, which is contrary to the original intention of the existence of video monitoring. Moreover, the traditional video monitoring system mostly only realizes acquisition of real-time video and saves the video to a local file. After some unexpected accidents occur, historical archive video is retrieved for verification, and the efficiency of such a monitoring system is very low. Most of the saved videos are useless information, which only consumes more local storage space, and the degree of automation of the system is also low. The face recognition technology is also limited and cannot timely and accurately judge the current personnel. In addition, in the era of rapid development of mainstream technologies such as deep learning, the current data volume is much larger than before, and even frequent replacement of personnel in the monitoring video picture cannot achieve a relatively ideal security monitoring effect.

[0003] Therefore, it is very meaningful to develop a face recognition technology of an intelligent security system, which guarantees the user's use autonomy through face recognition, and can timely perceive and judge the recognition result, thereby increasing the security privacy and reliability. SUMMARY

[0004] The application provides a face recognition method and device based on a security system and a storage medium, to solve the problem that people have increasingly high requirements for security monitoring systems in some actual scenarios in the current society, such as the deck on a ship in industrial production, an indoor processing workshop of a factory, a residential community, and the like. At present, people not only satisfy the monitoring of indoor and outdoor scenes by manpower, but also supervise video pictures all day long and uninterruptedly by employing and assigning designated personnel, which is costly, consumes a large amount of manpower and material resources, and causes misjudgment and missed judgment due to the decline in the acuteness of cognitive perception of the human eye after a long time of staring at the screen of an electronic device. This is extremely dangerous in some scenarios involving high confidentiality and high risk, which not only causes great property loss, but also threatens the safety of people's lives, which is contrary to the original intention of video monitoring. Moreover, most of the traditional video monitoring systems only realize the acquisition of real-time video and save the video to a local file, and after some unexpected accidents occur, historical archive video is retrieved to verify, which is very inefficient. Most of the saved video is useless information, which only consumes more local storage space, and the system has low automation and limited face recognition technology, which cannot accurately judge the current personnel. In addition, in the era of rapid development of mainstream technologies such as deep learning, the current data volume is much larger than before, and even frequent replacement of personnel in the monitoring video cannot achieve the ideal effect of security monitoring.

[0005] To achieve the above purpose, the application provides the following technical solutions.

[0006] The face recognition method based on the security system comprises the following steps.

[0007] S101: When a person approaches a security area, the security system automatically starts a face recognition function to collect face information data.

[0008] S102: The collected face information data is preprocessed and uploaded to the cloud, face recognition is performed based on an intelligent face database, and a recognition result is output.

[0009] S103: When the recognition result is 1, the face identity authentication is passed, and the person can enter the security area; when the recognition result is 0, the recognition result is displayed on a terminal and a corresponding warning is issued.

[0010] The S101 step comprises the following steps.

[0011] S1011: A human infrared sensor is installed at the entrance of the security area. When a person approaches the security area, the human infrared sensor detects the approach of the person, and the security system automatically starts the camera image acquisition function.

[0012] S1012: The original information data of the current person is acquired through the camera image acquisition function.

[0013] S1013: The face region in the original information data is detected and intercepted, and the useless background part in the original information data is removed to obtain the face information data.

[0014] The S102 step includes:

[0015] S1021: The collected face information data is subjected to a normalization and grayscale preprocessing operation, and the preprocessed face information data is uploaded to the cloud.

[0016] S1022: The face information data uploaded to the cloud is subjected to dimension reduction processing through a specified feature extraction method to obtain the to-be-identified face information data.

[0017] S1023: Face image recognition training is performed based on an intelligent face database to obtain a corresponding intelligent classification recognition model, the to-be-identified face information data is subjected to classification recognition based on the intelligent classification recognition model, and an identification result is output.

[0018] The S103 step includes:

[0019] S1031: The security worker checks the face image information data, analyzes the data, and identifies the result through the terminal. The identification result includes 1 or 0.

[0020] S1032: If the identification result is 1, the face identity authentication is passed, the person can enter the security area, and the information of the person entering the security area is recorded in the data record section of the cloud.

[0021] S1033: If the identification result is 0, the face identity authentication fails, the terminal issues a corresponding warning reminder, and the security worker re-identifies and confirms the current person.

[0022] The S1022 step includes: in the dimension reduction processing of the face information data uploaded to the cloud, the preprocessed face information data is first subjected to feature space dimension reduction and face recognition feature optimization processing, the original high-dimensional feature set is compressed, the dimension of the face image in the feature subspace is calculated, the discriminative features in the feature subspace are extracted, and the feature dimension is further reduced.

[0023] The S1023 step of obtaining a corresponding intelligent classification and recognition model based on the intelligent face database includes: collecting face information of people who can enter the security area and face information of several strangers, performing normalization and grayscale preprocessing operations on the collected face information, storing the preprocessed face information data in the intelligent face database, extracting double-layer MB-LBP features from the preprocessed face information data, counting MB-LBP histogram features, calculating Euclidean distance based on the histogram features, obtaining the best discrimination threshold value through the Euclidean distance, obtaining the parameters of the OC-SVM model based on the histogram features, and obtaining a first recognition classifier for identifying stranger faces based on the parameters of the OC-SVM model and the best discrimination threshold value.

[0024] Based on the training process of the first recognition classifier, the second type of classifier for identifying the identity information of the person entering the security area is obtained based on the face information of the person who can enter the security area and the personal identity information.

[0025] The S1023 step of obtaining a corresponding intelligent classification and recognition model based on the intelligent face database includes: collecting face information of people who can enter the security area and face information of several strangers, performing normalization and grayscale preprocessing operations on the collected face information, storing the preprocessed face information data in the intelligent face database, extracting double-layer MB-LBP features from the preprocessed face information data, counting MB-LBP histogram features, calculating Euclidean distance based on the histogram features, obtaining the best discrimination threshold value through the Euclidean distance, obtaining the parameters of the OC-SVM model based on the histogram features, and obtaining a first recognition classifier for identifying stranger faces based on the parameters of the OC-SVM model and the best discrimination threshold value.

[0026] The S1023 step of obtaining a corresponding intelligent classification and recognition model based on the intelligent face database includes: collecting face information of people who can enter the security area and face information of several strangers, performing normalization and grayscale preprocessing operations on the collected face information, storing the preprocessed face information data in the intelligent face database, extracting double-layer MB-LBP features from the preprocessed face information data, counting MB-LBP histogram features, calculating Euclidean distance based on the histogram features, obtaining the best discrimination threshold value through the Euclidean distance, obtaining the parameters of the OC-SVM model based on the histogram features, and obtaining a first recognition classifier for identifying stranger faces based on the parameters of the OC-SVM model and the best discrimination threshold value.

[0027] The S1023 step of obtaining a corresponding intelligent classification and recognition model based on the intelligent face database includes: collecting face information of people who can enter the security area and face information of several strangers, performing normalization and grayscale preprocessing operations on the collected face information, storing the preprocessed face information data in the intelligent face database, extracting double-layer MB-LBP features from the preprocessed face information data, counting MB-LBP histogram features, calculating Euclidean distance based on the histogram features, obtaining the best discrimination threshold value through the Euclidean distance, obtaining the parameters of the OC-SVM model based on the histogram features, and obtaining a first recognition classifier for identifying stranger faces based on the parameters of the OC-SVM model and the best discrimination threshold value.

[0028] The face recognition device based on the security system includes:

[0029] The face recognition starting device is used for automatically starting the face recognition function of the security system when a person approaches the security area, and collecting face information data;

[0030] The face information data recognition device is used for uploading the collected face information data to the cloud after preprocessing, performing face recognition based on the intelligent face database, and outputting the recognition result.

[0031] The recognition result output device is used for when the recognition result is 1, the face identity authentication is passed, and the person can enter the security area; when the recognition result is 0, the recognition result is displayed on the terminal and the corresponding warning is issued.

[0032] The computer readable storage medium stores a computer program, and the computer program is executed by the processor to realize the method of any one of claims 1-8.

[0033] Compared with the prior art, the present application has the following advantages:

[0034] When a person approaches the security area, the security system automatically starts the face recognition function, collects face information data, pre-processes the collected face information data, uploads it to the cloud, performs face recognition based on the intelligent face database, and outputs the recognition result. When the recognition result is 1, the face identity authentication is passed, and the person can enter the security area; when the recognition result is 0, the recognition result is displayed on the terminal and the corresponding warning is issued. The face recognition method of the security system has the characteristics of artificial intelligence, i.e. without arranging additional personnel for operation, the security system automatically judges and handles the situation according to the set recognition result; the face recognition method of the security system uses corresponding algorithms that are accurate and reliable, and can correctly distinguish between personnel who can enter the security area and strangers; in addition, when face recognition authentication cannot be effectively performed due to various reasons, the security system should provide other authentication methods to enable the system to continue to operate.

[0035] Other features and advantages of the present application will be set forth in the following description, and in part will become apparent to those skilled in the art from the description, or can be learned by practice of the present application. The objects and other advantages of the present application can be achieved and obtained by the structure particularly pointed out in the written description, claims, and drawings.

[0036] The technical solutions of the present application will be further described in detail below with the help of the drawings and examples. BRIEF DESCRIPTION OF DRAWINGS

[0037] The accompanying drawings are used to provide a further understanding of the present application, and constitute a part of the specification, together with the embodiments of the present application, to explain the present application, and do not constitute a limitation on the present application. In the drawings:

[0038] Figure 1 The flow chart of the face recognition method based on the security system in the embodiment of the present application is shown in the following figure.

[0039] Figure 2 The flow chart of the face recognition method based on the security system in the embodiment of the present application is shown in the following figure.

[0040] Figure 3 The flow chart of the face recognition method based on the security system in the embodiment of the present application is shown in the following figure. DETAILED DESCRIPTION

[0041] The preferred embodiments of the present application are described below in conjunction with the accompanying drawings, and it should be understood that the preferred embodiments described herein are only used to illustrate and explain the present application, and are not used to limit the present application.

[0042] The embodiment of the present application provides a face recognition method based on a security system, comprising:

[0043] S101: When a person approaches a security area, the security system automatically starts the face recognition function, and collects face information data;

[0044] S102: After the collected face information data is preprocessed, it is uploaded to the cloud, and face recognition is performed based on an intelligent face database, and the recognition result is output;

[0045] S103: When the recognition result is 1, the face identity authentication is passed, and the person can enter the security area; when the recognition result is 0, the recognition result is displayed on the terminal and a corresponding warning is issued.

[0046] The working principle of the technical solution is as follows: a human body infrared sensor is installed at the boundary of the security area, and the human body infrared function is turned on. When the human body infrared sensor detects a person, the security system automatically turns on the face recognition function (face recognition is performed by turning on the image acquisition function of the camera), and the collected face information data (i.e. face image data collected by the camera) is transmitted to the cloud after preprocessing. The cloud centrally processes and analyzes the collected face information data to complete face detection, recognition and other tasks (face recognition based on an intelligent face database). After the face identity authentication is passed (when the recognition result is 1, the identity authentication is passed, and when the recognition result is 0, the face identity authentication is not passed), the control module of the security system opens the door lock, and the person can enter the security area. The recognition result is uploaded to the cloud (the cloud is used to store face information data and analyze, recognize and detect face information data), and the historical record of the person entering the security area is saved by the data record block of the cloud, which can be checked by the security worker through the terminal at any time. When a stranger is recognized, the security system will prevent the stranger from entering the security area, and will timely notify the security worker through the terminal. In special cases, the security worker can verify the identity of the stranger through password authentication to open the door lock at the boundary of the security area, which is a backup identity authentication scheme of the security system.

[0047] The beneficial effects of the above technical solution are as follows: when a person approaches the security area, the security system automatically turns on the face recognition function and collects face information data; the collected face information data is preprocessed and uploaded to the cloud, face recognition is performed based on an intelligent face database, and a recognition result is output; when the recognition result is 1, the face identity authentication is passed, and the person can enter the security area; when the recognition result is 0, the recognition result is displayed on the terminal and a corresponding warning is issued. The face recognition method of the security system has the characteristics of artificial intelligence, i.e. without arranging additional personnel for operation, the security system automatically judges and handles the situation according to the set recognition result; the face recognition method of the security system uses corresponding algorithms that are accurate and reliable, and can correctly distinguish between personnel who can enter the security area and strangers; in addition, when face recognition authentication cannot be effectively performed due to various reasons, the security system should provide other authentication methods to enable the system to continue to operate.

[0048] In another embodiment, the S101 step includes:

[0049] S1011: A human body infrared sensor is installed at the entrance of the security area, and when a person approaches the security area, the human body infrared sensor detects the person approaching, and the security system automatically turns on the camera image acquisition function;

[0050] S1012: The original information data of the current person is collected through the camera image acquisition function;

[0051] S1013: The face region in the original information data is detected and intercepted, the useless background part in the original information data is removed, and face information data is obtained.

[0052] The working principle of the above technical solution is that a human body infrared sensor is installed at the entrance of the security area, when a person approaches the security area, the human body infrared sensor detects the approach of the person, and the security system automatically starts the camera image acquisition function; the original information data of the current person is collected through the camera image acquisition function; the face region in the original information data is detected and intercepted, the useless background part in the original information data is removed, and face information data is obtained. The face information data is collected to prepare for subsequent face recognition.

[0053] The beneficial effects of the above technical solution are that a human body infrared sensor is installed at the entrance of the security area, when a person approaches the security area, the human body infrared sensor detects the approach of the person, and the security system automatically starts the camera image acquisition function; the original information data of the current person is collected through the camera image acquisition function; the face region in the original information data is detected and intercepted, the useless background part in the original information data is removed, and face information data is obtained. The face information data is collected to prepare for subsequent face recognition.

[0054] In another embodiment, the S102 step includes:

[0055] S1021: The collected face information data is subjected to a normalization and grayscale preprocessing operation, and the preprocessed face information data is uploaded to the cloud;

[0056] S1022: The face information data uploaded to the cloud is subjected to dimension reduction processing through a specified feature extraction method, and the face information data to be recognized is obtained;

[0057] S1023: Face image recognition training is performed based on an intelligent face database, a corresponding intelligent classification recognition model is obtained, the face information data to be recognized is classified and recognized based on the intelligent classification recognition model, and a recognition result is output.

[0058] The working principle of the above technical solution is that the collected face information data is subjected to a normalization and grayscale preprocessing operation, and the preprocessed face information data is uploaded to the cloud; the face information data uploaded to the cloud is subjected to dimension reduction processing through a specified feature extraction method, and the face information data to be recognized is obtained; face image recognition training is performed based on an intelligent face database, a corresponding intelligent classification recognition model is obtained, the face information data to be recognized is classified and recognized based on the intelligent classification recognition model, and a recognition result is output.

[0059] The technical scheme has the beneficial effects that: the face information data collected is subjected to a normalization and grayscale preprocessing operation, and the preprocessed face information data is uploaded to the cloud; the face information data uploaded to the cloud is subjected to dimension reduction processing through a specified feature extraction manner, and the face information data to be recognized is obtained; face image recognition training is performed based on an intelligent face database, a corresponding intelligent classification recognition model is obtained, the face information data to be recognized is subjected to classification recognition based on the intelligent classification recognition model, and a recognition result is output.

[0060] In another embodiment, the S103 step includes:

[0061] S1031: The security worker views the face image information data, analysis data, and recognition result through the terminal, and the recognition result includes 1 or 0.

[0062] S1032: If the recognition result is 1, the face identity authentication is passed, the personnel can enter the security area, and the personnel information of the personnel entering the security area is recorded in the data record block of the cloud.

[0063] S1033: If the recognition result is 0, the face identity authentication fails, the terminal issues a corresponding warning reminder, and the security worker re-identifies and confirms the current personnel.

[0064] The working principle of the technical scheme is that: the security worker views the face image information data, analysis data, and recognition result through the terminal, and the recognition result includes 1 or 0; if the recognition result is 1, the face identity authentication is passed, the personnel can enter the security area, and the personnel information of the personnel entering the security area is recorded in the data record block of the cloud; if the recognition result is 0, the face identity authentication fails, the terminal issues a corresponding warning reminder, and the security worker re-identifies and confirms the current personnel; if the current personnel belongs to the personnel who can enter the security area but have not logged in the system in time, the current personnel is registered and the face information is registered, and the face information of the personnel is stored in the intelligent face database.

[0065] The technical scheme has the beneficial effects that: the security worker views the face image information data, analysis data, and recognition result through the terminal, and the recognition result includes 1 or 0; if the recognition result is 1, the face identity authentication is passed, the personnel can enter the security area, and the personnel information of the personnel entering the security area is recorded in the data record block of the cloud; if the recognition result is 0, the face identity authentication fails, the terminal issues a corresponding warning reminder, and the security worker re-identifies and confirms the current personnel. The face recognition method of the security system has the characteristics of artificial intelligence, that is, without arranging additional personnel for operation, the security system judges and processes the situation appearing by itself according to the set recognition result; the face recognition method of the security system adopts a corresponding algorithm which is accurate and reliable, and can correctly distinguish between the personnel who can enter the security area and the strangers.

[0066] In another embodiment, the S1022 step includes: in the dimensionality reduction processing of the face information data uploaded to the cloud, first performing feature space dimension reduction and face recognition feature optimization processing on the preprocessed face information data, compressing the original high-dimensional feature set, calculating the dimension of the face image in the feature subspace, extracting discriminative features on the feature subspace, and further reducing the feature dimension.

[0067] The working principle of the above technical solution is: although the higher the pixel of the geometric image obtained by wavelet transform is, the more the corresponding frequency domain information is, but the 512x512 geometric image corresponding to a three-dimensional face image has a wavelet feature magnitude of 10 6 Left and right, the calculation complexity is too high, which is not conducive to practical application. Therefore, dimensionality reduction processing is needed. In the dimensionality reduction processing of the face information data uploaded to the cloud, first perform feature space dimension reduction and face recognition feature optimization processing on the preprocessed face information data, compress the original high-dimensional feature set, calculate the dimension of the face image in the feature subspace, extract discriminative features on the feature subspace, and further reduce the feature dimension.

[0068] The beneficial effects of the above technical solution are: in the dimensionality reduction processing of the face information data uploaded to the cloud, first perform feature space dimension reduction and face recognition feature optimization processing on the preprocessed face information data, compress the original high-dimensional feature set, calculate the dimension of the face image in the feature subspace, extract discriminative features on the feature subspace, and further reduce the feature dimension. Thus, the calculation efficiency is improved.

[0069] In another embodiment, the S1023 step of obtaining a corresponding intelligent classification recognition model based on intelligent face database face image recognition training includes: after collecting face information of people who can enter the security area and face information of several strangers, performing normalization and grayscale preprocessing operations, storing the preprocessed face information data in the intelligent face database, extracting double-layer MB-LBP features from the preprocessed face information data, and counting the histogram features of the MB-LBP, then calculating the Euclidean distance using the histogram features, obtaining the best discrimination threshold through the Euclidean distance, and also obtaining the parameters of the OC-SVM model using the histogram features, obtaining a first recognition classifier dedicated to recognizing stranger faces based on the parameters of the OC-SVM model and the best discrimination threshold.

[0070] Based on the training process of the first recognition classifier, the second type of classifier that can be used to identify the identity information of the person entering the security area is obtained through the face information of the person who can enter the security area and the personal identity information.

[0071] The working principle of the above technical solution is that theoretically, the number of unknown faces to be identified is much larger than the number of faces of personnel who can enter the security area, and the security system is prone to mistakenly identifying strangers as personnel who can enter the security area. Therefore, in order to further improve the reliability of the system in identifying unknown faces, a method of fusing distance measurement based on the best threshold and OC-SVM two-class algorithm is used for unknown face recognition. Only when the face to be identified passes through both types of recognition algorithms will it be determined as personnel who can enter the security area, otherwise it will be determined as an unknown face.

[0072] In the distance measurement based on the best threshold, the best distinguishing threshold of two types of faces is calculated using feature parameters, and the calculation formula is:

[0073]

[0074] Where θ represents the best distinguishing threshold, μ1 represents the mean of the Euclidean distance of the face image statistics of personnel who can enter the security area, σ1 represents the mean square error value of the Euclidean distance of the face image statistics of personnel who can enter the security area, μ2 represents the mean of the Euclidean distance of the face image statistics of strangers, and σ2 represents the mean square error value of the Euclidean distance of the face image statistics of strangers.

[0075] The beneficial effects of the above technical solution are that after collecting the face information of personnel who can enter the security area and the face information of several strangers, the pre-processing operation of normalization and grayscale is performed, the pre-processed face information data is stored in the intelligent face database, the double-layer MB-LBP feature of the pre-processed face information data is extracted, the histogram feature of the MB-LBP is counted, the Euclidean distance is calculated using the histogram feature, the best distinguishing threshold is obtained through the Euclidean distance, the parameters of the OC-SVM model are also obtained using the histogram feature, and the first recognition classifier dedicated to identifying unknown faces is obtained based on the parameters of the OC-SVM model and the best distinguishing threshold; based on the training process of the first recognition classifier, the second type of classifier that can be used to identify the identity information of personnel entering the security area is obtained through the face information and personal identity information of personnel who can enter the security area. The reliability of the security system in identifying unknown faces is improved through the first recognition classifier and the second type of classifier.

[0076] In another embodiment, the step S1023 of classifying and identifying the to-be-identified face information data based on the intelligent classification and identification model comprises: performing a normalization and grayscale preprocessing operation on the to-be-identified face image information, performing double-layer MB-LBP feature extraction on the face image after the preprocessing, and counting the histogram data of the double-layer MB-LBP, and outputting the histogram data to the first identification classifier for identification, if the identification result of the first identification classifier is a stranger, directly outputting the result 0, stopping the next step, and determining that the identification result is a face identity authentication failure, and storing the identity of the to-be-identified face in the intelligent face database.

[0077] The working principle of the above technical solution is: performing a normalization and grayscale preprocessing operation on the to-be-identified face image information, performing double-layer MB-LBP feature extraction on the face image after the preprocessing, and counting the histogram data of the double-layer MB-LBP, and outputting the histogram data to the first identification classifier for identification, if the identification result of the first identification classifier is a stranger, directly outputting the result 0, stopping the next step, and determining that the identification result is a face identity authentication failure, and storing the identity of the to-be-identified face in the intelligent face database; if the identification result is a person who can enter the security area, outputting the result 1 and identifying the identity information of the person who can enter the security area through the second classifier, and displaying the identity information on the terminal. The first identification classifier and the second classifier improve the reliability of the security system for stranger face recognition.

[0078] The above technical solution has the beneficial effects that: performing a normalization and grayscale preprocessing operation on the to-be-identified face image information, performing double-layer MB-LBP feature extraction on the face image after the preprocessing, and counting the histogram data of the double-layer MB-LBP, and outputting the histogram data to the first identification classifier for identification, if the identification result of the first identification classifier is a stranger, directly outputting the result 0, stopping the next step, and determining that the identification result is a face identity authentication failure, and storing the identity of the to-be-identified face in the intelligent face database; if the identification result is a person who can enter the security area, outputting the result 1 and identifying the identity information of the person who can enter the security area through the second classifier, and displaying the identity information on the terminal. The first identification classifier and the second classifier improve the reliability of the security system for stranger face recognition.

[0079] In another embodiment, in the process of face feature point information detection and matching, the Euclidean distance is used for discrimination, the calculated Euclidean distance is compared with a preset Euclidean distance threshold, and the identity of the person is identified.

[0080] The recognition rule of the personnel identity includes: if the Euclidean distance between the feature information of a face in the face database and the actual feature information is less than a set value, it is indicated that the detected face is in the intelligent face database; if the Euclidean distance between each face feature vector in the intelligent face database and the detected face feature vector is greater than a set Euclidean distance value, it is indicated that the detected face is not in the intelligent face database, and the personnel identity is not recognized successfully.

[0081] The working principle of the technical solution is as follows: in the process of face feature point information detection and matching, the Euclidean distance is used for discrimination, the calculated Euclidean distance is compared with a preset Euclidean distance threshold, and the personnel identity is recognized; the recognition rule of the personnel identity includes: if the Euclidean distance between the feature information of a face in the face database and the actual feature information is less than a set value, it is indicated that the detected face is in the intelligent face database; if the Euclidean distance between each face feature vector in the intelligent face database and the detected face feature vector is greater than a set Euclidean distance value, it is indicated that the detected face is not in the intelligent face database, and the personnel identity is not recognized successfully. The face recognition method of the security system has the characteristics of artificial intelligence, that is, no additional personnel is arranged for operation, the security system judges and handles the situation according to the set recognition result; the face recognition method of the security system uses a corresponding algorithm which is accurate and reliable, and can correctly distinguish between personnel who can enter the security area and strangers; in addition, when face recognition authentication cannot be effectively performed due to various reasons, the security system should provide other authentication methods to enable the system to continue to operate.

[0082] The working principle of the technical solution is as follows: in the process of face feature point information detection and matching, the Euclidean distance is used for discrimination, the calculated Euclidean distance is compared with a preset Euclidean distance threshold, and the personnel identity is recognized; the recognition rule of the personnel identity includes: if the Euclidean distance between the feature information of a face in the face database and the actual feature information is less than a set value, it is indicated that the detected face is in the intelligent face database; if the Euclidean distance between each face feature vector in the intelligent face database and the detected face feature vector is greater than a set Euclidean distance value, it is indicated that the detected face is not in the intelligent face database, and the personnel identity is not recognized successfully. The face recognition method of the security system has the characteristics of artificial intelligence, that is, no additional personnel is arranged for operation, the security system judges and handles the situation according to the set recognition result; the face recognition method of the security system uses a corresponding algorithm which is accurate and reliable, and can correctly distinguish between personnel who can enter the security area and strangers; in addition, when face recognition authentication cannot be effectively performed due to various reasons, the security system should provide other authentication methods to enable the system to continue to operate.

[0083] In another embodiment, the face recognition device based on the security system comprises:

[0084] The face recognition starting device is used to automatically start the face recognition function of the security system when a person approaches the security area, and collect face information data;

[0085] The face information data recognition device is used to upload the collected face information data to the cloud after preprocessing, perform face recognition based on the intelligent face database, and output the recognition result;

[0086] The recognition result output device is used to pass the face identity authentication when the recognition result is 1, and the person can enter the security area; when the recognition result is 0, the recognition result is displayed on the terminal and the corresponding warning is issued.

[0087] The working principle of the above technical solution is as follows: the face recognition starting device is used to automatically start the face recognition function of the security system when a person approaches the security area, and collect face information data; the face information data recognition device is used to upload the collected face information data to the cloud after preprocessing, perform face recognition based on the intelligent face database, and output the recognition result; the recognition result output device is used to pass the face identity authentication when the recognition result is 1, and the person can enter the security area; when the recognition result is 0, the recognition result is displayed on the terminal and the corresponding warning is issued. The face recognition method of the security system has the characteristics of artificial intelligence, that is, without arranging additional personnel for operation, the security system can judge and handle the situation according to the set recognition result; the face recognition method of the security system is accurate and reliable, and can correctly distinguish between the personnel who can enter the security area and the strangers; in addition, when the face recognition authentication cannot be effectively performed due to various reasons, the security system should be able to provide other authentication methods, so that the system can continue to operate.

[0088] The beneficial effects of the above technical scheme are that the face recognition starting device is used for automatically starting the face recognition function of the security system when a person approaches the security area, and collecting face information data; the face information data recognition device is used for uploading the collected face information data to the cloud after preprocessing, performing face recognition based on an intelligent face database, and outputting a recognition result; and the recognition result output device is used for passing the face identity authentication when the recognition result is 1, and the person can enter the security area, and displaying the recognition result on the terminal and issuing a corresponding warning prompt when the recognition result is 0. The face recognition method of the security system has the characteristics of artificial intelligence, that is, no additional personnel needs to be arranged for operation, the security system automatically judges and handles the situation according to the set recognition result; the face recognition method of the security system is accurate and reliable by using a corresponding algorithm, and can correctly distinguish between the person who can enter the security area and the stranger; in addition, when the face recognition authentication cannot be effectively performed due to various reasons, the security system should be able to provide other authentication methods, so that the system can continue to run.

[0089] In another embodiment, a computer readable storage medium stores a computer program, and the computer program is executed by a processor to implement the method of any one of claims 1-8.

[0090] The working principle of the above technical scheme is that the computer readable storage medium stores a computer program, and the computer program is executed by a processor to implement the method of any one of claims 1-8.

[0091] The beneficial effects of the above technical scheme are that the computer readable storage medium stores a computer program, and the computer program is executed by a processor to implement the method of any one of claims 1-8. The running of the face recognition method of the security system is realized by the storage medium, and the security system has the characteristics of artificial intelligence, that is, no additional personnel needs to be arranged for operation, the security system automatically judges and handles the situation according to the set recognition result; the face recognition method of the security system is accurate and reliable by using a corresponding algorithm, and can correctly distinguish between the person who can enter the security area and the stranger; in addition, when the face recognition authentication cannot be effectively performed due to various reasons, the security system should be able to provide other authentication methods, so that the system can continue to run.

[0092] Obviously, those skilled in the art can make various modifications and variations to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application also intends to include these modifications and variations.

Claims

1. A face recognition method based on a security system, characterized in that, include: S101: When a person approaches the security area, the security system automatically activates the facial recognition function and collects facial information data; S102: After preprocessing the collected facial information data, upload it to the cloud, perform facial recognition based on the intelligent facial database, and output the recognition result; S103: When the recognition result is 1, the facial recognition is successful and the person can enter the security area. When the recognition result is 0, the recognition result will be displayed on the terminal and a corresponding warning will be issued. Step S102 includes: S1021: Perform normalization and grayscale preprocessing on the collected facial information data, and upload the preprocessed facial information data to the cloud; S1022: Dimensionality reduction processing is performed on the facial information data uploaded to the cloud using the specified feature extraction method to obtain the facial information data to be identified; S1023: Train face image recognition based on intelligent face database, obtain corresponding intelligent classification and recognition model, classify and recognize the face information data to be recognized based on intelligent classification and recognition model, and output recognition results; The step S1023, which involves training facial image recognition based on an intelligent facial database to obtain a corresponding intelligent classification and recognition model, includes: collecting facial information of people who can enter the security area and facial information of several strangers, performing normalization and grayscale preprocessing operations, storing the preprocessed facial information data in the intelligent facial database, extracting double-layer MB-LBP features from the preprocessed facial information data, calculating the histogram features of MB-LBP, calculating the Euclidean distance using the histogram features, obtaining the optimal discrimination threshold through the Euclidean distance, obtaining the parameters of the OC-SVM model using the histogram features, and obtaining a first recognition classifier specifically for recognizing stranger faces based on the parameters of the OC-SVM model and the optimal discrimination threshold. Based on the training process of the first recognition classifier, a second classifier is obtained by using the facial information and personal identity information of people who can enter the security area.

2. The face recognition method based on a security system according to claim 1, characterized in that, Step S101 includes: S1011: A human infrared sensor is installed at the entrance of the security area. When a person approaches the security area, the human infrared sensor detects the person's approach and the security system automatically turns on the camera to capture images. S1012: Collect the original information data of the current person through the camera image acquisition function; S1013: Detect and crop the face area in the original information data, remove the useless background parts in the original information data, and obtain the face information data.

3. The face recognition method based on a security system according to claim 1, characterized in that, Step S103 includes: S1031: Security workers can view facial image information data, analyze data, and identify results through a terminal. The identification result includes 1 or 0. S1032: If the recognition result is 1, the facial recognition is successful and the person can enter the security area. The information of the person entering the security area is recorded in the cloud data recording section. S1033: If the recognition result is 0, the facial recognition authentication fails, the terminal issues a corresponding warning, and the security personnel re-identify and confirm the current person.

4. The face recognition method based on a security system according to claim 1, characterized in that, The S1022 step includes: in the process of dimensionality reduction of the face information data uploaded to the cloud, the preprocessed face information data is first subjected to feature space dimensionality reduction and face recognition feature optimization processing, the original high-dimensional feature set is compressed, the dimension of the face image in the feature subspace is calculated, discriminative features are extracted in the feature subspace, and the feature dimensionality is further reduced.

5. The face recognition method based on a security system according to claim 1, characterized in that, In step S1023, the classification and recognition of the facial information data to be identified based on the intelligent classification and recognition model includes the following steps: First, the facial image information to be identified is preprocessed by normalization and grayscale conversion. Then, the preprocessed facial image is subjected to double-layer MB-LBP feature extraction, and the histogram data of the double-layer MB-LBP is statistically analyzed. The histogram data is output to the first recognition classifier for recognition. If the recognition result of the first recognition classifier is a stranger, the result 0 is output directly, the next step is stopped, and this recognition result is judged as a failure of facial identity authentication. It is stored as the identity of the face to be identified in the intelligent facial database. If the recognition result is a person who can enter the security area, the result is output as 1, and the identity information of the person who can enter the security area is identified by the second classifier. The identity information is displayed on the terminal.

6. The face recognition method based on a security system according to claim 1, characterized in that, Then, the Euclidean distance is calculated using histogram features to obtain the optimal discrimination threshold. This includes: in the process of detecting and matching facial feature points, discrimination is performed using Euclidean distance, and the calculated Euclidean distance is compared with the preset Euclidean distance threshold to identify the identity of the person. The rules for identifying individuals include: if the Euclidean distance between the feature information of a face in the face database and the actual feature information is less than a set value, it means that the detected face is in the intelligent face database; if the Euclidean distance between each face feature vector in the intelligent face database and the detected face feature vector is greater than the set Euclidean distance value, it means that the detected face is not in the intelligent face database and the individual's identity has not been successfully identified.

7. A facial recognition device based on a security system, characterized in that, include: The facial recognition activation device is used to automatically activate the facial recognition function of the security system and collect facial information data when a person approaches the security area; The facial information data recognition device is used to preprocess the collected facial information data and upload it to the cloud, perform facial recognition based on the intelligent facial database, and output the recognition result; The recognition result output device is used to ensure that when the recognition result is 1, the facial identity authentication is successful and the person can enter the security area; when the recognition result is 0, the recognition result is displayed on the terminal and a corresponding warning reminder is issued. The facial information data recognition device is used to preprocess the collected facial information data and upload it to the cloud, perform facial recognition based on an intelligent facial database, and output the recognition results, including: The collected facial information data is preprocessed by normalization and grayscale conversion, and then uploaded to the cloud. By using a specified feature extraction method, the facial information data uploaded to the cloud is dimensionality reduced to obtain the facial information data to be identified; Face image recognition training is performed based on an intelligent face database to obtain the corresponding intelligent classification and recognition model. The face information data to be recognized is classified and recognized based on the intelligent classification and recognition model, and the recognition result is output. The process of training a corresponding intelligent classification and recognition model for face image recognition based on an intelligent face database includes: collecting face information of people who can enter the security area and face information of several strangers, performing normalization and grayscale preprocessing operations, storing the preprocessed face information data in the intelligent face database, extracting double-layer MB-LBP features from the preprocessed face information data, calculating the histogram features of MB-LBP, calculating the Euclidean distance using the histogram features, obtaining the optimal discrimination threshold through the Euclidean distance, obtaining the parameters of the OC-SVM model using the histogram features, and obtaining a first recognition classifier specifically for recognizing stranger faces based on the parameters of the OC-SVM model and the optimal discrimination threshold. Based on the training process of the first recognition classifier, a second classifier is obtained by using the facial information and personal identity information of people who can enter the security area.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method as described in any one of claims 1 to 6.

Images