阅读说明：本技术 安防方法、安防系统和自动取款机设备 (Security method, security system and automatic teller machine equipment ) 是由 黄欣 宋荣 高燕煦 赵恺伦 于 2021-07-16 设计创作，主要内容包括：本申请公开了一种用于自动取款机的安防方法、安防系统和自动取款机,可用于金融领域,安防方法包括以下步骤：获取用户面部表情信息和语音音频；根据面部表情信息进行情绪类型分析,并输出风险值R1；根据语音音频进行情绪类型分析,并输出风险值V1；根据语音音频提取文字内容,并与风险语言库核对,输出风险值V2；根据风险值R1、风险值V1和风险值V2加权计算得出结果值T；当结果值T超过报警阈值时,输出报警通知。根据本申请的安防方法,能及时将风险情况通知网点安保值班人员,以帮助银行判断是否存在银行卡盗取、盗刷、诈骗、抢劫等风险行为,帮助银行识别潜在的安全风险事件,对维护社会稳定和杜绝违法犯罪有一定帮助。(The application discloses a security method, a security system and an automatic teller machine for the automatic teller machine, which can be used in the financial field, wherein the security method comprises the following steps: acquiring facial expression information and voice and audio of a user; performing emotion type analysis according to the facial expression information, and outputting a risk value R1; performing emotion type analysis according to the voice audio, and outputting a risk value V1; extracting character content according to the voice audio, checking the character content with a risk language library, and outputting a risk value V2; calculating a result value T in a weighting mode according to the risk value R1, the risk value V1 and the risk value V2; and when the result value T exceeds the alarm threshold value, outputting an alarm notice. According to the security method, the risk condition can be timely notified to security personnel on duty at the outlet, so that the bank can be helped to judge whether the risk behaviors such as stealing, swiping, cheating, robbery and the like of the bank card exist, the bank can be helped to identify potential security risk events, and certain help is provided for maintaining social stability and avoiding illegal crimes.)

1. A security method for an automatic teller machine, comprising the steps of:

acquiring facial expression information and voice and audio of a user;

performing emotion type analysis according to the facial expression information, and outputting a risk value R1;

performing emotion type analysis according to the voice audio, and outputting a risk value V1;

extracting character contents according to the voice audio, checking the character contents with a risk language library, and outputting a risk value V2;

calculating a result value T according to the risk value R1, the risk value V1 and the risk value V2 in a weighting mode;

and when the result value T exceeds an alarm threshold value, outputting an alarm notice.

2. The security method according to claim 1, further comprising:

judging whether the facial expression information is clear or not;

when the facial expression information is clear, performing emotion type analysis; and

when the facial expression information is unclear, the emotion type analysis is skipped, and the risk value R1 is assigned to 0.

3. The security method of claim 2, wherein performing emotion type analysis based on the facial expression information and outputting a risk value R1 comprises:

judging the emotion type of the user according to the facial expression information;

when the emotion type of the user is normal, assigning the risk value R1 as 0;

when the user emotion type is abnormal, the risk value R1 is assigned as a first constant C1.

4. The security method according to claim 1, wherein performing emotion type analysis based on the voice audio and outputting a risk value V1 comprises:

judging the emotion type of the user through the voice audio;

when the emotion type of the user is normal, the risk value V1 is assigned to be 0;

when the user emotion type is abnormal, the risk value V1 is assigned as a second constant C2.

5. The security method according to claim 1, wherein extracting text contents according to the voice audio and checking the text contents with a risk language library, and outputting a risk value V2 comprises:

checking the extracted text content with a risk language library;

assigning the risk value V2 as a third constant C3 when the textual content hits;

when the text content is not hit, the risk value V2 is assigned to 0.

6. The security method according to any one of claims 1 to 4, wherein the emotion type comprises a negative emotion.

7. The security method of claim 1, wherein the alert notification comprises: and the information is notified through a short message of a risk system or is notified through a large-screen alarm.

8. A security system for an automated teller machine comprising:

an identification module, the identification module comprising: the system comprises a facial recognition module and a voice recognition module, wherein the facial recognition module is used for acquiring facial expression information of a user, and the voice recognition module is used for acquiring voice and audio of the user and extracting text contents;

an emotional information analysis module, the emotional information analysis module comprising: the system comprises a facial emotion information analysis module and a voice emotion information analysis module, wherein the facial emotion information analysis module is used for calculating a risk value R1 of facial expression information of a user, and the voice emotion information analysis module is used for calculating a risk value V1 of voice audio of the user and calculating a text content risk value V2;

the comprehensive analysis module is used for weighting calculation of the risk value output by the emotion information analysis module and obtaining a result value T;

and the alarm judging module is used for judging whether the result value T exceeds an alarm threshold value or not, and outputting an alarm notice if the result value T exceeds the alarm threshold value.

9. The security system of claim 8, wherein the facial recognition module is further configured to determine whether the facial expression information is clear.

10. An automatic teller machine apparatus comprising:

a housing;

the camera is arranged around the shell and used for acquiring facial expression information of a user;

a microphone disposed about the housing for capturing user speech audio; and

the security system of claim 8 or 9, the security system in communicative connection with the camera and the microphone, respectively, to receive the user facial expression information and the user voice audio.

11. The atm apparatus of claim 10, wherein the housing has an escape slot disposed therein, the escape slot facing a face of the user, the camera and the microphone being embedded in the escape slot.

Technical Field

The application relates to the technical field of artificial intelligence, can be used in the financial field, and more particularly relates to a security method and a security system for an automatic teller machine and an automatic teller machine device.

Background

In recent years, many criminal sites or telecommunication forcing users to conduct involuntary transactions on the automatic teller machines, and money loss of the users is caused, so that the users can be judged whether to conduct the voluntary transactions in time.

Disclosure of Invention

The present application is directed to solving at least one of the problems in the prior art.

Therefore, a first objective of the present application is to provide a security method for an atm, which can help a bank identify a potential security risk event, and help to maintain social stability and prevent illegal crime;

a second objective of the present application is to provide a security system for an atm, which can carry the security method described above;

a third objective of the present application is to provide an atm apparatus, which includes the security system.

In order to achieve the above object, a first aspect of the present application provides a security method for an automatic teller machine, including the steps of:

acquiring facial expression information and voice and audio of a user;

performing emotion type analysis according to the facial expression information, and outputting a risk value R1;

performing emotion type analysis according to the voice audio, and outputting a risk value V1;

extracting character contents according to the voice audio, checking the character contents with a risk language library, and outputting a risk value V2;

calculating a result value T according to the risk value R1, the risk value V1 and the risk value V2 in a weighting mode;

and when the result value T exceeds an alarm threshold value, outputting an alarm notice.

According to the security method, the emotion types of the users are identified through the facial expressions and the voice audio, the risk conditions can be timely notified to security operators on duty, so that a bank is helped to judge whether risk behaviors such as stealing, swiping, cheating and robbing of a bank card exist, the bank is helped to identify potential security risk events, and certain help is provided for maintaining social stability and avoiding illegal crimes.

Further, still include:

judging whether the facial expression information is clear or not;

when the facial expression information is clear, performing emotion type analysis; and

when the facial expression information is unclear, the emotion type analysis is skipped, and the risk value R1 is assigned to 0.

Further, performing emotion type analysis based on the facial expression information, and outputting a risk value R1 includes:

judging the emotion type of the user according to the facial expression information;

when the emotion type of the user is normal, assigning the risk value R1 as 0;

when the user emotion type is abnormal, the risk value R1 is assigned as a first constant C1.

Further, performing emotion type analysis based on the voice audio, and outputting a risk value V1 includes:

judging the emotion type of the user through the voice audio;

when the emotion type of the user is normal, the risk value V1 is assigned to be 0;

when the user emotion type is abnormal, the risk value V1 is assigned as a second constant C2.

Further, extracting text contents according to the voice audio, and checking the text contents with a risk language library, wherein outputting a risk value V2 comprises:

checking the extracted text content with a risk language library;

assigning the risk value V2 as a third constant C3 when the textual content hits;

when the text content is not hit, the risk value V2 is assigned to 0.

Further, the emotion type includes a negative emotion.

Further, the alert notification includes: and the information is notified through a short message of a risk system or is notified through a large-screen alarm.

A second aspect of the present application provides a security system for an automated teller machine, comprising:

an identification module, the identification module comprising: the system comprises a facial recognition module and a voice recognition module, wherein the facial recognition module is used for acquiring facial expression information of a user, and the voice recognition module is used for acquiring voice and audio of the user and extracting text contents;

an emotional information analysis module, the emotional information analysis module comprising: the system comprises a facial emotion information analysis module and a voice emotion information analysis module, wherein the facial emotion information analysis module is used for calculating a risk value R1 of facial expression information of a user, and the voice emotion information analysis module is used for calculating a risk value V1 of voice audio of the user and calculating a text content risk value V2;

the comprehensive analysis module is used for weighting calculation of the risk value output by the emotion information analysis module and obtaining a result value T;

and the alarm judging module is used for judging whether the result value T exceeds an alarm threshold value or not, and outputting an alarm notice if the result value T exceeds the alarm threshold value.

Further, the facial recognition module may be further configured to determine whether the facial expression information is clear.

A third aspect of the present application provides an automated teller machine apparatus comprising: a housing; the camera is arranged around the shell and used for acquiring facial expression information of a user; a microphone disposed about the housing for capturing user speech audio; and the security system is in communication connection with the camera and the microphone respectively to receive the facial expression information of the user and the voice and audio of the user.

Further, be provided with on the casing and dodge the groove, it is just right with user's face to dodge the groove, the camera with the microphone is embedded into dodge the inslot.

Additional aspects and advantages of the present application will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the present application.

Drawings

The above and other objects, features and advantages of the present disclosure will become more apparent from the following description of embodiments of the present disclosure with reference to the accompanying drawings, in which:

FIG. 1 is a schematic flow chart of a security method according to an embodiment of the present application;

FIG. 2 is a schematic diagram of a security system according to an embodiment of the application;

fig. 3 is a block diagram of an automated teller machine apparatus according to an embodiment of the present application.

Reference numerals:

a recognition module 1, a face recognition module 11, a speech recognition module 12,

emotion information analysis module 2, facial emotion information analysis module 21, voice emotion information analysis module 22,

a comprehensive analysis module 3, an alarm discrimination module 4,

an automatic teller machine (atm) device 100,

so as to avoid the grooves 101 from being formed,

the automatic teller machine comprises a shell 10, a camera 20, a microphone 30 and an automatic teller machine body 40.

Detailed Description

Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. It should be understood that the description is illustrative only and is not intended to limit the scope of the present disclosure. In the following detailed description, for purposes of explanation, numerous specific details are set forth in order to provide a thorough understanding of the embodiments of the disclosure. It may be evident, however, that one or more embodiments may be practiced without these specific details. Moreover, in the following description, descriptions of well-known structures and techniques are omitted so as to not unnecessarily obscure the concepts of the present disclosure.

The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the disclosure. The terms "comprises," "comprising," and the like, as used herein, specify the presence of stated features, steps, operations, and/or components, but do not preclude the presence or addition of one or more other features, steps, operations, or components.

Where a convention analogous to "A, B or at least one of C, etc." is used, in general such a construction is intended in the sense one having skill in the art would understand the convention (e.g., "a system having at least one of A, B or C" would include but not be limited to systems that have a alone, B alone, C alone, a and B together, a and C together, B and C together, and/or A, B, C together, etc.). The terms "first", "second" and "first" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, features defined as "first", "second", may explicitly or implicitly include one or more of the described features.

In recent years, many criminal sites or telecommunication forcing users to conduct involuntary transactions on the automatic teller machines, and money loss of the users is caused, so that the users can be judged whether to conduct the voluntary transactions in time.

With the development of artificial intelligence technology, the technology for completing emotion analysis through face images is mature, the ability of analyzing the emotion change of a user according to the face images of the user at different time intervals is achieved, and the emotion of the user can be judged more easily through voice with the help of AI technology.

The method combines an artificial intelligence technology with the automatic teller machine, utilizes technologies such as face emotion recognition, voice emotion recognition and natural language processing to recognize the emotion state of a customer when the customer transacts on the automatic teller machine, and judges whether the customer is in an abnormal state, so as to help a bank judge whether security event risks such as stealing, embezzlement, fraud and robbery of a bank card exist.

A security method of an automatic teller machine according to an embodiment of the present application is described below with reference to fig. 1 to 3.

According to one embodiment of the present application, a security method for an automated teller machine is disclosed, and referring to fig. 1, the method may be performed according to the following steps or operations.

In step S10, user facial expression information and voice audio are acquired.

When a user station transacts business in front of the automatic deposit machine, the automatic deposit machine firstly acquires facial expression information and voice and audio of the transacted user, and specifically can acquire the facial expression information and the voice and audio by shooting videos and recording.

For example: when a user transacts business, the user shoots short videos for about 10 seconds at intervals and records the videos.

After step S10, it is necessary to first determine whether the facial expression information is clear; when the facial expression information is clear, performing emotion type analysis; and when the facial expression information is unclear, skipping the emotion type analysis and assigning the risk value R1 as 0.

After shooting, firstly, the definition of the acquired facial expression information of the user is judged, whether the facial expression information can be acquired or not is judged, step S20 can be executed in the case of definition, if the facial expression information cannot be acquired, emotion type analysis is directly skipped, the risk value R1 is assigned to 0, and step S30 is executed.

In the withdrawal process, the user may be in a shaking state all the time, so that a clear facial expression cannot be intercepted, or the facial expression cannot be intercepted because a device for acquiring the facial expression of the user has a stain, or the user may be blocked by a hat brim or other objects, so that facial expression information cannot be identified.

Step S20 may be performed in the case where the facial expression information is clear.

In step S20, emotion type analysis is performed according to the facial expression information, and a risk value R1 is output.

Wherein performing emotion type analysis based on the facial expression information and outputting the risk value R1 includes:

judging the emotion type of the user according to the facial expression information;

when the emotion type of the user is normal, assigning a risk value R1 as 0;

when the user's emotional type is abnormal, the risk value R1 is assigned to a first constant C1.

It is understood that after analyzing the facial expression information, the risk value R1 is output, and R1 may be two resulting values. One is that after the analysis of the facial expression, the type of the emotion of the user is good, and R1 is made 0, and the other is that after the analysis of the facial expression, the type of the emotion of the user is abnormal, and R1 is made a first constant C1.

When R1 is 0, it can be considered that the user is working normally in front of the atm, and there is no abnormality around the user, and the user is voluntary operation; when R1 is the first constant C1, it can be considered that the user may be involved in telephone fraud or robbery by thieves.

It should be noted that the first constant C1 may be a constant value or a value within an interval, representing the degree of importance.

In one embodiment, the first constant C1 may be a constant value of 1, that is, R1 only determines whether the user emotion type is abnormal. For example: when the user emotion type is good, direct output R1 is 0, and when the user emotion type is abnormal, direct output R1 is 1.

The method can be simple and convenient, whether the user is in the abnormal business handling condition or not can be obtained only through the facial expression of the user, and the calculation method is simple and convenient.

In another embodiment, the value range of the first constant C1 is set between 1 and 9 according to the degree of abnormality of the user emotion type, 9 represents that the user emotion type is abnormally severe, and 1 represents that the user emotion type is abnormally light, that is, R1 can not only determine whether the user emotion type is abnormal, but also display the degree of the user emotion type which is abnormally severe. For example: when the user's emotion type is good, output R1 is 0; and when the emotion type of the user is abnormal, if the emotion type of the user is abnormal seriously, the output R1 is 9, and if the emotion type of the user is abnormal lightly, the output R1 is 1.

The method can be more accurate, whether the user is in the abnormal business handling condition or not is judged through the facial expression of the user, and the probability of error judgment in the process of capturing the facial expression is reduced.

In the case where clear facial expression information cannot be obtained, or after step S20 is performed, step S30 may be performed.

In step S30, emotion type analysis is performed based on the voice audio, and a risk value V1 is output.

Performing emotion type analysis from the speech audio, and outputting a risk value V1 includes:

judging the emotion type of the user through voice audio;

when the emotion type of the user is normal, the risk value V1 is assigned to 0;

when the user emotion type is abnormal, the risk value V1 is assigned as a second constant C2.

It is understood that after analyzing the voice audio, the risk value V1 is output, and V1 may be two resulting values. One is to make V1 0 if the user emotion type is good after the voice audio analysis is performed, and the other is to make V1 a second constant C2 if the user emotion type is abnormal after the voice audio analysis is performed.

When V1 is 0, it can be considered that the user is working normally in front of the atm, there is no abnormality around, and it is voluntary operation; when V1 is the second constant C2, it can be considered that the user is likely to be involved in telephone fraud or robbery by thieves.

Speech audio is understood to mean the speech sounds of a user when speaking, the tone of speech fluctuating, the sentence continuation, and so on.

It should be noted that the second constant C2 may be a constant value or a value within an interval, representing the degree of importance.

In one embodiment, the second constant C2 may be a constant value of 1, that is, V1 only determines whether the user emotion type is abnormal. For example: when the user emotion type is good, the direct output V1 is 0, and when the user emotion type is abnormal, the direct output V1 is 1.

The method can be simple and convenient, the condition whether the user is in abnormal business handling or not can be obtained only through the voice audio, and the calculation method is simple and convenient.

In another embodiment, the value range of the second constant C2 is set between 1 and 9 according to the degree of abnormality of the user emotion type, 9 represents that the user emotion type is abnormally severe, and 1 represents that the user emotion type is abnormally light, that is, the V1 can not only determine whether the user emotion type is abnormal, but also display the degree of the user emotion type which is abnormally severe. For example: when the user emotion type is good, the output V1 is 0; and when the user emotion type is abnormal, if the user emotion type is abnormal seriously, the output V1 is 9, and if the user emotion type is abnormal lightly, the output V1 is 1.

The method can be more accurate, whether the user is in the condition of handling the service abnormally is judged through the voice audio, and the probability of error judgment existing when the user listens to the voice audio is reduced.

After the risk value V1 is obtained, step S40 is performed.

In step S40, the text content is extracted from the speech audio and checked against the risk language library, and a risk value V2 is output.

Extracting character content according to the voice audio and checking the character content with a risk language library, wherein the outputting of the risk value V2 comprises the following steps:

checking the extracted text content with a risk language library;

when the text content hits, the risk value V2 is assigned to the third constant C3;

when the text content is not hit, the risk value V2 is assigned to 0.

It is understood that after analyzing the extracted text content, the risk value V2 is output, and V2 may be two result values. One is to make V2 0 if the result is no hit after analysis of the extracted text and checking with the risk language library, and the other is to make V2 a third constant C3 if the result is hit after analysis of the extracted text and checking with the risk language library.

When V2 is 0, it can be considered that the user is working normally in front of the atm, there is no abnormality around, and it is voluntary operation; when V2 is the third constant C3, it is considered that the user may be involved in telephone fraud or robbery by thieves.

The extracted text content is the text content of the voice audio, which can be described by the user, or the content recorded by the voice audio transmitted by the surrounding people or the telephone.

The risk language library is a collection of risk language libraries formed by integrating the languages commonly used by criminals in the practice of fraud in the previous case, and in case of hit, it indicates that the user may be suffering from telephone fraud or robbery by thieves.

It should be noted that the third constant C3 may be a constant value or a value within an interval, representing the degree of importance.

In one embodiment, the third constant C3 may be a constant value of 1, i.e., V2 only determines whether the extracted textual content hits in the risk language library. For example: when the extracted text content does not have the same sentence after being checked against the risk language base, direct output V2 is 0, and when the extracted text content hits a sentence in the risk language base, direct output V2 is 1.

The method can be simple, whether the user is in abnormal business handling condition can be obtained only through the extracted text content, and the calculation method is simple.

In another embodiment, the value range of the third constant C3 is set between 1 and 9 according to the hit degree of the extracted text content checked with the risk language library, where 9 represents that the hit rate of the text in the speech audio is very high, and 1 represents that the hit rate of the text in the speech audio is low, that is, V2 can not only determine whether the extracted text content hits the risk language library, but also display the hit probability, that is, the severity degree, of the extracted text content. For example: when the extracted text content does not have the same words after being checked with the risk language library, the output V2 is 0; when the extracted text content hits a word in the risk language base, if the hit rate of the extracted text content with respect to the risk language base is extremely high, the output V2 is 9, and if the hit rate of the extracted text content with respect to the risk language base is low, the output V2 is 1. The problem of high and low statement hit rate can be set according to bank risk requirements.

The method can be more accurate, whether the user is in the condition of abnormal business handling is judged through the extracted text content, and the probability of error judgment of the extracted text content is reduced.

According to one embodiment of the application, the type of emotion comprises a negative emotion.

For example, negative emotions include: fear, tension, anxiety, worry, falling, etc.

More specifically, the first constant C1 is output when the user's facial expression is one of the above, and the second constant C2 is output when the voice audio appears in one of the above.

After the risk value V2 is obtained, step S50 is performed.

At step S50, a result value T is calculated by weighting according to the risk value R1, the risk value V1, and the risk value V2.

The weighting calculation means that weights taken in the square error calculation are different depending on the accuracy of the risk value, and the weighting means multiplication by a coefficient, and the larger the coefficient, the larger the weight.

And weighting and calculating the risk value R1, the risk value V1 and the risk value V2 to obtain a result value T, so that the calculated result is more suitable for the actual situation.

The weighting on the risk value R1, the risk value V1 and the risk value V2 may be based on the actual situation. For example: the probability of wrong judgment of the facial expression information after acquisition is high, or the facial expression information is often unclear, the weight of the risk value R1 can be reduced, the accuracy of the text content extracted through the voice audio is high, whether the text content hits the risk language library or not can be easily checked through keywords, and the weight of the risk value V2 can be increased.

In one embodiment, the sum of the weighting coefficients before risk value R1, risk value V1, and risk value V2 is given as unit 1, the coefficient is added 0.2 before risk value R1, the coefficient is added 0.2 before risk value V1, the coefficient is added 0.6 before risk value V2,

the final result value T will be:

T＝0.2R1+0.2V1+0.6V2

in this embodiment, the weights of the facial expression information and the speech audio are the same and both are 0.2, and the weight of the extracted text content is 0.6, which is greater than the weights of the facial expression information and the speech audio, which indicates that the extracted text content is more emphasized on the risk value for checking with the risk language library than the facial expression information and the speech audio, and the result value T calculated in this way is considered to be more accurate and more precise.

After the result value T is obtained, step S60 is executed.

In step S60, when the result value T exceeds the alarm threshold, an alarm notification is output.

According to one embodiment of the present application, an alert notification includes: and the information is notified through a short message of a risk system or is notified through a large-screen alarm.

Comparing the finally obtained result value T with an alarm threshold value, and outputting an alarm notification when the result value T exceeds the alarm threshold value; when the result value T is lower than the alarm threshold, the process may return to step S10 after a certain time interval.

It should be noted that the time interval between the shooting and the recording and the time length of the shooting and the recording can be set by the bank or the state of the user, for example: if the initial detection result is that the emotional state of the user is good, the time interval for restarting shooting and recording next time can be lengthened, and the time for recording and shooting is kept unchanged; if the emotional state of the user is unstable in the primary detection, the time interval for restarting shooting and recording next time can be shortened, and the time for recording shooting is prolonged.

Through the control to the interval time of shooting and recording and the record time of shooting and recording, can be when guaranteeing the utilization of energy maximize, saved the electric energy according to the demand, monitor the condition of user's actual emotional state to prevent missing necessary information, cause user's money and property loss.

According to the security method, the emotion types of the users are identified through the facial expressions and the voice audio, the risk conditions can be timely notified to security operators on duty, so that a bank is helped to judge whether risk behaviors such as stealing, swiping, cheating and robbing of a bank card exist, the bank is helped to identify potential security risk events, and certain help is provided for maintaining social stability and avoiding illegal crimes.

Referring to fig. 2, a security system for an automatic teller machine according to an embodiment of the present application includes:

identification module 1, identification module 1 includes: the system comprises a facial recognition module 11 and a voice recognition module 12, wherein the facial recognition module 11 is used for acquiring facial expression information of a user, and the voice recognition module 12 is used for acquiring voice and audio of the user and extracting character contents;

emotion information analysis module 2, emotion information analysis module 2 includes: the system comprises a facial emotion information analysis module 21 and a voice emotion information analysis module 22, wherein the facial emotion information analysis module 21 is used for calculating a risk value R1 of facial expression information of a user, and the voice emotion information analysis module 22 is used for calculating a risk value V1 of voice audio of the user and calculating a risk value V2 of text content;

the comprehensive analysis module 3 is used for carrying out weighted calculation on the risk value output by the emotion information analysis module and obtaining a result value T;

and the alarm judging module 4 is used for judging whether the result value T exceeds an alarm threshold value or not, and outputting an alarm notification if the result value T exceeds the alarm threshold value.

According to an embodiment of the present application, the facial recognition module 11 may also be used to determine whether the facial expression information is clear.

According to the security system, the risk condition can be timely reported to security personnel on duty at a website, security personnel can follow up security events, a bank can be helped to identify potential security risk events, and certain help is provided for maintaining social stability and avoiding illegal crimes.

According to one embodiment of the application, the overall process of the security system is as follows:

1. the user transacts business before the ATM, and the face recognition module 11 and the voice recognition module 12 shoot short videos for about 10 seconds at intervals and record the videos.

2. The facial emotion information analysis module 21 processes the video, analyzes facial expression information of the user in the video, calculates an emotion type of the user, and obtains a risk value R1 of the facial expression information of the user.

3. The voice emotion information analysis module 22 processes the voice file, and the voice emotion information analysis module 22 calculates a risk value V1 of the voice audio.

4. The speech recognition module 12 processes and analyzes the speech content in natural language, extracts the text content in the recording, and checks the keywords with the risk language library, and the speech emotion information analysis module 22 calculates the text content risk value V2 according to the check result.

5. The comprehensive analysis module 3 performs weighted calculation on the risk value R1 of the facial expression information of the user, the risk value V1 of the voice audio and the risk value V2 of the text content to finally obtain a risk result value T.

6. The alarm discrimination module 4 compares the result value T with an alarm threshold value, if the result value T exceeds the alarm threshold value, the alarm discrimination module 4 informs the network point where the automatic teller machine is located of the risk information, and security personnel of the network point further confirms and processes the risk information.

An automatic teller machine apparatus 100 according to an embodiment of the present application includes: the casing 10, camera 20, microphone 30 and the security protection system.

Specifically, the camera 20 is disposed around the housing 10 and is used for acquiring facial expression information of the user; a microphone 30 is arranged around the housing 10 for acquiring voice audio of a user; the security system is in communication connection with the camera 20 and the microphone 30 respectively to receive the user facial expression information and the user voice and audio.

An automatic teller machine body 40 is arranged in the shell 10, a user needs to stand facing the automatic teller machine body 40, the camera 20 and the microphone 30 are arranged around the shell 10 and used for acquiring facial expression information and voice audio of the user and then sending the acquired facial expression information and voice audio of the user to the security system, and the security system is used for monitoring and analyzing whether illegal criminal behaviors occur or not, can judge in time and takes necessary safety measures.

According to an embodiment of the present application, an avoiding groove is formed on the housing 10, the avoiding groove is opposite to the face of the user, and the camera 20 and the microphone 30 are embedded in the avoiding groove.

Because camera 20 and microphone 30 need gather user's facial expression information and pronunciation audio frequency, the setting is more favorable to clearly, the accurate information that obtains needs when just facing the face, camera 20 and microphone 30 expose outside, just facing the face, it is inconvenient that the user handles the business before ATM body 40 on the one hand, there is the danger of colliding with the head, on the other hand is that criminal or other personnel intentionally destroy camera 20 and microphone 30, or some personnel damage camera 20 and microphone 30 when operating in front of the ATM in the accident, cause camera 20 and microphone 30 unable to use, cost of maintenance and replacement increase.

In the present application, an avoiding groove is formed in the housing 10, and the camera 20 and the microphone 30 are installed in the avoiding groove.

In an embodiment of the present application, as shown in fig. 3, an avoiding groove is formed in the housing 10 above the automatic teller machine body 40, the camera 20 and the microphone 30 are installed in the avoiding groove, and when a user stands facing the automatic teller machine body 40, the camera 20 and the microphone 30 are just at the position of the face of the user, which is beneficial for the camera 20 and the microphone 30 to clearly take facial expression information and voice and audio of the user.

The security system is installed in the automatic teller machine 100, the security method can be operated, and whether the risk behaviors such as stealing, cheating, robbing and the like exist is deduced by judging whether a user is in a passive state such as tension, anxiety, fear and the like.

It should be noted that the implementation, solved technical problems, implemented functions, and achieved technical effects of each module/unit/subunit and the like in the apparatus part embodiment are respectively the same as or similar to the implementation, solved technical problems, implemented functions, and achieved technical effects of each corresponding step in the method part embodiment.

Those skilled in the art will appreciate that various combinations and/or combinations of features recited in the various embodiments and/or claims of the present disclosure can be made, even if such combinations or combinations are not expressly recited in the present disclosure. In particular, various combinations and/or combinations of the features recited in the various embodiments and/or claims of the present disclosure may be made without departing from the spirit or teaching of the present disclosure. All such combinations and/or associations are within the scope of the present disclosure.

The embodiments of the present disclosure have been described above. However, these examples are for illustrative purposes only and are not intended to limit the scope of the present disclosure. Although the embodiments are described separately above, this does not mean that the measures in the embodiments cannot be used in advantageous combination. The scope of the disclosure is defined by the appended claims and equivalents thereof. Various alternatives and modifications can be devised by those skilled in the art without departing from the scope of the present disclosure, and such alternatives and modifications are intended to be within the scope of the present disclosure.