WO2021068616A1 - 身份验证方法、装置、计算机设备和存储介质 - Google Patents

身份验证方法、装置、计算机设备和存储介质 Download PDF

Info

Publication number
WO2021068616A1
WO2021068616A1 PCT/CN2020/106238 CN2020106238W WO2021068616A1 WO 2021068616 A1 WO2021068616 A1 WO 2021068616A1 CN 2020106238 W CN2020106238 W CN 2020106238W WO 2021068616 A1 WO2021068616 A1 WO 2021068616A1
Authority
WO
WIPO (PCT)
Prior art keywords
detected
lip
display code
face detection
random
Prior art date
Application number
PCT/CN2020/106238
Other languages
English (en)
French (fr)
Inventor
余龙龙
Original Assignee
深圳壹账通智能科技有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 深圳壹账通智能科技有限公司 filed Critical 深圳壹账通智能科技有限公司
Publication of WO2021068616A1 publication Critical patent/WO2021068616A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F21/00Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
    • G06F21/30Authentication, i.e. establishing the identity or authorisation of security principals
    • G06F21/31User authentication
    • G06F21/32User authentication using biometric data, e.g. fingerprints, iris scans or voiceprints

Definitions

  • This application relates to the field of artificial intelligence technology, in particular to an identity verification method, device, computer equipment and storage medium.
  • Traditional identity authentication schemes include the use of passwords or passwords to determine personal identity. For example, when a user uses an application, the background of the application can confirm the user’s identity according to the login password entered by the user, or the SMS verification code when the user forgets the password. It also includes the use of face recognition technology to confirm the user's identity, by extracting the user's facial features, and comparing the pre-stored facial features with the facial features extracted during verification to confirm the user's identity.
  • the inventor realizes that in the traditional identity authentication scheme, there is still the problem of password or password being stolen, and in the face recognition technology, there is also a pre-recorded video including the user’s face, which illegally obtains the user’s head and eyes. The movement of the head and mouth, etc., attack the face recognition system, resulting in the problem of low identity authentication security.
  • an identity verification method e.g., an identity verification method, device, computer equipment, and storage medium are provided.
  • An identity verification method including:
  • a face detection instruction is triggered, and face detection is performed on the object to be detected according to the face detection instruction to obtain a face detection result;
  • the random display code sequence is composed according to at least one of random numbers, random letters, or random Chinese characters;
  • the lip movement feature display code and the voice text display code are compared with the random display code sequence, when the lip movement feature display code and the voice text display code both conform to the random display code sequence When, perform a live detection on the object to be detected;
  • the face image file of the object to be detected pre-stored in the database is called, compared with the face detection result, identity authentication is performed, and the identity authentication result is obtained.
  • An identity verification device includes:
  • the face detection module is configured to trigger a face detection instruction when the login operation of the object to be detected is detected, and perform face detection on the object to be detected according to the face detection instruction to obtain a face detection result;
  • a random display code sequence generating module configured to generate a random display code sequence according to the face detection result; the random display code sequence is composed of any two forms or more of random numbers, random letters or random Chinese characters;
  • the lip motion feature extraction module is used to extract lip motion features of the object to be detected, and use a pre-trained lip language recognition model to perform feature analysis on the lip motion features to obtain a corresponding lip motion feature display code;
  • a voice recognition module configured to extract the interpretation audio of the object to be detected, perform voice recognition on the interpretation audio, and convert the interpretation audio into a voice text display code
  • the comparison module is used to compare the lip action feature display code and the voice text display code with the random display code sequence, when the lip action feature display code and the voice text display code both match When the code sequence is randomly displayed, live detection is performed on the object to be detected;
  • the identity authentication module is used for when it is confirmed that the object to be detected is a living object, call the face image file of the object to be detected prestored in the database, compare with the result of the face detection, perform identity authentication, and obtain identity authentication result.
  • a computer device including a memory and one or more processors, the memory stores computer readable instructions, and when the computer readable instructions are executed by the processor, the one or more processors execute The following steps:
  • a face detection instruction is triggered, and face detection is performed on the object to be detected according to the face detection instruction to obtain a face detection result;
  • the random display code sequence is composed according to at least one of random numbers, random letters, or random Chinese characters;
  • the lip movement feature display code and the voice text display code are compared with the random display code sequence, when the lip movement feature display code and the voice text display code both conform to the random display code sequence When, perform a live detection on the object to be detected;
  • the face image file of the object to be detected pre-stored in the database is called, compared with the face detection result, identity authentication is performed, and the identity authentication result is obtained.
  • One or more computer-readable storage media storing computer-readable instructions.
  • the one or more processors perform the following steps:
  • a face detection instruction is triggered, and face detection is performed on the object to be detected according to the face detection instruction to obtain a face detection result;
  • the random display code sequence is composed according to at least one of random numbers, random letters, or random Chinese characters;
  • the lip movement feature display code and the voice text display code are compared with the random display code sequence, when the lip movement feature display code and the voice text display code both conform to the random display code sequence When, perform a live detection on the object to be detected;
  • the face image file of the object to be detected pre-stored in the database is called, compared with the face detection result, identity authentication is performed, and the identity authentication result is obtained.
  • the above-mentioned identity verification method, device, computer equipment and storage medium when the login operation of the object to be detected is detected, face detection of the object to be detected is performed to obtain the face detection result, and a random display code sequence is generated according to the face detection result,
  • the random display code sequence is composed of at least one form of random numbers, random letters or random Chinese characters. Due to the use of randomly generated numbers, letters and Chinese character sequences as identifiers, it is difficult to record audio files and video files in advance, which can reduce the risk of being attacked.
  • the corresponding lip motion feature display code is obtained.
  • extract the interpretation audio of the object to be detected perform voice recognition on the interpretation audio, convert the interpretation audio into a voice text display code, and then compare the lip movement feature display code and the voice text display code with the random display code sequence.
  • the object to be detected is subjected to live detection.
  • the face image file of the object to be detected pre-stored in the database is called, and The face detection results are compared, identity authentication is performed, and the identity authentication result is obtained. Due to the combination of face recognition, lip language recognition and voice recognition at the same time, they can complement and improve each other, further enhancing the reliability and security of identity authentication.
  • Fig. 1 is an application scenario diagram of an identity verification method according to one or more embodiments
  • Fig. 2 is a schematic flowchart of an identity verification method according to one or more embodiments
  • Figure 3 is a block diagram of an identity verification device according to one or more embodiments.
  • Figure 4 is a block diagram of a computer device according to one or more embodiments.
  • the identity verification method provided in this application can be applied to the application environment as shown in FIG. 1.
  • the terminal 102 and the server 104 communicate through the network.
  • the server 104 detects the login operation of the object to be detected on the terminal 102, it triggers a face detection instruction, performs face detection on the object to be detected according to the face detection instruction, obtains the face detection result, and generates a random display based on the face detection result
  • the code sequence wherein the random display code sequence is composed of at least one form of random numbers, random letters or random Chinese characters.
  • the server 104 extracts the lip motion features of the object to be detected, and uses a pre-trained lip language recognition model to perform feature analysis on the lip motion features to obtain a corresponding lip motion feature display code.
  • the terminal 102 may be, but is not limited to, various personal computers, notebook computers, smart phones, tablet computers, and portable wearable devices.
  • the server 104 may be implemented by an independent server or a server cluster composed of multiple servers.
  • an identity verification method is provided. Taking the method applied to the server in FIG. 1 as an example for description, the method includes the following steps:
  • Step S202 When the login operation of the object to be detected is detected, a face detection instruction is triggered, and face detection of the object to be detected is performed according to the face detection instruction to obtain a face detection result.
  • the login status of the object to be detected is acquired, where the login status includes login success and login failure.
  • the face detection instruction is triggered, and the camera program is opened according to the face detection instruction, and the camera program is used to detect the face of the object to be detected.
  • the face detection result is generated.
  • the login operation of the object to be detected in the application program includes a login operation in which the object to be detected inputs an account password to log in, or obtains a real-time verification code and logs in by inputting an account.
  • the face detection instruction is triggered when the login is successful, that is, there is no error in the account and password. Enter the account number of the object to be detected and enter the verification code obtained in real time to log in.
  • the face detection instruction is also triggered when the login is successful.
  • the face detection instruction is used to detect the face part of the object to be detected that is currently performing the login operation.
  • the camera program on the terminal can be opened according to the face detection instruction, and the camera can be used to detect the face to determine whether the to-be-detected object is detected.
  • the face part of the object, and the face detection result is obtained.
  • the interface prompts "please aim at the camera” until the camera detects the face of the object to be detected.
  • Step S204 Generate a random display code sequence according to the face detection result; the random display code sequence is composed of at least one form of random numbers, random letters or random Chinese characters.
  • the preset requirements include the length of the random display code sequence and the composition of the random display code sequence, and when the face detection result is determined to be detected When detecting the face of the object, a random display code sequence is randomly generated according to the preset requirements of the random display code sequence.
  • the random display code sequence can be composed of at least one form of random numbers, random letters, and random Chinese characters.
  • the random number sequence is generated based on numbers from 0 to 9, and the number of numbers included is greater than or equal to 6 digits. Including but not limited to 6 bits, 8 bits and 10 bits, etc.
  • a 6-digit random number sequence can be 156871, 589624, 896547, etc.
  • an 8-digit random number sequence can be 85412369, 78452369, etc.
  • the number sequence generation does not follow a certain rule and is generated randomly, which is different from storing the number sequence in advance. The situation of calling at time can avoid the problem of digital sequence being misappropriated.
  • the random letter sequence is generated based on 26 English letters, and the number of digits in the sequence is not limited, more than 6 digits, including but not limited to 6 digits, 8 digits, and 10 digits.
  • a 6-digit random letter sequence can be oputrd, ruighd, swertg, etc. The same applies to 8-digit and 10-digit random letter sequences.
  • the generated random sequence length is relatively long, and there are many types of sequences.
  • the length of the random sequence can be set to 8 or 10 bits, and the sequence type It can be set to include two or more of numbers, letters and Chinese characters at the same time.
  • an 8-digit random display code sequence that includes both numbers and letters can be in various forms such as soer5648, 9867ogsi, ru98yt03, 75un09eg, wr8750jt, and 36thfb68.
  • the 10-digit random display code sequence that includes numbers, letters, and Chinese characters at the same time can be in various forms such as "49pr if 06et", "ty start 73rb05" and "update 34wh96rs".
  • step S206 the lips motion features of the object to be detected are extracted, and the pre-trained lip language recognition model is used to perform feature analysis on the lips motion features to obtain the corresponding lip motion feature display code.
  • the lip motion features are compared with the preset The lip movement feature samples are compared.
  • the display code corresponding to the lip action feature sample is used as the lip action feature display code.
  • the server extracts the lip motion features of the object to be detected in real time, and the lip shape picture of the object to be detected during the entire interpretation process can be taken by using a camera.
  • the lip shape picture of the object to be detected is recognized and analyzed, and the lip motion feature display code corresponding to the lip shape is obtained.
  • the randomly generated display code sequence is a 6-digit random number sequence 569841
  • multiple lip-shaped pictures are captured by the camera, and when the trained lip-language recognition model is used to identify and analyze each lip-shaped picture,
  • the corresponding lip movement feature numbers can be analyzed separately, which can be 8, 9, 5, 6, 1, and 4. According to the generation time of the lip picture, the lip movement feature numbers corresponding to the lip picture are sorted to get the lip movement Characteristic number sequence.
  • the randomly generated display code sequence is an 8-digit alphanumeric sequence 9852tyrd
  • using multiple lip diagrams captured by the camera when using the trained lip language recognition model, perform an analysis on each lip image.
  • the corresponding lip action feature numbers and letters can be parsed separately, including 9, 8, 5, 2, t, y, r, and d, and according to the generation time of the lip picture, the lips corresponding to the lip picture
  • the movement characteristic numbers/letters are sorted, and the sequence of lip movement characteristic numbers/letters is obtained.
  • Step S208 Extract the interpretation audio of the object to be detected, perform voice recognition on the interpretation audio, and convert the interpretation audio into a voice text display code.
  • Step S210 comparing the lip motion feature display code and the voice text display code with the random display code sequence, and when both the lip motion feature display code and the voice text display code conform to the random display code sequence, live detection of the object to be detected is performed.
  • the lip action feature display code and the voice text display code are compared with the random display code sequence respectively. And only when the lip motion feature display code and the voice text display code are in accordance with the random display code sequence, the next step of live detection is performed. Only the voice and text display code meets the random display code sequence, or only the lip motion feature display code meets the random display code sequence. When displaying the code sequence, you need to return to the corresponding step to re-test.
  • the server needs to regenerate the random display code sequence, and the object to be detected needs to be interpreted again, and the lip motion feature display code and voice text display are repeated Obtain the code and compare it again until the lip movement feature display code and the voice text display code are in line with the corresponding random display code sequence.
  • living body detection uses combined actions such as blinking, opening the mouth, shaking the head, and nodding, using technologies such as facial key point positioning and face tracking to verify whether the object to be detected is a real living body. For example, by instructing the object to be detected to blink, shake the head, and open the mouth, determine whether the object to be detected can perform the corresponding instruction operation. If an error occurs, the object to be detected needs to perform the corresponding operation again according to the instructions, such as repeated 3 times The above error means that the object to be detected cannot perform the instruction operation and cannot pass the live detection.
  • Step S212 When it is confirmed that the object to be detected is a living object, call the face image file of the object to be detected pre-stored in the database, compare with the face detection result, perform identity authentication, and obtain the identity authentication result.
  • the object to be detected passes the living body detection, it means that the object to be detected that is currently logged in to the terminal application is a living body, and the next step of identity authentication can be performed.
  • the face image is detected and compared with the face image file of the object to be detected pre-stored in the database. When the two match, the identity authentication is successful. If the two do not match, a prompt will be issued to re-authenticate.
  • the object to be detected is subjected to face detection to obtain the face detection result, and a random display code sequence is generated according to the face detection result; the random display code sequence is based on random numbers , Random letters or at least one form of random Chinese characters. Due to the use of randomly generated numbers, letters and Chinese character sequences as identifiers, it is difficult to record audio files and video files in advance, which can reduce the risk of being attacked.
  • the corresponding lip motion feature display code is obtained.
  • extract the interpretation audio of the object to be detected perform voice recognition on the interpretation audio, convert the interpretation audio into a voice text display code, and then compare the lip movement feature display code and the voice text display code with the random display code sequence.
  • the object to be detected is subjected to live detection.
  • the face image file of the object to be detected pre-stored in the database is called, and The face detection results are compared, identity authentication is performed, and the identity authentication result is obtained. Due to the combination of face recognition, lip language recognition and voice recognition at the same time, they can complement and improve each other, further enhancing the reliability and security of identity authentication.
  • the pre-trained lip language recognition model before extracting the lip motion features of the object to be detected, and using the pre-trained lip language recognition model to analyze the lip motion features, before obtaining the corresponding lip motion feature display code, it also includes: according to the sample The data trains the deep learning model to obtain the trained lip recognition model.
  • the deep learning model is trained according to the sample data to obtain the trained lip language recognition model.
  • the specific process includes: collecting lip pictures corresponding to each number and each letter, and extracting from the database more than the preset frequency of use. Collect the lip pictures corresponding to multiple Chinese characters, extract the lip motion features on each lip picture, generate sample data, and use the sample data to train the convolutional neural network model to obtain the corresponding lip language recognition model .
  • the lip pictures that need to be collected need to involve different types of objects to be detected, including lip pictures of objects to be detected in different age stages and different regions.
  • sample data is generated by extracting the lip motion features on the collected lip-shaped pictures, and the convolutional neural network model is trained using the sample data to obtain the corresponding lip language recognition model.
  • the collected lip pictures of each number can be used to train the convolutional neural network model to obtain the corresponding lip language recognition model.
  • Tensorflow is a computational graph model, namely A model of the calculation process is expressed in the form of a graph, and the Tensorflow program is generally divided into two stages: the construction of the graph and the execution of the graph. The construction of the graph is also called the definition phase of the graph. The process will be defined in the graph model. The required operation, the result of each operation and the original input data can all be called a node.
  • the face image file of the object to be detected pre-stored in the database is called, and the result is compared with the face detection result to perform identity authentication, and after the identity authentication result is obtained ,Also includes:
  • behavior records include browsing records, favorite records, and purchase records;
  • the basic information of the object to be detected is compared with the pre-stored user information, and when the basic information of the object to be detected matches the pre-stored user information, the behavior record of the object to be detected is verified;
  • the identity authentication performed after the living body test is passed in addition to verifying the face image to be detected, it also needs to obtain the user's basic personal information and behavior habits on the application, including browsing records, favorite records, and purchase records And so on, set up corresponding verification links respectively.
  • users need to perform inquiries, such as inquiring about the balance of a personal account, they need to confirm the user's personal information, and the user needs to fill in the personal real name and account name on the verification interface. If the user needs to perform the transfer service, after confirming the user's name and account number, the user's ID number and mobile phone number must also be verified, and the transfer service can only be performed after all are passed.
  • the user needs to purchase funds and other services, he needs to confirm the user's favorite record and purchase record on the application, and include the user's favorite record or purchase record on the application.
  • a number of different products are displayed for the user who is currently making a purchase operation to choose to determine whether the user making the purchase business is himself. When the selection is correct, it means that the verification is passed and the user can purchase products such as funds.
  • the basic information of the object to be detected is compared with the pre-stored user information.
  • the behavior record of the object to be detected is verified.
  • the behavior record of is verified, it means that the identity authentication is successful, which further guarantees the safety factor of identity authentication.
  • steps in the flowchart of FIG. 2 are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless there is a clear description in this article, there is no strict order for the execution of these steps, and these steps can be executed in other orders. Moreover, at least part of the steps in FIG. 2 may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution of these sub-steps or stages The sequence is not necessarily performed sequentially, but may be performed alternately or alternately with at least a part of other steps or sub-steps or stages of other steps.
  • an identity verification device which includes: a face detection module 302, a random display code sequence generation module 304, a lip motion feature extraction module 306, a speech recognition module 308, and a comparison Pair module 310 and identity authentication module 312, where:
  • the face detection module 302 is configured to trigger a face detection instruction when the login operation of the object to be detected is detected, and perform face detection on the object to be detected according to the face detection instruction to obtain a face detection result.
  • the random display code sequence generating module 304 is used to generate a random display code sequence according to the face detection result; the random display code sequence is composed of any two forms or more of random numbers, random letters or random Chinese characters.
  • the lip action feature extraction module 306 is used to extract lip action features of the object to be detected, and use a pre-trained lip language recognition model to perform feature analysis on the lip action features to obtain a corresponding lip action feature display code.
  • the voice recognition module 308 is used to extract the interpretation audio of the object to be detected, perform voice recognition on the interpretation audio, and convert the interpretation audio into a voice text display code.
  • the comparison module 310 is used to compare the lip motion feature display code and the voice text display code with the random display code sequence. When the lip motion feature display code and the voice text display code both conform to the random display code sequence, the object to be detected is treated Perform a live test.
  • the identity authentication module 312 is configured to, when it is confirmed that the object to be detected is a living object, call the face image file of the object to be detected prestored in the database, compare with the face detection result, perform identity authentication, and obtain the identity authentication result.
  • the above identity verification device detects the face of the object to be detected when the login operation of the object to be detected is detected, obtains the face detection result, and generates a random display code sequence according to the face detection result.
  • the random display code sequence is based on random numbers, It is composed of at least one form of random letters or random Chinese characters. Due to the use of randomly generated numbers, letters and Chinese character sequences as identifiers, it is difficult to record audio files and video files in advance, which can reduce the risk of being attacked.
  • extract the interpretation audio of the object to be detected perform voice recognition on the interpretation audio, convert the interpretation audio into a voice text display code, and then compare the lip movement feature display code and the voice text display code with the random display code sequence.
  • the object to be detected is subjected to live detection.
  • the face image file of the object to be detected pre-stored in the database is called, and The face detection results are compared, identity authentication is performed, and the identity authentication result is obtained. Due to the combination of face recognition, lip language recognition and voice recognition at the same time, they can complement and improve each other, further enhancing the reliability and security of identity authentication.
  • the face detection module is also used to:
  • the login status of the object to be detected is obtained; the login status includes login success and login failure; when it is determined that the login status of the object to be detected is login success, the face detection instruction is triggered; according to the face The detection instruction opens the camera program, and uses the camera program to perform face detection on the object to be detected. When the face part of the object to be detected is detected, a face detection result is generated.
  • the above-mentioned face detection module can determine whether to trigger the face recognition instruction by recognizing the login operation of the object to be detected and judging the login status. When the face part of the object to be detected is detected, the corresponding face recognition result is generated , Improve the efficiency of identity verification.
  • the random display code sequence generation module is also used to:
  • the preset requirements include the length of the random display code sequence and the composition of the random display code sequence; when the face part of the object to be detected is determined according to the face detection result , According to the preset requirements of the random display code sequence, randomly generate a random display code sequence.
  • the above-mentioned random display code sequence generation module sets corresponding random sequence codes according to different application scenarios, which can be read by the object to be detected, avoiding the situation of recording audio files and video files in advance, reducing the risk of being attacked and further improving The security and reliability of identity verification.
  • an identity verification device which further includes a lip language recognition model training module for:
  • Collect lip pictures corresponding to each number and each letter extract multiple Chinese characters that exceed the preset frequency of use from the database, and collect lip pictures corresponding to multiple Chinese characters; extract lip motion features on each lip picture, and generate samples Data; use sample data to train the convolutional neural network model to obtain the corresponding lip language recognition model.
  • the above-mentioned identity verification device collects the lip-shaped pictures corresponding to each number and each letter, extracts multiple Chinese characters beyond the preset frequency of use from the database, and collects the lip-shaped pictures corresponding to the multiple Chinese characters, and extracts the lip-shaped pictures on each lip picture.
  • Lip action features generate sample data, and use the sample data to train the convolutional neural network model to obtain the corresponding lip language recognition model.
  • the obtained lip language recognition model can be applied to the lip motion features of the object to be detected. Improve the safety factor of identity authentication.
  • the lip motion feature extraction module is also used to:
  • the above-mentioned lip motion feature extraction module uses the lip language recognition model to perform feature analysis on lip motion features.
  • the display code corresponding to the lip motion feature sample is used as the lip motion feature display code, Improve the effectiveness of lip recognition and further improve the efficiency of identity verification.
  • an identity verification device is provided, and further includes a secondary authentication module for:
  • behavior records include browsing records, favorite records, and purchase records.
  • the behavior record of the object to be detected is verified; when the behavior record of the object to be detected passes the verification, it means that the identity authentication is successful.
  • the aforementioned identity verification device compares the basic information of the object to be detected with the pre-stored user information. When the basic information of the object to be detected matches the pre-stored user information, the behavior record of the object to be detected is verified. When the behavior record of the object is verified, it indicates that the identity authentication is successful, which further guarantees the safety factor of identity authentication.
  • Each module in the above-mentioned identity verification device can be implemented in whole or in part by software, hardware, and a combination thereof.
  • the above-mentioned modules may be embedded in the form of hardware or independent of the processor in the computer equipment, or may be stored in the memory of the computer equipment in the form of software, so that the processor can call and execute the operations corresponding to the above-mentioned modules.
  • a computer device is provided.
  • the computer device may be a server, and its internal structure diagram may be as shown in FIG. 4.
  • the computer equipment includes a processor, a memory, a network interface, and a database connected through a system bus.
  • the processor of the computer device is used to provide calculation and control capabilities.
  • the memory of the computer device includes a non-volatile or volatile storage medium and internal memory.
  • the non-volatile or volatile storage medium stores an operating system, computer readable instructions, and a database.
  • the internal memory provides an environment for the operation of the operating system and computer-readable instructions in the non-volatile storage medium.
  • the database of the computer equipment is used to store relevant data of the object to be detected.
  • the network interface of the computer device is used to communicate with an external terminal through a network connection.
  • the computer-readable instructions are executed by the processor to implement an identity verification method.
  • FIG. 4 is only a block diagram of part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied.
  • the specific computer device may Including more or fewer parts than shown in the figure, or combining some parts, or having a different arrangement of parts.
  • a computer device includes a memory and one or more processors.
  • the memory stores computer-readable instructions.
  • the one or more processors perform the following steps:
  • the face detection instruction is triggered, and the face detection of the object to be detected is performed according to the face detection instruction to obtain the face detection result;
  • the random display code sequence is composed of at least one form of random numbers, random letters or random Chinese characters;
  • Extract the interpretation audio of the object to be detected perform voice recognition on the interpretation audio, and convert the interpretation audio into a voice text display code
  • the face image file of the object to be detected pre-stored in the database is called, and the result of the face detection is compared, identity authentication is performed, and the identity authentication result is obtained.
  • the processor further implements the following steps when executing the computer-readable instructions:
  • the login status of the object to be detected is obtained; the login status includes login success and login failure;
  • Open the camera program according to the face detection instruction use the camera program to detect the face of the object to be detected, and generate the face detection result when the face part of the object to be detected is detected.
  • the processor further implements the following steps when executing the computer-readable instructions:
  • the preset requirements include the length of the random display code sequence and the composition of the random display code sequence;
  • a random display code sequence is randomly generated according to the preset requirement of the random display code sequence.
  • the processor further implements the following steps when executing the computer-readable instructions:
  • the deep learning model is trained according to the sample data, and the trained lip recognition model is obtained.
  • the processor further implements the following steps when executing the computer-readable instructions:
  • the processor further implements the following steps when executing the computer-readable instructions:
  • lip language recognition model uses the lip language recognition model to perform feature analysis on lip action features, and compare the lip action features with preset lip action feature samples;
  • the display code corresponding to the lip motion feature sample is used as the lip motion feature display code.
  • the processor further implements the following steps when executing the computer-readable instructions:
  • behavior records include browsing records, favorite records, and purchase records;
  • the basic information of the object to be detected is compared with the pre-stored user information, and when the basic information of the object to be detected matches the pre-stored user information, the behavior record of the object to be detected is verified;
  • One or more computer-readable storage media storing computer-readable instructions.
  • the one or more processors perform the following steps:
  • the face detection instruction is triggered, and the face detection of the object to be detected is performed according to the face detection instruction to obtain the face detection result;
  • the random display code sequence is composed of at least one form of random numbers, random letters or random Chinese characters;
  • Extract the interpretation audio of the object to be detected perform voice recognition on the interpretation audio, and convert the interpretation audio into a voice text display code
  • the face image file of the object to be detected pre-stored in the database is called, and the result of face detection is compared to perform identity authentication, and the identity authentication result is obtained.
  • the computer-readable storage medium may be non-volatile or volatile.
  • the login status of the object to be detected is obtained; the login status includes login success and login failure;
  • Open the camera program according to the face detection instruction use the camera program to detect the face of the object to be detected, and generate the face detection result when the face part of the object to be detected is detected.
  • the preset requirements include the length of the random display code sequence and the composition of the random display code sequence;
  • a random display code sequence is randomly generated according to the preset requirement of the random display code sequence.
  • the deep learning model is trained according to the sample data, and the trained lip recognition model is obtained.
  • lip language recognition model uses the lip language recognition model to perform feature analysis on lip action features, and compare the lip action features with preset lip action feature samples;
  • the display code corresponding to the lip motion feature sample is used as the lip motion feature display code.
  • behavior records include browsing records, favorite records, and purchase records;
  • the basic information of the object to be detected is compared with the pre-stored user information, and when the basic information of the object to be detected matches the pre-stored user information, the behavior record of the object to be detected is verified;
  • Non-volatile memory may include read only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory.
  • Volatile memory may include random access memory (RAM) or external cache memory.
  • RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous chain Channel (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Security & Cryptography (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Hardware Design (AREA)
  • Software Systems (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Collating Specific Patterns (AREA)
  • Image Analysis (AREA)

Abstract

一种身份验证方法,包括:对待检测对象进行人脸检测,获得人脸检测结果(S202),根据人脸检测结果生成随机展示码序列(S204);对待检测对象的嘴唇动作特征进行提取,利用唇语识别模型对嘴唇动作特征进行特征解析,获得嘴唇动作特征展示码(S206);提取待检测对象的解读音频,转换成语音文本展示码(S208);将嘴唇动作特征展示码和语音文本展示码与随机展示码序列进行比对,当两者均符合随机展示码序列时,对待检测对象进行活体检测(S210);当通过活体检测时,将待检测对象的脸部图像文件与人脸检测结果进行比对,得到身份认证结果(S212)。

Description

身份验证方法、装置、计算机设备和存储介质
相关申请的交叉引用
本申请要求于2019年10月12日提交中国专利局,申请号为2019109698312,申请名称为“身份验证方法、装置、计算机设备和存储介质”的中国专利申请的优先权,其全部内容通过引用结合在本申请中。
技术领域
本申请涉及人工智能技术领域,特别是涉及一种身份验证方法、装置、计算机设备和存储介质。
背景技术
随着信息技术的日益发展,人们的生活方式也在不断发生变化,而在移动支付和身份验证等领域,如何快速而有效地确认用户的个人身份,并保护用户的个人信息安全,变得尤为重要。
传统的身份认证方案,包括利用密码或者口令来确定个人身份,比如用户使用某一应用程序时,应用程序后台可根据用户输入的登录密码,或者用户忘记密码时输入短信验证码来确认用户身份,还包括利用人脸识别技术来对确认用户身份,通过提取用户的人脸特征,并将预先存储的人脸特征和进行验证时提取的人脸特征进行比对,确认用户身份。
然而,发明人意识到,传统的身份认证方案中,还存在密码或者口令被盗用的问题,以及人脸识别技术中,也存在提前录制包括用户的脸部视频,非法获得用户的头部、眼部以及嘴部运动等,对人脸识别系统进行攻击,导致身份认证安全性太低的问题。
发明内容
根据本申请公开的各种实施例,提供一种身份验证方法、装置、计算机设备和存储介质。
一种身份验证方法,包括:
当检测到待检测对象的登录操作时,触发人脸检测指令,根据所述人脸检测指令对所述待检测对象进行人脸检测,获得人脸检测结果;
根据所述人脸检测结果生成随机展示码序列;所述随机展示码序列根据随机数字、随机字母或随机汉字中至少一种形式组成;
对所述待检测对象的嘴唇动作特征进行提取,并利用预先训练的唇语识别模型对所述嘴唇动作特征进行特征解析,获得对应的嘴唇动作特征展示码;
提取所述待检测对象的解读音频,并对所述解读音频进行语音识别,将所述解读音频 转换成语音文本展示码;
将所述嘴唇动作特征展示码和所述语音文本展示码,与所述随机展示码序列进行比对,当所述嘴唇动作特征展示码和所述语音文本展示码均符合所述随机展示码序列时,对所述待检测对象进行活体检测;及
当确认所述待检测对象为活体对象时,调用数据库中预存的待检测对象的脸部图像文件,与所述人脸检测结果进行比对,进行身份认证,得到身份认证结果。
一种身份验证装置,包括:
人脸检测模块,用于当检测到待检测对象的登录操作时,触发人脸检测指令,根据所述人脸检测指令对所述待检测对象进行人脸检测,获得人脸检测结果;
随机展示码序列生成模块,用于根据所述人脸检测结果生成随机展示码序列;所述随机展示码序列根据随机数字、随机字母或随机汉字中任意两种形式或两种以上的形式组成;
嘴唇动作特征提取模块,用于对所述待检测对象的嘴唇动作特征进行提取,并利用预先训练的唇语识别模型对所述嘴唇动作特征进行特征解析,获得对应的嘴唇动作特征展示码;
语音识别模块,用于提取所述待检测对象的解读音频,并对所述解读音频进行语音识别,将所述解读音频转换成语音文本展示码;
比对模块,用于将所述嘴唇动作特征展示码和所述语音文本展示码,与所述随机展示码序列进行比对,当所述嘴唇动作特征展示码和所述语音文本展示码均符合所述随机展示码序列时,对所述待检测对象进行活体检测;及
身份认证模块,用于当确认所述待检测对象为活体对象时,调用数据库中预存的待检测对象的脸部图像文件,与所述人脸检测结果进行比对,进行身份认证,得到身份认证结果。
一种计算机设备,包括存储器和一个或多个处理器,所述存储器中储存有计算机可读指令,所述计算机可读指令被所述处理器执行时,使得所述一个或多个处理器执行以下步骤:
当检测到待检测对象的登录操作时,触发人脸检测指令,根据所述人脸检测指令对所述待检测对象进行人脸检测,获得人脸检测结果;
根据所述人脸检测结果生成随机展示码序列;所述随机展示码序列根据随机数字、随机字母或随机汉字中至少一种形式组成;
对所述待检测对象的嘴唇动作特征进行提取,并利用预先训练的唇语识别模型对所述嘴唇动作特征进行特征解析,获得对应的嘴唇动作特征展示码;
提取所述待检测对象的解读音频,并对所述解读音频进行语音识别,将所述解读音频转换成语音文本展示码;
将所述嘴唇动作特征展示码和所述语音文本展示码,与所述随机展示码序列进行比 对,当所述嘴唇动作特征展示码和所述语音文本展示码均符合所述随机展示码序列时,对所述待检测对象进行活体检测;及
当确认所述待检测对象为活体对象时,调用数据库中预存的待检测对象的脸部图像文件,与所述人脸检测结果进行比对,进行身份认证,得到身份认证结果。
一个或多个存储有计算机可读指令的计算机可读存储介质,计算机可读指令被一个或多个处理器执行时,使得一个或多个处理器执行以下步骤:
当检测到待检测对象的登录操作时,触发人脸检测指令,根据所述人脸检测指令对所述待检测对象进行人脸检测,获得人脸检测结果;
根据所述人脸检测结果生成随机展示码序列;所述随机展示码序列根据随机数字、随机字母或随机汉字中至少一种形式组成;
对所述待检测对象的嘴唇动作特征进行提取,并利用预先训练的唇语识别模型对所述嘴唇动作特征进行特征解析,获得对应的嘴唇动作特征展示码;
提取所述待检测对象的解读音频,并对所述解读音频进行语音识别,将所述解读音频转换成语音文本展示码;
将所述嘴唇动作特征展示码和所述语音文本展示码,与所述随机展示码序列进行比对,当所述嘴唇动作特征展示码和所述语音文本展示码均符合所述随机展示码序列时,对所述待检测对象进行活体检测;及
当确认所述待检测对象为活体对象时,调用数据库中预存的待检测对象的脸部图像文件,与所述人脸检测结果进行比对,进行身份认证,得到身份认证结果。
上述身份验证方法、装置、计算机设备和存储介质,当检测到待检测对象的登录操作时,对待检测对象进行人脸检测,获得人脸检测结果,并根据人脸检测结果生成随机展示码序列,随机展示码序列根据随机数字、随机字母或随机汉字中至少一种形式组成。由于利用随机生成的数字、字母及汉字序列作为标识,难以被提前录制音频文件和视频文件,可降低被攻击的风险。通过对待检测对象的嘴唇动作特征进行提取,并利用预先训练的唇语识别模型对嘴唇动作特征进行特征解析,获得对应的嘴唇动作特征展示码。同时提取待检测对象的解读音频,并对解读音频进行语音识别,将解读音频转换成语音文本展示码,进而将嘴唇动作特征展示码和语音文本展示码,与随机展示码序列进行比对,当嘴唇动作特征展示码和语音文本展示码均符合随机展示码序列时,对待检测对象进行活体检测,当确认待检测对象为活体对象时,调用数据库中预存的待检测对象的脸部图像文件,与人脸检测结果进行比对,进行身份认证,得到身份认证结果。由于同时结合了人脸识别、唇语识别以及语音识别,可互相补充和完善,进一步增强了身份认证的可靠性和安全性。
本申请的一个或多个实施例的细节在下面的附图和描述中提出。本申请的其它特征和优点将从说明书、附图以及权利要求书变得明显。
附图说明
为了更清楚地说明本申请实施例中的技术方案,下面将对实施例中所需要使用的附图作简单地介绍,显而易见地,下面描述中的附图仅仅是本申请的一些实施例,对于本领域普通技术人员来讲,在不付出创造性劳动的前提下,还可以根据这些附图获得其它的附图。
图1为根据一个或多个实施例中身份验证方法的应用场景图;
图2为根据一个或多个实施例中身份验证方法的流程示意图;
图3为根据一个或多个实施例中身份验证装置的框图;
图4为根据一个或多个实施例中计算机设备的框图。
具体实施方式
为了使本申请的技术方案及优点更加清楚明白,以下结合附图及实施例,对本申请进行进一步详细说明。应当理解,此处描述的具体实施例仅仅用以解释本申请,并不用于限定本申请。
本申请提供的身份验证方法,可以应用于如图1所示的应用环境中。其中,终端102与服务器104通过网络进行通信。当服务器104在终端102检测到待检测对象的登录操作时,触发人脸检测指令,根据人脸检测指令对待检测对象进行人脸检测,获得人脸检测结果,并根据人脸检测结果生成随机展示码序列,其中,随机展示码序列根据随机数字、随机字母或随机汉字中至少一种形式组成。服务器104通过对待检测对象的嘴唇动作特征进行提取,并利用预先训练的唇语识别模型对嘴唇动作特征进行特征解析,获得对应的嘴唇动作特征展示码。在身份验证过程中,同时提取待检测对象的解读音频,并对解读音频进行语音识别,将解读音频转换成语音文本展示码,并将嘴唇动作特征展示码和语音文本展示码,与随机展示码序列进行比对,当嘴唇动作特征展示码和语音文本展示码均符合随机展示码序列时,对待检测对象进行活体检测。当确认待检测对象为活体对象时,调用数据库中预存的待检测对象的脸部图像文件,与人脸检测结果进行比对,进行身份认证,得到身份认证结果。其中,终端102可以但不限于是各种个人计算机、笔记本电脑、智能手机、平板电脑和便携式可穿戴设备,服务器104可以用独立的服务器或者是多个服务器组成的服务器集群来实现。
在其中一个实施例中,如图2所示,提供了一种身份验证方法,以该方法应用于图1中的服务器为例进行说明,包括以下步骤:
步骤S202,当检测到待检测对象的登录操作时,触发人脸检测指令,根据人脸检测指令对待检测对象进行人脸检测,获得人脸检测结果。
具体地,当检测到待检测对象的登录操作时,获取待检测对象的登录状态,其中,登录状态包括登录成功和登录失败。当确定待检测对象的登录状态为登陆成功时,触发人脸检测指令,并根据人脸检测指令打开相机程序,利用相机程序对待检测对象进行人脸检测,当检测到待检测对象的人脸部分时,生成人脸检测结果。
其中,待检测对象在应用程序的登录操作包括待检测对象输入账号密码进行登录,或者以输入账号获取实时验证码并登录的登录操作。针对待检测对象输入账号密码进行登录的操作,在登录成功即账号密码均无错误时,触发人脸检测指令。针对待检测对象输入账号并输入实时获取的验证码进行登录的操作,在待检测对象输入的验证码符合实时获取到的验证码,登录成功时,同样触发人脸检测指令。
进一步地,人脸检测指令用于检测当前执行登录操作的待检测对象的人脸部分,可根据人脸检测指令打开终端上的相机程序,利用相机对人脸进行检测,判断是否检测到待检测对象的人脸部分,得到人脸检测结果。当未检测到人脸时,界面发出“请对准相机”的提示,直到相机检测到待检测对象的人脸部分。
步骤S204,根据人脸检测结果生成随机展示码序列;随机展示码序列根据随机数字、随机字母或随机汉字中至少一种形式组成。
具体地,通过获取当前应用场景下的随机展示码序列的预设要求,其中,预设要求包括随机展示码序列长度,以及随机展示码序列组成,且当根据人脸检测结果,确定检测到待检测对象的人脸部分时,根据随机展示码序列的预设要求,随机生成随机展示码序列。
其中,随机展示码序列可以由随机数字、随机字母以及随机汉字中的至少一种形式等组成,其中,随机数字序列是根据0至9包括的数字生成,包括的数字个数大于等于6位,包括但不限于6位、8位以及10位等。比如6位的随机数字序列可以是156871、589624以及896547等,8位的随机数字序列可以是85412369以及78452369等,数字序列生成不遵循某种规律,随机生成,不同于提前存储数字序列,在需要时进行调用的情况,可避免数字序列被盗用的问题。
同样地,随机字母序列是根据26个英文字母生成,序列位数也不进行限定,大于6位,包括但不限于6位、8位以及10位等。比如,6位的随机字母序列可以是oputrd、ruighd以及swertg等,8位和10位的随机字母序列也是相同情况。
进一步地,在安全级别要求较高的场景,比如办理银行业务等情况下,生成的随机序列长度比较长,并且序列种类多,其中,可将随机序列长度设置成8位或者10位,序列种类可以设置同时包括数字、字母及汉字中的两种或两种以上。比如,同时包括数字和字母的8位随机展示码序列可以是soer5648、9867ogsi、ru98yt03、75un09eg、wr8750jt以及36thfb68等多种形式。同样地,同时包括数字、字母以及汉字的10位随机展示码序列可以是:“49pr如果06et”、“ty开始73rb05”以及“更新34wh96rs”等多种形式。
步骤S206,对待检测对象的嘴唇动作特征进行提取,并利用预先训练的唇语识别模型对嘴唇动作特征进行特征解析,获得对应的嘴唇动作特征展示码。
具体地,通过对待检测对象的嘴唇动作特征进行提取,并将嘴唇动作特征输入预先训练的唇语识别模型,并利用唇语识别模型对嘴唇动作特征进行特征解析,将嘴唇动作特征与预设的嘴唇动作特征样本进行比对。当嘴唇动作特征符合嘴唇动作特征样本一致时,将嘴唇动作特征样本对应的展示码,作为嘴唇动作特征展示码。
进一步地,服务器在待检测对象的解读过程中,实时对待检测对象的嘴唇动作特征进行提取,可通过利用相机拍摄整个解读过程中待检测对象的唇形图片。利用预先训练好的唇语识别模型,对待检测对象的唇形图片进行识别和解析,得到与唇形对应的嘴唇动作特征展示码。比如,当随机生成的展示码序列为6位的随机数字序列569841,利用相机拍摄到多个唇形图片,当利用训练好的唇语识别模型,对每一张唇形图片进行识别解析时,可分别解析得到对应的嘴唇动作特征数字,可以是8、9、5、6、1以及4,按照唇形图片的生成时间,对与唇形图片对应的嘴唇动作特征数字进行排序,得到嘴唇动作特征数字序列。
同样地,针对随机生成的展示码序列为8位的数字字母组合序列9852tyrd时,利用相机拍摄到的多个唇形图,当利用训练好的唇语识别模型,对每一张唇形图片进行识别解析时,可分别解析得到对应得嘴唇动作特征数字和字母,包括9、8、5、2、t、y、r以及d,并按照唇形图片的生成时间,与唇形图片对应的嘴唇动作特征数字/字母进行排序,得到嘴唇动作特征数字/字母序列。
步骤S208,提取待检测对象的解读音频,并对解读音频进行语音识别,将解读音频转换成语音文本展示码。
步骤S210,将嘴唇动作特征展示码和语音文本展示码,与随机展示码序列进行比对,当嘴唇动作特征展示码和语音文本展示码均符合随机展示码序列时,对待检测对象进行活体检测。
具体地,通过将嘴唇动作特征展示码和语音文本展示码,分别与随机展示码序列进行比对。且只有当嘴唇动作特征展示码以及语音文本展示码,均符合随机展示码序列时,才进行下一步的活体检测,仅语音文本展示码符合随机展示码序列,或仅嘴唇动作特征展示码符合随机展示码序列时,需要返回对应步骤进行重新检测。比如,当嘴唇动作特征展示码,或语音文本展示码不符合随机展示码序列时,服务器需要重新生成随机展示码序列,待检测对象需要再次进行解读,重复执行嘴唇动作特征展示码和语音文本展示码的获取,并再次比对,直到嘴唇动作特征展示码和语音文本展示码,均符合相应的随机展示码序列。
其中,活体检测是通过眨眼、张嘴、摇头、点头等组合动作,使用人脸关键点定位和人脸追踪等技术,验证待检测对象是否为真实活体本人操作。比如,通过指示待检测对象进行眨眼、摇头以及张嘴的操作,判断待检测对象是否能执行对应的指示操作,如出现错误,则需要待检测对象重新按照指示执行相应的操作,如重复出现3次以上的错误,则表示该待检测对象无法执行指示操作,不能通过活体检测。
步骤S212,当确认待检测对象为活体对象时,调用数据库中预存的待检测对象的脸部图像文件,与人脸检测结果进行比对,进行身份认证,得到身份认证结果。
具体地,当待检测对象通过活体检测时,表示当前在终端的应用程序进行登录操作的待检测对象,为活体,可进行下一步的身份认证,通过将解读过程中存储的待检测对象的待检测人脸图像,与数据库中预存的待检测对象的脸部图像文件进行比对,两者符合时, 则表示身份认证成功。如两者不相符合,则发出需要重新进行身份认证的提示。
上述身份验证方法中,当检测到待检测对象的登录操作时,对待检测对象进行人脸检测,获得人脸检测结果,并根据人脸检测结果生成随机展示码序列;随机展示码序列根据随机数字、随机字母或随机汉字中至少一种形式组成。由于利用随机生成的数字、字母及汉字序列作为标识,难以被提前录制音频文件和视频文件,可降低被攻击的风险。通过对待检测对象的嘴唇动作特征进行提取,并利用预先训练的唇语识别模型对嘴唇动作特征进行特征解析,获得对应的嘴唇动作特征展示码。同时提取待检测对象的解读音频,并对解读音频进行语音识别,将解读音频转换成语音文本展示码,进而将嘴唇动作特征展示码和语音文本展示码,与随机展示码序列进行比对,当嘴唇动作特征展示码和语音文本展示码均符合随机展示码序列时,对待检测对象进行活体检测,当确认待检测对象为活体对象时,调用数据库中预存的待检测对象的脸部图像文件,与人脸检测结果进行比对,进行身份认证,得到身份认证结果。由于同时结合了人脸识别、唇语识别以及语音识别,可互相补充和完善,进一步增强了身份认证的可靠性和安全性。
在其中一个实施例中,在对待检测对象的嘴唇动作特征进行提取,并利用预先训练的唇语识别模型对嘴唇动作特征进行特征解析,获得对应的嘴唇动作特征展示码之前,还包括:根据样本数据对深度学习模型进行训练,获得训练后的唇语识别模型。
具体地,根据样本数据对深度学习模型进行训练,获得训练后的唇语识别模型的具体过程,包括:收集各数字以及各字母对应的唇形图片,从数据库中提取超出预设使用频率的多个汉字,并收集多个汉字对应的唇形图片,提取各唇形图片上的嘴唇动作特征,生成样本数据,并利用样本数据,对卷积神经网络模型进行训练,得到对应的唇语识别模型。
其中,通过收集数字0至9对应的唇形图片,以及26个英文字母对应的唇形图片,从数据库中收集超过预设使用频率的多个汉字,比如以一天作为周期,获取一天内使用频率超过预设频率的多个汉字,并获取该些汉字对应的唇形图片。其中,需要收集的唇形图片需要涉及不同类型的待检测对象,包括不同年龄阶段以及不同地区的待检测对象的唇形图片。
进一步地,通过提取所收集的各唇形图片上的嘴唇动作特征,生成样本数据,并利用样本数据对卷积神经网络模型进行训练,得到对应的唇语识别模型。其中,在本实施例中,可基于Tensorflow平台,利用所收集的各数字的唇形图片,对卷积神经网络模型进行训练,得到对应的唇语识别模型,Tensorflow是一种计算图模型,即用图的形式来表示运算过程的一种模型,而Tensorflow程序一般分为图的构建和图的执行两个阶段,图的构建阶段也称为图的定义阶段,该过程会在图模型中定义所需的运算,每次运算的结果以及原始的输入数据都可称为一个节点。
上述步骤中,通过收集各数字以及各字母对应的唇形图片,从数据库中提取超出预设使用频率的多个汉字,并收集多个汉字对应的唇形图片,提取各唇形图片上的嘴唇动作特征,生成样本数据,并利用样本数据,对卷积神经网络模型进行训练,得到对应的唇语 识别模型,可将得到的唇语识别模型应用于对待检测对象的嘴唇动作特征进行识别,提高身份认证的安全系数。
在其中一个实施例中,在当确认待检测对象为活体对象时,调用数据库中预存的待检测对象的脸部图像文件,与人脸检测结果进行比对,进行身份认证,得到身份认证结果之后,还包括:
获取待检测对象的基本信息和行为记录;行为记录包括浏览记录、收藏记录以及购买记录;
将待检测对象的基本信息和预先存储的用户信息进行比对,当待检测对象的基本信息符合预先存储的用户信息时,对待检测对象的行为记录进行验证;
当待检测对象的行为记录通过验证时,表示身份认证成功。
具体地,在活体检测通过后进行的身份认证,除需对待检测人脸图像进行验证,还需获取用户的基本个人信息以及在该应用程序上的行为习惯,包括浏览记录,收藏记录以及购买记录等,分别设置相应的验证环节。比如,针对银行业务的办理,用户需要进行查询业务,比如查询个人账号余额,则需要对用户的个人信息进行确认,用户需要在验证界面填写个人真实姓名以及账号名等。如用户需要进行转账业务,则在进行用户姓名以及账号确认后的基础上,还需对用户身份证号和手机号码进行验证,均通过之后才可进行转账业务。另一种情况下,如用户需要进行基金购买等业务,则需要对用户在该应用程序上的收藏记录以及购买记录等进行确认,将包括用户在应用程序上的收藏记录或购买记录在内的多个不同产品进行展示,供当前进行购买操作的用户进行选择,以判断进行购买业务的用户是否为本人,当选择无误时,则表示通过验证,用户可进行对基金等产品的购买操作。
上述步骤中,通过将待检测对象的基本信息和预先存储的用户信息进行比对,当待检测对象的基本信息符合预先存储的用户信息时,对待检测对象的行为记录进行验证,当待检测对象的行为记录通过验证时,表示身份认证成功,进一步保证了身份认证的安全系数。
应该理解的是,虽然图2的流程图中的各个步骤按照箭头的指示依次显示,但是这些步骤并不是必然按照箭头指示的顺序依次执行。除非本文中有明确的说明,这些步骤的执行并没有严格的顺序限制,这些步骤可以以其它的顺序执行。而且,图2中的至少一部分步骤可以包括多个子步骤或者多个阶段,这些子步骤或者阶段并不必然是在同一时刻执行完成,而是可以在不同的时刻执行,这些子步骤或者阶段的执行顺序也不必然是依次进行,而是可以与其它步骤或者其它步骤的子步骤或者阶段的至少一部分轮流或者交替地执行。
在其中一个实施例中,如图3所示,提供了一种身份验证装置,包括:人脸检测模块302、随机展示码序列生成模块304、嘴唇动作特征提取模块306、语音识别模块308、比对模块310以及身份认证模块312,其中:
人脸检测模块302,用于当检测到待检测对象的登录操作时,触发人脸检测指令,根 据人脸检测指令对待检测对象进行人脸检测,获得人脸检测结果。
随机展示码序列生成模块304,用于根据人脸检测结果生成随机展示码序列;随机展示码序列根据随机数字、随机字母或随机汉字中任意两种形式或两种以上的形式组成。
嘴唇动作特征提取模块306,用于对待检测对象的嘴唇动作特征进行提取,并利用预先训练的唇语识别模型对嘴唇动作特征进行特征解析,获得对应的嘴唇动作特征展示码。
语音识别模块308,用于提取待检测对象的解读音频,并对解读音频进行语音识别,将解读音频转换成语音文本展示码。
比对模块310,用于将嘴唇动作特征展示码和语音文本展示码,与随机展示码序列进行比对,当嘴唇动作特征展示码和语音文本展示码均符合随机展示码序列时,对待检测对象进行活体检测.
身份认证模块312,用于当确认待检测对象为活体对象时,调用数据库中预存的待检测对象的脸部图像文件,与人脸检测结果进行比对,进行身份认证,得到身份认证结果。
上述身份验证装置,当检测到待检测对象的登录操作时,对待检测对象进行人脸检测,获得人脸检测结果,并根据人脸检测结果生成随机展示码序列,随机展示码序列根据随机数字、随机字母或随机汉字中至少一种形式组成。由于利用随机生成的数字、字母及汉字序列作为标识,难以被提前录制音频文件和视频文件,可降低被攻击的风险。通过对待检测对象的嘴唇动作特征进行提取,并利用预先训练的唇语识别模型对嘴唇动作特征进行特征解析,获得对应的嘴唇动作特征展示码。同时提取待检测对象的解读音频,并对解读音频进行语音识别,将解读音频转换成语音文本展示码,进而将嘴唇动作特征展示码和语音文本展示码,与随机展示码序列进行比对,当嘴唇动作特征展示码和语音文本展示码均符合随机展示码序列时,对待检测对象进行活体检测,当确认待检测对象为活体对象时,调用数据库中预存的待检测对象的脸部图像文件,与人脸检测结果进行比对,进行身份认证,得到身份认证结果。由于同时结合了人脸识别、唇语识别以及语音识别,可互相补充和完善,进一步增强了身份认证的可靠性和安全性。
在其中一个实施例中,人脸检测模块还用于:
当检测到待检测对象的登录操作时,获取待检测对象的登录状态;登录状态包括登录成功和登录失败;当确定待检测对象的登录状态为登陆成功时,触发人脸检测指令;根据人脸检测指令打开相机程序,利用相机程序对待检测对象进行人脸检测,当检测到待检测对象的人脸部分时,生成人脸检测结果。
上述人脸检测模块,可通过对待检测对象的登录操作进行识别以及登录状态的判断,来确定是否触发人脸识别指令,当检测到待检测对象的人脸部分时,生成相应的人脸识别结果,提高了身份验证的效率。
在其中一个实施例中,随机展示码序列生成模块还用于:
获取当前应用场景下的随机展示码序列的预设要求;预设要求包括随机展示码序列长度,以及随机展示码序列组成;当根据人脸检测结果,确定检测到待检测对象的人脸部 分时,根据随机展示码序列的预设要求,随机生成随机展示码序列。
上述随机展示码序列生成模块,根据不同应用场景设置对应的随机序列码,可供待检测对象进行读取,避免出现被提前录制音频文件和视频文件的情况,可降低被攻击的风险,进一步提高身份验证的安全性和可靠性。
在其中一个实施例中,提供了一种身份验证装置,还包括唇语识别模型训练模块,用于:
收集各数字以及各字母对应的唇形图片;从数据库中提取超出预设使用频率的多个汉字,并收集多个汉字对应的唇形图片;提取各唇形图片上的嘴唇动作特征,生成样本数据;利用样本数据,对卷积神经网络模型进行训练,得到对应的唇语识别模型。
上述身份验证装置,通过收集各数字以及各字母对应的唇形图片,从数据库中提取超出预设使用频率的多个汉字,并收集多个汉字对应的唇形图片,提取各唇形图片上的嘴唇动作特征,生成样本数据,并利用样本数据,对卷积神经网络模型进行训练,得到对应的唇语识别模型,可将得到的唇语识别模型应用于对待检测对象的嘴唇动作特征进行识别,提高身份认证的安全系数。
在其中一个实施例中,嘴唇动作特征提取模块还用于:
对待检测对象的嘴唇动作特征进行提取,并将嘴唇动作特征输入预先训练的唇语识别模型;利用唇语识别模型对嘴唇动作特征进行特征解析,将嘴唇动作特征与预设的嘴唇动作特征样本进行比对;当嘴唇动作特征符合嘴唇动作特征样本时,将嘴唇动作特征样本对应的展示码,作为嘴唇动作特征展示码。
上述嘴唇动作特征提取模块,通过利用唇语识别模型对嘴唇动作特征进行特征解析,当确定嘴唇动作特征符合嘴唇动作特征样本时,将唇动作特征样本对应的展示码,作为嘴唇动作特征展示码,提高唇语识别的有效性,进一步提高身份验证的效率。
在其中一个实施例中,提供了一种身份验证装置,还包括二次认证模块,用于:
获取待检测对象的基本信息和行为记录;行为记录包括浏览记录、收藏记录以及购买记录将待检测对象的基本信息和预先存储的用户信息进行比对,当待检测对象的基本信息符合预先存储的用户信息时,对待检测对象的行为记录进行验证;当待检测对象的行为记录通过验证时,表示身份认证成功。
上述身份验证装置,通过将待检测对象的基本信息和预先存储的用户信息进行比对,当待检测对象的基本信息符合预先存储的用户信息时,对待检测对象的行为记录进行验证,当待检测对象的行为记录通过验证时,表示身份认证成功,进一步保证了身份认证的安全系数。
关于身份验证装置的具体限定可以参见上文中对于身份验证方法的限定,在此不再赘述。上述身份验证装置中的各个模块可全部或部分通过软件、硬件及其组合来实现。上述各模块可以硬件形式内嵌于或独立于计算机设备中的处理器中,也可以以软件形式存储于计算机设备中的存储器中,以便于处理器调用执行以上各个模块对应的操作。
在一个实施例中,提供了一种计算机设备,该计算机设备可以是服务器,其内部结构图可以如图4所示。该计算机设备包括通过系统总线连接的处理器、存储器、网络接口和数据库。其中,该计算机设备的处理器用于提供计算和控制能力。该计算机设备的存储器包括非易失性或易失性存储介质、内存储器。该非易失性或易失性存储介质存储有操作系统、计算机可读指令和数据库。该内存储器为非易失性存储介质中的操作系统和计算机可读指令的运行提供环境。该计算机设备的数据库用于存储待检测对象相关数据。该计算机设备的网络接口用于与外部的终端通过网络连接通信。该计算机可读指令被处理器执行时以实现一种身份验证方法。
本领域技术人员可以理解,图4中示出的结构,仅仅是与本申请方案相关的部分结构的框图,并不构成对本申请方案所应用于其上的计算机设备的限定,具体的计算机设备可以包括比图中所示更多或更少的部件,或者组合某些部件,或者具有不同的部件布置。
一种计算机设备,包括存储器和一个或多个处理器,存储器中储存有计算机可读指令,计算机可读指令被处理器执行时,使得一个或多个处理器执行以下步骤:
当检测到待检测对象的登录操作时,触发人脸检测指令,根据人脸检测指令对待检测对象进行人脸检测,获得人脸检测结果;
根据人脸检测结果生成随机展示码序列;随机展示码序列根据随机数字、随机字母或随机汉字中至少一种形式组成;
对待检测对象的嘴唇动作特征进行提取,并利用预先训练的唇语识别模型对嘴唇动作特征进行特征解析,获得对应的嘴唇动作特征展示码;
提取待检测对象的解读音频,并对解读音频进行语音识别,将解读音频转换成语音文本展示码;
将嘴唇动作特征展示码和语音文本展示码,与随机展示码序列进行比对,当嘴唇动作特征展示码和语音文本展示码均符合随机展示码序列时,对待检测对象进行活体检测;及
当确认待检测对象为活体对象时,调用数据库中预存的待检测对象的脸部图像文件,与人脸检测结果进行比对,进行身份认证,得到身份认证结果。
在一个实施例中,处理器执行计算机可读指令时还实现以下步骤:
当检测到待检测对象的登录操作时,获取待检测对象的登录状态;登录状态包括登录成功和登录失败;
当确定待检测对象的登录状态为登陆成功时,触发人脸检测指令;及
根据人脸检测指令打开相机程序,利用相机程序对待检测对象进行人脸检测,当检测到待检测对象的人脸部分时,生成人脸检测结果。
在一个实施例中,处理器执行计算机可读指令时还实现以下步骤:
获取当前应用场景下的随机展示码序列的预设要求;预设要求包括随机展示码序列长度,以及随机展示码序列组成;及
当根据人脸检测结果,确定检测到待检测对象的人脸部分时,根据随机展示码序列的预设要求,随机生成随机展示码序列。
在一个实施例中,处理器执行计算机可读指令时还实现以下步骤:
根据样本数据对深度学习模型进行训练,获得训练后的唇语识别模型。
在一个实施例中,处理器执行计算机可读指令时还实现以下步骤:
收集各数字以及各字母对应的唇形图片;
从数据库中提取超出预设使用频率的多个汉字,并收集多个汉字对应的唇形图片;
提取各唇形图片上的嘴唇动作特征,生成样本数据;及
利用样本数据,对卷积神经网络模型进行训练,得到对应的唇语识别模型。
在一个实施例中,处理器执行计算机可读指令时还实现以下步骤:
对待检测对象的嘴唇动作特征进行提取,并将嘴唇动作特征输入预先训练的唇语识别模型;
利用唇语识别模型对嘴唇动作特征进行特征解析,将嘴唇动作特征与预设的嘴唇动作特征样本进行比对;及
当嘴唇动作特征符合嘴唇动作特征样本时,将嘴唇动作特征样本对应的展示码,作为嘴唇动作特征展示码。
在一个实施例中,处理器执行计算机可读指令时还实现以下步骤:
获取待检测对象的基本信息和行为记录;行为记录包括浏览记录、收藏记录以及购买记录;
将待检测对象的基本信息和预先存储的用户信息进行比对,当待检测对象的基本信息符合预先存储的用户信息时,对待检测对象的行为记录进行验证;及
当待检测对象的行为记录通过验证时,表示身份认证成功。
一个或多个存储有计算机可读指令的计算机可读存储介质,计算机可读指令被一个或多个处理器执行时,使得一个或多个处理器执行以下步骤:
当检测到待检测对象的登录操作时,触发人脸检测指令,根据人脸检测指令对待检测对象进行人脸检测,获得人脸检测结果;
根据人脸检测结果生成随机展示码序列;随机展示码序列根据随机数字、随机字母或随机汉字中至少一种形式组成;
对待检测对象的嘴唇动作特征进行提取,并利用预先训练的唇语识别模型对嘴唇动作特征进行特征解析,获得对应的嘴唇动作特征展示码;
提取待检测对象的解读音频,并对解读音频进行语音识别,将解读音频转换成语音文本展示码;
将嘴唇动作特征展示码和语音文本展示码,与随机展示码序列进行比对,当嘴唇动作特征展示码和语音文本展示码均符合随机展示码序列时,对待检测对象进行活体检测;及
当确认待检测对象为活体对象时,调用数据库中预存的待检测对象的脸部图像文件, 与人脸检测结果进行比对,进行身份认证,得到身份认证结果。
其中,该计算机可读存储介质可以是非易失性,也可以是易失性的。
在一个实施例中,计算机可读指令被处理器执行时还实现以下步骤:
当检测到待检测对象的登录操作时,获取待检测对象的登录状态;登录状态包括登录成功和登录失败;
当确定待检测对象的登录状态为登陆成功时,触发人脸检测指令;及
根据人脸检测指令打开相机程序,利用相机程序对待检测对象进行人脸检测,当检测到待检测对象的人脸部分时,生成人脸检测结果。
在一个实施例中,计算机可读指令被处理器执行时还实现以下步骤:
获取当前应用场景下的随机展示码序列的预设要求;预设要求包括随机展示码序列长度,以及随机展示码序列组成;及
当根据人脸检测结果,确定检测到待检测对象的人脸部分时,根据随机展示码序列的预设要求,随机生成随机展示码序列。
在一个实施例中,计算机可读指令被处理器执行时还实现以下步骤:
根据样本数据对深度学习模型进行训练,获得训练后的唇语识别模型。
在一个实施例中,计算机可读指令被处理器执行时还实现以下步骤:
收集各数字以及各字母对应的唇形图片;
从数据库中提取超出预设使用频率的多个汉字,并收集多个汉字对应的唇形图片;
提取各唇形图片上的嘴唇动作特征,生成样本数据;及
利用样本数据,对卷积神经网络模型进行训练,得到对应的唇语识别模型。
在一个实施例中,计算机可读指令被处理器执行时还实现以下步骤:
对待检测对象的嘴唇动作特征进行提取,并将嘴唇动作特征输入预先训练的唇语识别模型;
利用唇语识别模型对嘴唇动作特征进行特征解析,将嘴唇动作特征与预设的嘴唇动作特征样本进行比对;及
当嘴唇动作特征符合嘴唇动作特征样本时,将嘴唇动作特征样本对应的展示码,作为嘴唇动作特征展示码。
在一个实施例中,计算机可读指令被处理器执行时还实现以下步骤:
获取待检测对象的基本信息和行为记录;行为记录包括浏览记录、收藏记录以及购买记录;
将待检测对象的基本信息和预先存储的用户信息进行比对,当待检测对象的基本信息符合预先存储的用户信息时,对待检测对象的行为记录进行验证;及
当待检测对象的行为记录通过验证时,表示身份认证成功。
本领域普通技术人员可以理解实现上述实施例方法中的全部或部分流程,是可以通过计算机可读指令来指令相关的硬件来完成,所述的计算机可读指令可存储于一计算机可 读取存储介质中,该计算机可读指令在执行时,可包括如上述各方法的实施例的流程。其中,本申请所提供的各实施例中所使用的对存储器、存储、数据库或其它介质的任何引用,均可包括非易失性和/或易失性存储器。非易失性存储器可包括只读存储器(ROM)、可编程ROM(PROM)、电可编程ROM(EPROM)、电可擦除可编程ROM(EEPROM)或闪存。易失性存储器可包括随机存取存储器(RAM)或者外部高速缓冲存储器。作为说明而非局限,RAM以多种形式可得,诸如静态RAM(SRAM)、动态RAM(DRAM)、同步DRAM(SDRAM)、双数据率SDRAM(DDRSDRAM)、增强型SDRAM(ESDRAM)、同步链路(Synchlink)DRAM(SLDRAM)、存储器总线(Rambus)直接RAM(RDRAM)、直接存储器总线动态RAM(DRDRAM)、以及存储器总线动态RAM(RDRAM)等。
以上实施例的各技术特征可以进行任意的组合,为使描述简洁,未对上述实施例中的各个技术特征所有可能的组合都进行描述,然而,只要这些技术特征的组合不存在矛盾,都应当认为是本说明书记载的范围。
以上所述实施例仅表达了本申请的几种实施方式,其描述较为具体和详细,但并不能因此而理解为对发明专利范围的限制。应当指出的是,对于本领域的普通技术人员来说,在不脱离本申请构思的前提下,还可以做出若干变形和改进,这些都属于本申请的保护范围。因此,本申请专利的保护范围应以所附权利要求为准。

Claims (20)

  1. 一种身份验证方法,包括:
    当检测到待检测对象的登录操作时,触发人脸检测指令,根据所述人脸检测指令对所述待检测对象进行人脸检测,获得人脸检测结果;
    根据所述人脸检测结果生成随机展示码序列;所述随机展示码序列根据随机数字、随机字母或随机汉字中至少一种形式组成;
    对所述待检测对象的嘴唇动作特征进行提取,并利用预先训练的唇语识别模型对所述嘴唇动作特征进行特征解析,获得对应的嘴唇动作特征展示码;
    提取所述待检测对象的解读音频,并对所述解读音频进行语音识别,将所述解读音频转换成语音文本展示码;
    将所述嘴唇动作特征展示码和所述语音文本展示码,与所述随机展示码序列进行比对,当所述嘴唇动作特征展示码和所述语音文本展示码均符合所述随机展示码序列时,对所述待检测对象进行活体检测;及
    当确认所述待检测对象为活体对象时,调用数据库中预存的待检测对象的脸部图像文件,与所述人脸检测结果进行比对,进行身份认证,得到身份认证结果。
  2. 根据权利要求1所述的方法,其中,所述当检测到待检测对象的登录操作时,触发人脸检测指令,根据所述人脸检测指令对所述待检测对象进行人脸检测,获得人脸检测结果,包括:
    当检测到所述待检测对象的登录操作时,获取所述待检测对象的登录状态;所述登录状态包括登录成功和登录失败;
    当确定所述待检测对象的登录状态为登陆成功时,触发人脸检测指令;及
    根据所述人脸检测指令打开相机程序,利用所述相机程序对所述待检测对象进行人脸检测,当检测到所述待检测对象的人脸部分时,生成人脸检测结果。
  3. 根据权利要求1所述的方法,其中,所述根据所述人脸检测结果生成随机展示码序列,包括:
    获取当前应用场景下的所述随机展示码序列的预设要求;所述预设要求包括所述随机展示码序列长度,以及所述随机展示码序列组成;及
    当根据所述人脸检测结果,确定检测到所述待检测对象的人脸部分时,根据所述随机展示码序列的预设要求,随机生成所述随机展示码序列。
  4. 根据权利要求1所述的方法,其中,在所述对所述待检测对象的嘴唇动作特征进行提取,并利用预先训练的唇语识别模型对所述嘴唇动作特征进行特征解析,获得对应的嘴唇动作特征展示码之前,还包括:根据样本数据对深度学习模型进行训练,获得训练后的唇语识别模型。
  5. 根据权利要求4所述的方法,其中,所述根据样本数据对深度学习模型进行训练,获得训练后的唇语识别模型,包括;
    收集各数字以及各字母对应的唇形图片;
    从数据库中提取超出预设使用频率的多个汉字,并收集多个所述汉字对应的唇形图片;
    提取各所述唇形图片上的嘴唇动作特征,生成样本数据;及
    利用所述样本数据,对卷积神经网络模型进行训练,得到对应的唇语识别模型。
  6. 根据权利要求5所述的方法,其中,所述对所述待检测对象的嘴唇动作特征进行提取,并利用预先训练的唇语识别模型对所述嘴唇动作特征进行特征解析,获得对应的嘴唇动作特征展示码,包括:
    对所述待检测对象的嘴唇动作特征进行提取,并将所述嘴唇动作特征输入所述预先训练的唇语识别模型;
    利用所述唇语识别模型对所述嘴唇动作特征进行特征解析,将所述嘴唇动作特征与预设的嘴唇动作特征样本进行比对;及
    当所述嘴唇动作特征符合所述嘴唇动作特征样本时,将所述嘴唇动作特征样本对应的展示码,作为所述嘴唇动作特征展示码。
  7. 根据权利要求1所述的方法,其中,在所述当确认所述待检测对象为活体对象时,调用数据库中预存的待检测对象的脸部图像文件,与所述人脸检测结果进行比对,进行身份认证,得到身份认证结果之后,还包括:
    获取所述待检测对象的基本信息和行为记录;所述行为记录包括浏览记录、收藏记录以及购买记录;
    将所述待检测对象的基本信息和预先存储的用户信息进行比对,当所述待检测对象的基本信息符合所述预先存储的用户信息时,对所述待检测对象的行为记录进行验证;及
    当所述待检测对象的行为记录通过验证时,表示所述身份认证成功。
  8. 一种身份验证装置,包括:
    人脸检测模块,用于当检测到待检测对象的登录操作时,触发人脸检测指令,根据所述人脸检测指令对所述待检测对象进行人脸检测,获得人脸检测结果;
    随机展示码序列生成模块,用于根据所述人脸检测结果生成随机展示码序列;所述随机展示码序列根据随机数字、随机字母或随机汉字中任意两种形式或两种以上的形式组成;
    嘴唇动作特征提取模块,用于对所述待检测对象的嘴唇动作特征进行提取,并利用预先训练的唇语识别模型对所述嘴唇动作特征进行特征解析,获得对应的嘴唇动作特征展示码;
    语音识别模块,用于提取所述待检测对象的解读音频,并对所述解读音频进行语音识别,将所述解读音频转换成语音文本展示码;
    比对模块,用于将所述嘴唇动作特征展示码和所述语音文本展示码,与所述随机展示码序列进行比对,当所述嘴唇动作特征展示码和所述语音文本展示码均符合所述随机展示 码序列时,对所述待检测对象进行活体检测;及
    身份认证模块,用于当确认所述待检测对象为活体对象时,调用数据库中预存的待检测对象的脸部图像文件,与所述人脸检测结果进行比对,进行身份认证,得到身份认证结果。
  9. 一种计算机设备,包括存储器及一个或多个处理器,所述存储器中储存有计算机可读指令,所述计算机可读指令被所述一个或多个处理器执行时,使得所述一个或多个处理器执行以下步骤:
    当检测到待检测对象的登录操作时,触发人脸检测指令,根据所述人脸检测指令对所述待检测对象进行人脸检测,获得人脸检测结果;
    根据所述人脸检测结果生成随机展示码序列;所述随机展示码序列根据随机数字、随机字母或随机汉字中至少一种形式组成;
    对所述待检测对象的嘴唇动作特征进行提取,并利用预先训练的唇语识别模型对所述嘴唇动作特征进行特征解析,获得对应的嘴唇动作特征展示码;
    提取所述待检测对象的解读音频,并对所述解读音频进行语音识别,将所述解读音频转换成语音文本展示码;
    将所述嘴唇动作特征展示码和所述语音文本展示码,与所述随机展示码序列进行比对,当所述嘴唇动作特征展示码和所述语音文本展示码均符合所述随机展示码序列时,对所述待检测对象进行活体检测;及
    当确认所述待检测对象为活体对象时,调用数据库中预存的待检测对象的脸部图像文件,与所述人脸检测结果进行比对,进行身份认证,得到身份认证结果。
  10. 根据权利要求9所述的计算机设备,其中,所述处理器执行所述计算机可读指令时还执行以下步骤:
    当检测到所述待检测对象的登录操作时,获取所述待检测对象的登录状态;所述登录状态包括登录成功和登录失败;
    当确定所述待检测对象的登录状态为登陆成功时,触发人脸检测指令;及
    根据所述人脸检测指令打开相机程序,利用所述相机程序对所述待检测对象进行人脸检测,当检测到所述待检测对象的人脸部分时,生成人脸检测结果。
  11. 根据权利要求9所述的计算机设备,其中,所述处理器执行所述计算机可读指令时还执行以下步骤:
    获取当前应用场景下的所述随机展示码序列的预设要求;所述预设要求包括所述随机展示码序列长度,以及所述随机展示码序列组成;及
    当根据所述人脸检测结果,确定检测到所述待检测对象的人脸部分时,根据所述随机展示码序列的预设要求,随机生成所述随机展示码序列。
  12. 根据权利要求9所述的计算机设备,其中,所述处理器执行所述计算机可读指令时还执行以下步骤;
    根据样本数据对深度学习模型进行训练,获得训练后的唇语识别模型。
  13. 根据权利要求12所述的计算机设备,其中,所述处理器执行所述计算机可读指令时还执行以下步骤:
    收集各数字以及各字母对应的唇形图片;
    从数据库中提取超出预设使用频率的多个汉字,并收集多个所述汉字对应的唇形图片;
    提取各所述唇形图片上的嘴唇动作特征,生成样本数据;及
    利用所述样本数据,对卷积神经网络模型进行训练,得到对应的唇语识别模型。
  14. 根据权利要求13所述的计算机设备,其中,所述处理器执行所述计算机可读指令时还执行以下步骤:
    对所述待检测对象的嘴唇动作特征进行提取,并将所述嘴唇动作特征输入所述预先训练的唇语识别模型;
    利用所述唇语识别模型对所述嘴唇动作特征进行特征解析,将所述嘴唇动作特征与预设的嘴唇动作特征样本进行比对;及
    当所述嘴唇动作特征符合所述嘴唇动作特征样本时,将所述嘴唇动作特征样本对应的展示码,作为所述嘴唇动作特征展示码。
  15. 一个或多个存储有计算机可读指令的计算机可读存储介质,所述计算机可读指令被一个或多个处理器执行时,使得所述一个或多个处理器执行以下步骤:
    当检测到待检测对象的登录操作时,触发人脸检测指令,根据所述人脸检测指令对所述待检测对象进行人脸检测,获得人脸检测结果;
    根据所述人脸检测结果生成随机展示码序列;所述随机展示码序列根据随机数字、随机字母或随机汉字中至少一种形式组成;
    对所述待检测对象的嘴唇动作特征进行提取,并利用预先训练的唇语识别模型对所述嘴唇动作特征进行特征解析,获得对应的嘴唇动作特征展示码;
    提取所述待检测对象的解读音频,并对所述解读音频进行语音识别,将所述解读音频转换成语音文本展示码;
    将所述嘴唇动作特征展示码和所述语音文本展示码,与所述随机展示码序列进行比对,当所述嘴唇动作特征展示码和所述语音文本展示码均符合所述随机展示码序列时,对所述待检测对象进行活体检测;及
    当确认所述待检测对象为活体对象时,调用数据库中预存的待检测对象的脸部图像文件,与所述人脸检测结果进行比对,进行身份认证,得到身份认证结果。
  16. 根据权利要求15所述的存储介质,其中,所述计算机可读指令被所述处理器执行时还执行以下步骤:
    当检测到所述待检测对象的登录操作时,获取所述待检测对象的登录状态;所述登录状态包括登录成功和登录失败;
    当确定所述待检测对象的登录状态为登陆成功时,触发人脸检测指令;及
    根据所述人脸检测指令打开相机程序,利用所述相机程序对所述待检测对象进行人脸检测,当检测到所述待检测对象的人脸部分时,生成人脸检测结果。
  17. 根据权利要求15所述的存储介质,其中,所述计算机可读指令被所述处理器执行时还执行以下步骤:
    获取当前应用场景下的所述随机展示码序列的预设要求;所述预设要求包括所述随机展示码序列长度,以及所述随机展示码序列组成;及
    当根据所述人脸检测结果,确定检测到所述待检测对象的人脸部分时,根据所述随机展示码序列的预设要求,随机生成所述随机展示码序列。
  18. 根据权利要求15所述的存储介质,其中,所述计算机可读指令被所述处理器执行时还执行以下步骤:
    根据样本数据对深度学习模型进行训练,获得训练后的唇语识别模型。
  19. 根据权利要求18所述的存储介质,其中,所述计算机可读指令被所述处理器执行时还执行以下步骤:
    收集各数字以及各字母对应的唇形图片;
    从数据库中提取超出预设使用频率的多个汉字,并收集多个所述汉字对应的唇形图片;
    提取各所述唇形图片上的嘴唇动作特征,生成样本数据;及
    利用所述样本数据,对卷积神经网络模型进行训练,得到对应的唇语识别模型。
  20. 根据权利要求19所述的存储介质,其中,所述计算机可读指令被所述处理器执行时还执行以下步骤:
    对所述待检测对象的嘴唇动作特征进行提取,并将所述嘴唇动作特征输入所述预先训练的唇语识别模型;
    利用所述唇语识别模型对所述嘴唇动作特征进行特征解析,将所述嘴唇动作特征与预设的嘴唇动作特征样本进行比对;及
    当所述嘴唇动作特征符合所述嘴唇动作特征样本时,将所述嘴唇动作特征样本对应的展示码,作为所述嘴唇动作特征展示码。
PCT/CN2020/106238 2019-10-12 2020-07-31 身份验证方法、装置、计算机设备和存储介质 WO2021068616A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910969831.2 2019-10-12
CN201910969831.2A CN110955874A (zh) 2019-10-12 2019-10-12 身份验证方法、装置、计算机设备和存储介质

Publications (1)

Publication Number Publication Date
WO2021068616A1 true WO2021068616A1 (zh) 2021-04-15

Family

ID=69975586

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/106238 WO2021068616A1 (zh) 2019-10-12 2020-07-31 身份验证方法、装置、计算机设备和存储介质

Country Status (2)

Country Link
CN (1) CN110955874A (zh)
WO (1) WO2021068616A1 (zh)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113242551A (zh) * 2021-06-08 2021-08-10 中国银行股份有限公司 手机银行登录验证方法及装置
CN114696988A (zh) * 2022-03-11 2022-07-01 支付宝(杭州)信息技术有限公司 活体检测方法、装置、设备及系统
CN116259100A (zh) * 2022-09-28 2023-06-13 北京中关村科金技术有限公司 基于唇动跟踪的识别方法、装置、存储介质及电子设备
CN116319100A (zh) * 2023-05-22 2023-06-23 交通运输部水运科学研究所 基于gis空间数据分析的港区危险源安全准入方法及系统

Families Citing this family (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110955874A (zh) * 2019-10-12 2020-04-03 深圳壹账通智能科技有限公司 身份验证方法、装置、计算机设备和存储介质
CN113743160A (zh) * 2020-05-29 2021-12-03 北京中关村科金技术有限公司 活体检测的方法、装置以及存储介质
CN112101931A (zh) * 2020-08-27 2020-12-18 江西台德智慧科技有限公司 一种增强刷脸支付技术安全性的方法
CN112329750B (zh) * 2021-01-05 2021-04-20 北京远鉴信息技术有限公司 基于数字序列的身份识别方法、装置、电子设备及介质
CN113011301A (zh) * 2021-03-10 2021-06-22 北京百度网讯科技有限公司 一种活体识别方法、装置及电子设备
CN114731280B (zh) * 2022-02-25 2024-02-09 百果园技术(新加坡)有限公司 身份认证方法、装置、终端及存储介质
CN115880789B (zh) * 2023-02-08 2023-06-30 中昊芯英(杭州)科技有限公司 基于加密的人脸识别方法、相关装置和存储介质
CN117171694B (zh) * 2023-11-02 2024-01-30 北京龙德缘电力科技发展有限公司 一种基于ai技术的配电场景安全识别系统

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104598796A (zh) * 2015-01-30 2015-05-06 科大讯飞股份有限公司 身份识别方法及系统
CN105119872A (zh) * 2015-02-13 2015-12-02 腾讯科技(深圳)有限公司 身份验证方法、客户端和服务平台
CN106407783A (zh) * 2016-09-23 2017-02-15 武汉必腾移动科技有限责任公司 一种在线验证方法及系统
CN109214820A (zh) * 2018-07-06 2019-01-15 厦门快商通信息技术有限公司 一种基于音视频结合的商户收款系统及方法
CN110955874A (zh) * 2019-10-12 2020-04-03 深圳壹账通智能科技有限公司 身份验证方法、装置、计算机设备和存储介质

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108427874A (zh) * 2018-03-12 2018-08-21 平安科技(深圳)有限公司 身份认证方法、服务器及计算机可读存储介质
CN109815665A (zh) * 2018-12-25 2019-05-28 深圳供电局有限公司 身份认证方法及系统、电子设备、计算机可读存储介质
CN110113319A (zh) * 2019-04-16 2019-08-09 深圳壹账通智能科技有限公司 身份认证方法、装置、计算机设备和存储介质

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104598796A (zh) * 2015-01-30 2015-05-06 科大讯飞股份有限公司 身份识别方法及系统
CN105119872A (zh) * 2015-02-13 2015-12-02 腾讯科技(深圳)有限公司 身份验证方法、客户端和服务平台
CN106407783A (zh) * 2016-09-23 2017-02-15 武汉必腾移动科技有限责任公司 一种在线验证方法及系统
CN109214820A (zh) * 2018-07-06 2019-01-15 厦门快商通信息技术有限公司 一种基于音视频结合的商户收款系统及方法
CN110955874A (zh) * 2019-10-12 2020-04-03 深圳壹账通智能科技有限公司 身份验证方法、装置、计算机设备和存储介质

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113242551A (zh) * 2021-06-08 2021-08-10 中国银行股份有限公司 手机银行登录验证方法及装置
CN114696988A (zh) * 2022-03-11 2022-07-01 支付宝(杭州)信息技术有限公司 活体检测方法、装置、设备及系统
CN116259100A (zh) * 2022-09-28 2023-06-13 北京中关村科金技术有限公司 基于唇动跟踪的识别方法、装置、存储介质及电子设备
CN116319100A (zh) * 2023-05-22 2023-06-23 交通运输部水运科学研究所 基于gis空间数据分析的港区危险源安全准入方法及系统
CN116319100B (zh) * 2023-05-22 2023-07-28 交通运输部水运科学研究所 基于gis空间数据分析的港区危险源安全准入方法及系统

Also Published As

Publication number Publication date
CN110955874A (zh) 2020-04-03

Similar Documents

Publication Publication Date Title
WO2021068616A1 (zh) 身份验证方法、装置、计算机设备和存储介质
WO2020077885A1 (zh) 身份验证方法、装置、计算机设备和存储介质
CN110557376B (zh) 电子合同签署方法、装置、计算机设备及存储介质
US10275672B2 (en) Method and apparatus for authenticating liveness face, and computer program product thereof
CN109165940B (zh) 一种防盗方法、装置及电子设备
WO2020140665A1 (zh) 双录视频质量检测方法、装置、计算机设备和存储介质
WO2020024398A1 (zh) 生物特征辅助支付方法、装置、计算机设备及存储介质
CN107800672B (zh) 一种信息验证方法、电子设备、服务器及信息验证系统
US20210089635A1 (en) Biometric identity verification and protection software solution
CN111861240A (zh) 可疑用户识别方法、装置、设备及可读存储介质
Thomas et al. A broad review on non-intrusive active user authentication in biometrics
CN109389028A (zh) 基于动作分析的人脸识别方法、装置、设备及存储介质
CN110795714A (zh) 一种身份验证方法、装置、计算机设备及存储介质
EP3848830A1 (en) User authentication device and recording medium
CN110929244A (zh) 数字化身份识别方法、装置、设备及存储介质
Kim et al. Reconstruction of fingerprints from minutiae using conditional adversarial networks
Lesani et al. Mobile phone security using automatic lip reading
Sharma et al. Behavioral biometrics: past, present and future
CN113627387A (zh) 基于人脸识别的并行身份认证方法、装置、设备和介质
Stylios et al. BioPrivacy: a behavioral biometrics continuous authentication system based on keystroke dynamics and touch gestures
CN110738499B (zh) 用户身份验证方法、装置、计算机设备和存储介质
WO2016139655A1 (en) Method and system for preventing uploading of faked photos
CN116886315A (zh) 一种web3.0下的基于生物特征和零知识证明的身份验证方法
CN111339829B (zh) 用户身份鉴定方法、装置、计算机设备和存储介质
Zolotarev et al. Liveness detection methods implementation to face identification reinforcement in gaming services

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20874218

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 16/08/2022)

122 Ep: pct application non-entry in european phase

Ref document number: 20874218

Country of ref document: EP

Kind code of ref document: A1