Intelligent face recognition identity verification system and method for examination environment

Abstract

The invention relates to the technical field related to data processing, in particular to an intelligent face recognition identity verification system and method for an examination environment, and the system comprises the steps that a face recognition device collects face data of examinees, and establishes an image set after preprocessing; the authentication database calls examinee information, performs feature extraction and comparison on the preprocessed image, and generates authentication feedback; the technical problem that a single static image comparison mode is poor in adaptability in a complex illumination environment and a dynamic face scene and cannot adapt to complex environments and dynamic changes is solved, face data of the examinees are collected through a high-precision linkage camera, and the face data of the examinees can be accurately compared with the face data of the examinees. And in combination with illumination compensation and a dynamic feature extraction mechanism, the method is suitable for a large-scale examination environment with high security requirements, and the security and reliability of identity verification are enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field related to data processing, and in particular to an intelligent face recognition identity authentication system and method for use in an examination environment. Background Art

[0002] With the advancement of science and technology, biometric technology, especially face recognition technology, has been widely used in identity authentication, security monitoring and other fields, such as finance, security, education and other industries. Among them, in the examination scenarios in the education field, accurate identification and management of candidates' identities is particularly important. With the expansion of large-scale examinations such as college entrance examinations and postgraduate examinations, higher requirements are placed on the candidate identity verification system: it is necessary to ensure the accuracy of identity authentication and to be able to adapt to efficient and batch examination management scenarios.

[0003] The existing examination identity verification system mainly relies on the verification of identity cards or admission tickets, which makes it difficult to effectively prevent fraudulent behaviors such as identity forgery and proxy examinations. At the same time, the single static image comparison mode has poor adaptability in complex lighting environments and dynamic face scenes, and is easily affected by factors such as the external environment and changes in the examinee's face (such as expression, angle).

[0004] In summary, the existing technology has the technical problem that a single static image comparison mode has poor adaptability in complex lighting environments and dynamic face scenes and cannot adapt to complex environments and dynamic changes. Summary of the invention

[0005] This application provides an intelligent face recognition identity authentication system and method for use in an examination environment, aiming to solve the technical problem that the single static image comparison mode in the prior art has poor adaptability in complex lighting environments and dynamic face scenes, and cannot adapt to complex environments and dynamic changes.

[0006] In view of the above problems, the technical solution to implement this application is: On the one hand, the present application provides an intelligent face recognition identity authentication system for an examination environment, wherein the system includes: a face recognition hardware device, wherein the face recognition hardware device is configured with a linked acquisition camera, wherein the linked acquisition camera includes at least two high-definition cameras and one infrared camera, and the face recognition hardware device is configured with a fill light, and the face recognition hardware device is used to perform data acquisition of the examinee and establish a set of examinee face images; a preprocessing module, wherein the preprocessing module is used to perform joint image preprocessing on the examinee face image set and establish a preprocessed image; an authentication database, wherein the authentication database is used to call the examinee data, establish a verification data set based on the examinee data, perform feature extraction on the preprocessed image after performing face positioning, establish a numerical vector of facial features, use the verification data set to verify and compare the numerical vector, and establish authentication feedback; an examination module, wherein the examination module is used to receive the authentication feedback and perform examinee examination management according to the authentication feedback.

[0007] On the other hand, the present application provides an intelligent face recognition identity authentication method for an examination environment, wherein the method includes: a face recognition hardware device is configured with a linked acquisition camera, the linked acquisition camera includes at least two high-definition cameras and one infrared camera, and the face recognition hardware device is configured with a fill light, the face recognition hardware device is used to perform data acquisition of the examinee, and a face image set of the examinee is established; joint image preprocessing is performed on the face image set of the examinee to establish a preprocessed image; the examinee data is called, a verification data set is established based on the examinee data, face positioning is performed on the preprocessed image, feature extraction is performed, a numerical vector of face features is established, the numerical vector is verified and compared using the verification data set, and authentication feedback is established; the authentication feedback is received, and the examinee examination is managed according to the authentication feedback.

[0008] In summary, one or more technical solutions provided in this application solve the technical problem that a single static image comparison mode has poor adaptability in complex lighting environments and dynamic face scenes, and cannot adapt to complex environments and dynamic changes. It realizes the collection of examinee facial data through high-precision linked cameras, and combines lighting compensation and dynamic feature extraction mechanisms. It is suitable for large-scale, high-security examination environments, and enhances the technical effect of the security and reliability of identity authentication. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 A schematic diagram of the structure of an intelligent face recognition identity authentication system for an examination environment is provided for this application; Figure 2 A flow chart of an intelligent face recognition identity authentication method for an examination environment is provided for this application.

[0010] Explanation of the reference numerals: face recognition hardware device M100, preprocessing module M200, authentication database M300, examination module M400. DETAILED DESCRIPTION

[0011] Embodiment 1: The present application is described in detail below with reference to the accompanying drawings. Figure 1 As shown, the embodiment of the present application provides an intelligent face recognition identity authentication system for an examination environment, wherein the system includes: The face recognition hardware device M100 is configured with a linkage acquisition camera, which includes at least two high-definition cameras and one infrared camera, and the face recognition hardware device is configured with a fill light. The face recognition hardware device is used to perform data collection of candidates and establish a candidate face image set.

[0012] Specifically, the linked collection camera refers to multiple cameras working together, including at least two high-definition cameras and one infrared camera, which are used to capture the facial images of candidates from different angles and spectral ranges; the infrared camera is a camera capable of capturing images within the infrared spectrum, which is suitable for low-light or no-light environments; the fill light is used to provide additional light source when the light is insufficient to ensure image clarity; candidate data collection refers to the collection of candidate facial image data through hardware equipment; the candidate face image set refers to the collected set of candidate facial image data for subsequent identity verification and analysis.

[0013] The face recognition hardware equipment performs data collection of candidates through the configured linked collection cameras. At least two high-definition cameras can capture the facial details of the candidates from different angles, while the infrared camera can provide additional image information in low-light conditions, enhancing the system's adaptability under various lighting conditions. The configuration of the fill light further ensures that clear facial images can be obtained even in low-light conditions. The collected images are used to establish the candidate's face image set, which is the basis for subsequent preprocessing, feature extraction and identity authentication, and provides support for accurate identity authentication.

[0014] The preprocessing module M200 is used to perform joint image preprocessing on the examinee's face image set to establish a preprocessed image.

[0015] Specifically, the preprocessing module is responsible for the preliminary processing of the collected original images during the facial recognition process; joint image preprocessing refers to the unified processing flow of images from multiple sources or types to improve the efficiency and accuracy of subsequent processing steps; preprocessed images refer to images processed by the preprocessing module, which are ready for the next step of feature extraction and recognition.

[0016] The role of the preprocessing module is to perform a series of preprocessing operations on the images in the candidate's face image set to improve the accuracy and efficiency of face recognition, including image cropping (for example, removing unnecessary background parts and retaining only the facial area), scaling, rotation correction, grayscale, histogram equalization (increasing the contrast of the image to make facial features more obvious), etc.; joint image preprocessing means that these operations are not only for images captured by high-definition cameras, but also for images captured by infrared cameras to ensure that the two images are consistent and comparable before feature extraction. In this way, the preprocessing module ensures that images captured under any lighting conditions can be effectively used in the recognition process, thereby improving the overall performance and reliability of the system.

[0017] The authentication database M300 is used to call the examinee data, establish a verification data set based on the examinee data, perform feature extraction on the pre-processed image after face positioning, establish a numerical vector of facial features, use the verification data set to verify and compare the numerical vector, and establish authentication feedback.

[0018] Specifically, the authentication database is used to store the candidate's identity information and the database for verification, which is used for data comparison during the identity verification process; the candidate's data refers to the candidate's personal information, including name, ID number, pre-collected facial images, etc.; the verification data set refers to the database for verification established based on the candidate's data, which contains feature data for comparison; face positioning is the process of identifying and locating the face in the image; feature extraction is the process of extracting key feature points and feature vectors from the face image; the numerical vector represents the mathematical vector of facial features, which is used for subsequent comparison and recognition.

[0019] The examinee data is called, including the examinee's basic information and the previously collected face images; based on these data, a verification data set is established for the subsequent comparison process; when the preprocessing module completes the image preprocessing, the authentication database will receive these preprocessed images and perform face positioning to determine the position of the face in the image; perform feature extraction to convert the face image into a numerical vector, which contains the key features of the face; use the feature vectors in the verification data set to verify and compare the newly extracted numerical vectors, compare the extracted feature vectors with the feature vectors in the database to verify the authenticity of the examinee's identity, and establish authentication feedback, which refers to the comparison result, indicating whether it matches or not, and is used for the final identity authentication decision. If the comparison result is significantly different, it may trigger further verification processes or alarms to prevent identity fraud. In this way, the authentication database ensures the accuracy and efficiency of identity authentication during the examination process.

[0020] The examination module M400 and the examination module M500 are used to receive the certification feedback and perform candidate examination management according to the certification feedback.

[0021] Specifically, the examination module is responsible for managing the candidate's examination process based on the certification results; certification feedback is the result provided by the certification database, indicating whether the candidate's identity authentication is successful; candidate examination management is to manage the candidate's examination process based on the identity authentication results, including allowing or refusing candidates to take the examination, monitoring examination behavior, etc.

[0022] The examination module is responsible for receiving the certification feedback provided by the certification database and managing the candidates' examinations accordingly; when the certification feedback confirms that the candidate's identity is correct, the examination module will allow the candidate to enter the examination room or continue the examination; if the certification feedback shows that the identity authentication failed or there is a question, the examination module can take further verification measures or refuse the candidate to take the examination, thereby ensuring that only verified candidates can take the examination and improving the level of automation and intelligence of examination management.

[0023] Furthermore, the authentication database M300 also includes: A reading module, the reading module is used to read the candidate's ID card and establish basic candidate information; an information calling module, the information calling module is used to call the candidate's photo based on the basic candidate information and establish a time-series candidate photo data set; an authentication module, the authentication module is used to perform time-series trust authentication on the time-series candidate photo data set and establish a trusted photo set; a feature reconstruction module, the feature reconstruction module is used to perform time-series feature extraction based on the trusted photo set, and perform feature evolution verification on the time-series feature extraction results to establish the verification data set.

[0024] Specifically, the reading module is used to read and parse the candidate's ID card information; the candidate's ID card is the candidate's identity document, which usually contains personal information such as name, ID number, etc.; basic candidate information refers to the candidate's basic information, which is used to establish the candidate's file and perform identity verification; the information call module is used to retrieve the candidate's photo based on the basic candidate information; the time series candidate photo dataset contains a dataset of historical photos of the candidates, which are arranged in chronological order; the authentication module is used to analyze and authenticate the time series candidate photo dataset. Correspondingly, the time series trust authentication takes into account the changes in the candidate's photos over time to establish a trusted photo set; the feature reconstruction module is used to extract facial features from the trusted photo set and perform feature evolution verification; time series feature extraction is to extract facial features from the time series of the candidate's photos; feature evolution verification includes verifying whether the changes of the extracted features over time are reasonable.

[0025] The main function of the reading module is to read the ID card information provided by the candidate and convert this information into electronic format and store it in the system. It can establish basic candidate information, including the candidate's name, ID card number, date of birth and other key data; the information call module retrieves the candidate's registration photo from the database based on the basic candidate information provided by the reading module, and establishes a time-series candidate photo data set, which includes photos of the candidate at different time points, for subsequent identity verification and feature analysis.

[0026] The authentication module analyzes the time-series candidate photo dataset retrieved by the information call module, and identifies the consistency and changes in the candidate's face by comparing photos at different time points, thereby performing time-series trust authentication. The goal of the authentication module is to establish a trusted photo set, which contains candidate photos that are authenticated as authentic and reliable, for subsequent feature extraction and identity authentication, to improve the security of identity authentication, and to ensure that only authenticated candidate photos are used in the final verification process; the feature reconstruction module performs time-series feature extraction based on the trusted photo set established by the authentication module, analyzes the changes in the candidate's facial features over time, and performs feature evolution verification to establish a verification dataset. The verification dataset contains detailed information on the candidate's facial features, which is used to compare with the candidate's facial features collected in real time to verify the candidate's identity, improve the accuracy and dynamic adaptability of identity authentication, and adapt to changes in the candidate's face over time.

[0027] Furthermore, the feature reconstruction module includes: A time tagging module, the time tagging module is used to perform time sequence tagging on the trusted photo set to establish a time sequence; a dynamic feature extraction module, the dynamic feature extraction module includes: The identity feature extraction channel is used to extract the facial key point coordinates of the trusted photo set according to the time series to establish the identity feature set; the illumination feature extraction channel is used to extract the illumination features of the trusted photo set according to the time series to establish the illumination feature set; the displacement analysis channel is used to receive the identity feature set and the illumination feature set, perform facial feature displacement analysis under illumination difference elimination, establish facial feature dynamic trajectory, and establish the verification data set according to the facial feature dynamic trajectory.

[0028] Specifically, the time labeling module is used to add a time tag to each photo in the trusted photo set; the trusted photo set refers to a collection of candidate photos that have been authenticated by the authentication module and are considered to be authentic and reliable; time series labeling is the process of marking the shooting time on the photo to facilitate the establishment of a time series of photos; a time series refers to a data sequence arranged in chronological order, which is used to analyze the changes in data over time; the dynamic feature extraction module is used to extract the dynamic features of the candidate's face from the time series, and the dynamic feature extraction module includes an identity feature extraction channel, an illumination feature extraction channel, and a displacement analysis channel, wherein the identity feature extraction channel is used to extract the coordinates of facial key points for identifying the identity; the illumination feature extraction channel is used to extract illumination features for easy identification under different lighting conditions; the displacement analysis channel is used to analyze the displacement of facial features and establish dynamic trajectories.

[0029] The main task of the time annotation module is to perform time series annotation on each photo in the trusted photo set, that is, to record the specific time when each photo was taken. The time series can reflect the changes in the candidate's facial features over time, and provide an accurate time reference for subsequent dynamic feature extraction and analysis; the dynamic feature extraction module includes three channels: identity feature extraction channel, illumination feature extraction channel and displacement analysis channel. The identity feature extraction channel extracts the coordinates of facial key points of the trusted photo set according to the time series. Among them, the identity feature extraction channel identifies and tracks the positions of key facial features of the candidate, such as the eyes, corners of the mouth and the tip of the nose; the illumination feature extraction channel analyzes the lighting conditions of each photo and extracts relevant features for effective identification under different lighting conditions; the displacement analysis channel analyzes the changes in these features over time, such as slight movements in the position of the eyes, and establishes a dynamic trajectory to deeply understand the natural changes in the candidate's facial features and provide more accurate data support during the identity verification process.

[0030] Furthermore, the face recognition hardware device M100 includes: A perception module, which is used to sense the presence of the examinee, locate the examinee's face, perform facial light detection based on the facial positioning result, and establish light feedback; a fill light control module, which is used to configure a fill light signal based on the light feedback, use the fill light signal to control the fill light, and then synchronously activate the linkage acquisition camera; an information acquisition module, which is used to collect the examinee's facial data based on the linkage acquisition camera, and establish a set of examinee face images.

[0031] Specifically, the perception module is used to detect the presence of the examinee and locate his / her face; examinee presence perception is used to detect whether the examinee is within the shooting range of the camera; face positioning is used to determine the position of the examinee's face in the image; face light detection is used to evaluate the lighting conditions of the facial area; the fill light control module is used to adjust the brightness of the fill light according to light feedback; the fill light signal is a signal generated based on light feedback and is used to control the fill light; fill light control is to adjust the brightness of the fill light to adapt to different lighting conditions; the information acquisition module is used to collect the examinee's facial data from the linked acquisition cameras; the linked acquisition cameras are multiple cameras working in coordination to capture images from different angles and spectra; the examinee's face image set refers to a data set consisting of facial images acquired by multiple cameras.

[0032] Detect whether the examinee has entered the camera's shooting range. Once the examinee is detected, the face is immediately located to determine the specific position of the face; detect the lighting conditions of the facial area to ensure that the best image quality can be obtained in the subsequent image acquisition process; according to the results of facial light detection, the perception module will establish light feedback, inform the fill light control module through light feedback that the lighting needs to be enhanced, and provide the fill light control module with necessary data; the fill light control module receives light feedback from the perception module and configures the fill light signal according to these feedbacks to control the brightness of the fill light; if the light detected by the perception module is insufficient, the fill light control module will increase the brightness of the fill light to ensure that the facial image is clear and visible, and ensure that the linkage acquisition camera captures the examinee's facial image under the best lighting conditions; the information acquisition module is responsible for collecting the examinee's facial data from the linkage acquisition camera, including high-definition images and infrared images, and synchronously activates the linkage acquisition camera. The collected images are used to establish the examinee's face image set to provide accurate visual data for subsequent identity verification. In this way, clear and accurate facial images can be obtained under various lighting conditions, thereby improving the accuracy and reliability of identity verification.

[0033] Furthermore, the preprocessing module M200 includes: An alignment module, which is used to perform image alignment after key point detection on the candidate's face image set; a joint fusion module, which is used to obtain high-definition images in the candidate's face image set, extract image features on the high-definition images, establish position trust, use the position trust and image alignment to perform image trust fusion, and establish a trust fusion result; an additional fusion module, which is used to perform compensatory fusion of the trust fusion results based on the image alignment and infrared images to establish the preprocessed image.

[0034] Specifically, the alignment module is used to adjust the images in the candidate's face image set to make them conform to specific standards or templates; key point detection refers to identifying specific points in the face image, such as the corners of the eyes, the tip of the nose, the corners of the mouth, etc., for aligning the image; image alignment refers to adjusting the image so that the key points of the face are consistent with the predefined standard positions; the joint fusion module is used to combine the high-definition image and position trust information to generate a fused image; image feature extraction refers to extracting key features that aid identification from the high-definition image; position trust is an accuracy assessment based on image features and key point positions, which is used to measure the credibility of the image; the additional fusion module is used to further optimize the trust fusion results by combining infrared images for compensatory fusion; compensatory fusion refers to adjusting and optimizing the fused image to compensate for possible recognition errors or missing information.

[0035] Key point detection is performed on each image in the candidate's facial image set to identify the key feature points of the face; image alignment is performed based on these key points to ensure that the faces in all images are arranged according to a unified standard, reducing recognition errors caused by different head postures. In this way, no matter how the candidate's head is tilted or rotated, it can be adjusted to a standard posture for subsequent processing.

[0036] Obtain high-definition images from the candidate's facial image set, extract image features, and establish positional trust; use the positional trust and image alignment results to perform image trust fusion to generate trust fusion results, obtain the candidate's facial images from high-definition cameras at different angles, extract the key features of each image, and evaluate the positional trust of the key features, combine multiple facial images corresponding to high-definition cameras at different angles and trust information, and generate a fused image. The fused image integrates information from multiple perspectives and can more comprehensively represent the candidate's facial features. Through the above steps, while retaining the details of the high-definition image, the recognition accuracy and robustness of the image are improved by fusing images under multiple perspectives or conditions.

[0037] Based on the image alignment results and infrared images generated by the alignment module, the trust fusion results are compensated and fused. On the basis of high-definition image fusion, the information provided by the infrared image is further used to enhance the image recognition ability. Especially in the case of unsatisfactory lighting conditions, the trust fusion results of the high-definition image are compared with the infrared image to identify and compensate for the missing facial features caused by lighting changes or occlusion. In this way, the quality and recognition accuracy of the pre-processed image are improved, providing a solid foundation for subsequent feature extraction and identity authentication.

[0038] Furthermore, the authentication database M300 includes: A positioning module, the positioning module is used to locate the face and establish the face area; a feature extraction module, the feature extraction module is used to extract the high-dimensional feature vector after locating the key points of the face area; a dimensionality reduction module, the dimensionality reduction module is used to establish the numerical vector after performing feature dimensionality reduction on the high-dimensional feature vector.

[0039] Specifically, the positioning module is used to identify and determine the position of the face in the image; the face area refers to the part of the image that is identified as containing the face, providing a specific operation area for subsequent processing; the feature extraction module is used to extract key information from the face area that helps with identification and verification; key point positioning refers to the key feature points in the face, such as the corners of the eyes, the tip of the nose, the corners of the mouth, etc.; the high-dimensional feature vector contains a vector of multiple eigenvalues ​​to quickly distinguish different individuals, which is used to represent the unique features of the face; the dimensionality reduction module is used to reduce the dimension of the feature vector while retaining the most important information as much as possible; feature dimensionality reduction refers to reducing the dimension of the feature vector for easy processing and comparison; the numerical vector refers to the feature vector after dimensionality reduction, which is used for final identity verification and comparison.

[0040] Identify the face in the image and determine its position, so as to establish the face area. This involves face detection algorithms, such as Haar features or deep learning models, which can quickly and accurately locate the face in various complex backgrounds, provide precise face areas for subsequent feature extraction, ensure the accuracy and effectiveness of feature extraction, locate key points based on the face area determined by the positioning module, and extract high-dimensional feature vectors, which involves extracting features that can represent personal identity from the face image, such as geometric features, texture features, depth features, etc. The function of the feature extraction module is to convert the face image into a numerical form that can be used for comparison and identification, and provide key information for identity authentication.

[0041] The dimension reduction module receives the high-dimensional feature vector generated by the feature extraction module and performs dimension reduction processing on it to establish a numerical vector, reduce the complexity of the data, improve the calculation efficiency, and try to retain the key information required to distinguish different individuals. The role of the dimension reduction module is to optimize the feature vector without losing the recognition accuracy, making it more suitable for real-time processing and comparison. Specifically, PCA (Principal Component Analysissi) or other dimension reduction techniques are used to process high-dimensional feature vectors. In this way, the high-dimensional feature vector is converted into a lower-dimensional numerical vector while retaining the most important feature information, providing efficient and accurate data support for the final identity authentication.

[0042] Furthermore, the intelligent face recognition identity authentication system for an examination environment further comprises: A continuous acquisition module, which is used to identify the examinee of interest and dynamically monitor the examinee's facial features during the answering process to establish a dynamic data set; an additional verification module, which is used to perform additional verification based on the dynamic data set and the verification data set, and manage the examinee according to the additional verification results.

[0043] Specifically, the continuous acquisition module is used to continuously collect facial dynamic data of specific candidates over a period of time; candidates of concern refer to candidates who need special monitoring in the examination environment, perhaps because the candidate's behavior or history has attracted extra attention; the dynamic data set contains a data set of dynamic facial changes of candidates over a period of time; the additional verification module is used to further verify the identity of the candidate after preliminary verification; the dynamic data set contains a data set of the candidate's facial dynamic information during the examination process; the verification data set contains a data set of facial features and identity information collected when the candidate registered; candidate management refers to the management operations performed on the candidate based on the verification results, such as verbal warnings and disqualification from the examination.

[0044] Preferably, VPN (Virtual Private Network) and other technologies are used to ensure the security of data transmission. After identifying a candidate who needs special attention, the candidate's facial dynamics are monitored throughout the entire answering process, including collecting dynamic information such as the candidate's facial expressions, eye movements, head posture, and establishing a dynamic data set. The dynamic data set can be used to analyze the candidate's behavior patterns and detect abnormal behaviors, such as frequent changes in facial expressions or unusual eye movements, which indicate the candidate's nervousness and restlessness. Through facial dynamic monitoring, changes in the candidate's facial expressions and movements are tracked and recorded in real time, providing real-time facial dynamic data to improve the intensity and efficiency of test monitoring; on the basis of preliminary verification, further verification is performed using additional data. Specifically, additional verification is performed using the dynamic data set and the verification data set, analyzing the dynamic facial changes of the candidate during the test and comparing them with the facial features collected when the candidate registered. If an abnormality or inconsistency is found, the candidate is marked to trigger additional verification processes, such as re-identification or notifying the invigilator to conduct a manual check, to improve the accuracy of identity verification.

[0045] In summary, the beneficial effects of the embodiments of the present application are: The face recognition hardware device is equipped with a linkage acquisition camera, which includes at least two high-definition cameras and one infrared camera, and the face recognition hardware device is equipped with a fill light. The face recognition hardware device is used to perform data collection on the examinee and establish a face image set of the examinee; the preprocessing module is used to perform joint image preprocessing on the face image set of the examinee and establish a preprocessed image; the authentication database is used to call the examinee data, establish a verification data set based on the examinee data, perform feature extraction on the preprocessed image after face positioning, establish a numerical vector of the face feature, use the verification data set to verify and compare the numerical vector, and establish authentication feedback; the examination module is used to receive authentication feedback and manage the examinee's examination according to the authentication feedback. It realizes the collection of examinee face data through high-precision linkage cameras, and combines illumination compensation and dynamic feature extraction mechanisms, which is suitable for large-scale, high-security examination environments, and enhances the technical effect of the security and reliability of identity authentication.

[0046] Embodiment 2: Based on the same inventive concept as the intelligent face recognition identity authentication system for examination environment in the aforementioned embodiment, Figure 2 As shown, the present application provides an intelligent face recognition identity authentication method for an examination environment, wherein the method comprises: S1: The face recognition hardware device is configured with a linkage acquisition camera, which includes at least two high-definition cameras and one infrared camera, and the face recognition hardware device is configured with a fill light. The face recognition hardware device is used to perform data acquisition of the examinee and establish a examinee face image set; S2: The examinee face image set is subjected to joint image preprocessing to establish a preprocessed image; S3: The examinee data is called, a verification data set is established based on the examinee data, and after locating the face of the preprocessed image, feature extraction is performed to establish a numerical vector of the face feature, and the verification data set is used to verify and compare the numerical vector to establish authentication feedback; S4: The authentication feedback is received, and the examinee examination management is performed according to the authentication feedback.

[0047] Furthermore, the present application method also includes: S31: Read the candidate's ID card and establish basic candidate information; S32: Call the candidate's photo based on the basic candidate information and establish a time-series candidate photo data set; S33: Perform time-series trust authentication on the time-series candidate photo data set to establish a trusted photo set; S34: Perform time-series feature extraction based on the trusted photo set, and perform feature evolution verification on the time-series feature extraction results to establish the verification data set.

[0048] Furthermore, the present application method comprises: S34: Time-sequence annotation of the trusted photo set to establish a time series; S35: Dynamic feature extraction module, the dynamic feature extraction module includes: S351: Extract the facial key point coordinates of the trusted photo set according to the time series and establish an identity feature set; S352: Extract the lighting features of the trusted photo set according to the time series and establish an lighting feature set; S353: Receive the identity feature set and the lighting feature set, perform facial feature displacement analysis under lighting difference elimination, establish facial feature dynamic trajectories, and establish the verification data set according to the facial feature dynamic trajectories.

[0049] Furthermore, the present application method comprises: S11: After sensing the presence of the examinee, locate the examinee's face, perform facial light detection based on the facial location result, and establish light feedback; S12: configure a fill light signal based on the light feedback, use the fill light signal to control the fill light, and then synchronously activate the linkage acquisition camera; S13: Based on the linkage acquisition camera, collect the examinee's facial data and establish a examinee's face image set.

[0050] Furthermore, the present application method comprises: S21: After performing key point detection on the candidate's face image set, perform image alignment; S22: Obtain high-definition images from the candidate's face image set, and perform image feature extraction on the high-definition images, establish position trust, and use the position trust and image alignment to perform image trust fusion to establish a trust fusion result; S23: Perform compensatory fusion of the trust fusion results based on the image alignment and infrared images to establish the preprocessed image.

[0051] Furthermore, the present application method comprises: S35: locate the face and establish the face area; S36: extract the high-dimensional feature vector after locating the key points of the face area; S37: establish the numerical vector after performing feature dimensionality reduction on the high-dimensional feature vector.

[0052] Furthermore, the present application method also includes: S51: After identifying the candidate of interest, the facial dynamics of the candidate during the answering process are monitored to establish a dynamic data set; S52: additional verification is performed based on the dynamic data set and the verification data set, and the candidate is managed according to the additional verification result.

[0053] In summary, any step can be stored as a computer instruction or program in an unlimited computer memory, and can be called and recognized by an unlimited computer processor, and no unnecessary restrictions are made here.

[0054] Furthermore, the above-mentioned technical scheme only reflects the preferred technical scheme of the technical scheme of the embodiment of the present application. Some changes that may be made to some parts thereof by technicians in this technical field all reflect the novel principles of the embodiment of the present application. Obviously, technicians in this field can make various changes and modifications to the present application without departing from the scope of the present application.

Claims

1. An intelligent face recognition identity authentication system for an examination environment, characterized in that: The system comprises: A face recognition hardware device, wherein the face recognition hardware device is configured with a linkage collection camera, wherein the linkage collection camera includes at least two high-definition cameras and one infrared camera, and the face recognition hardware device is configured with a fill light, and the data of the examinee is collected by the face recognition hardware device to establish a face image set of the examinee; A preprocessing module, the preprocessing module is used to perform joint image preprocessing on the examinee's face image set to establish a preprocessed image; An authentication database, wherein the authentication database is used to call the examinee data, establish a verification data set based on the examinee data, perform feature extraction after performing face location on the preprocessed image, establish a numerical vector of facial features, verify and compare the numerical vector using the verification data set, and establish authentication feedback; An examination module is used to receive the certification feedback and perform candidate examination management according to the certification feedback.

2. The intelligent face recognition identity authentication system for an examination environment as claimed in claim 1, characterized in that: The authentication database also includes: A reading module, which is used to read the candidate's ID card and establish basic candidate information; An information calling module, the information calling module is used to call candidates' photos based on the basic candidate information and establish a time-series candidate photo data set; An authentication module, the authentication module is used to perform time-series trust authentication on the time-series candidate photo data set to establish a trusted photo set; A feature reconstruction module is used to extract temporal features based on the trusted photo set, and perform feature evolution verification on the temporal feature extraction results to establish the verification data set.

3. The intelligent face recognition identity authentication system for an examination environment as claimed in claim 2, characterized in that: The feature reconstruction module comprises: A time marking module, the time marking module is used to perform time sequence marking on the trusted photo set to establish a time sequence; A dynamic feature extraction module, the dynamic feature extraction module comprising: An identity feature extraction channel, used to extract the facial key point coordinates of the trusted photo set according to the time series to establish an identity feature set; An illumination feature extraction channel, used to extract illumination features of a trusted photo set according to the time series, and establish an illumination feature set; The displacement analysis channel is used to receive the identity feature set and the illumination feature set, perform facial feature displacement analysis under illumination difference elimination, establish facial feature dynamic trajectory, and establish the verification data set according to the facial feature dynamic trajectory.

4. The intelligent face recognition identity authentication system for an examination environment as claimed in claim 1, characterized in that: The face recognition hardware device includes: A sensing module, wherein the sensing module is used to sense the presence of the examinee, locate the examinee's face, perform facial light detection based on the face location result, and establish light feedback; A fill light control module, the fill light control module is used to configure a fill light signal according to the light feedback, and after using the fill light signal to control the fill light, synchronously activate the linkage acquisition camera; The information collection module is used to collect the facial data of the examinee based on the linkage collection camera and establish a facial image set of the examinee.

5. The intelligent face recognition identity authentication system for an examination environment as claimed in claim 4, characterized in that: The preprocessing module comprises: An alignment module, the alignment module is used to perform image alignment after performing key point detection on the examinee's face image set; A joint fusion module, the joint fusion module is used to obtain a high-definition image in the examinee's face image set, perform image feature extraction on the high-definition image, establish position trust, perform image trust fusion using the position trust and image alignment, and establish a trust fusion result; An additional fusion module is used to perform compensatory fusion of trust fusion results based on the image alignment and the infrared image to establish the pre-processed image.

6. The intelligent face recognition identity authentication system for an examination environment as claimed in claim 1, characterized in that: The authentication database includes: A positioning module, the positioning module is used to locate the face and establish the face area; A feature extraction module, wherein the feature extraction module is used to extract a high-dimensional feature vector after locating key points of the face area; A dimension reduction module is used to establish the numerical vector after performing feature dimension reduction on the high-dimensional feature vector.

7. The intelligent face recognition identity authentication system for an examination environment as claimed in claim 1, characterized in that: The system further comprises: A continuous acquisition module, which is used to monitor the facial dynamics of the examinee during the answering process after identifying the examinee of interest, and establish a dynamic data set; An additional verification module is used to perform additional verification based on the dynamic data set and the verification data set, and to manage examinees according to the additional verification results.

8. An intelligent face recognition identity authentication method for an examination environment, characterized in that: The method for implementing the intelligent face recognition identity authentication system for an examination environment according to any one of claims 1 to 7 comprises: The face recognition hardware device is equipped with a linkage collection camera, which includes at least two high-definition cameras and one infrared camera, and the face recognition hardware device is equipped with a fill light. The face recognition hardware device is used to collect data of the examinee and establish a face image set of the examinee; Performing joint image preprocessing on the examinee's face image set to establish a preprocessed image; Calling the examinee data, establishing a verification data set based on the examinee data, performing feature extraction on the preprocessed image after performing face location, establishing a numerical vector of facial features, using the verification data set to verify and compare the numerical vector, and establishing authentication feedback; The certification feedback is received, and the candidate's examination management is performed according to the certification feedback.

Images