Face multi-feature fusion charging card sender identity identification authentication method and system

Abstract

The invention discloses a face multi-feature fusion charging card sender identification identity authentication method and system, and belongs to the technical field of card sender identification identity authentication. When it is detected that a collected person enters a collection area, after adaptive light source light supplement is carried out, parameters of a camera of a charging card sender are automatically adjusted; the position and the angle of the collected person are automatically detected, and a face image is collected; feature point detection is carried out on the collected face image, the face features of the face image are extracted, feature fusion is carried out on the face features, and a face feature vector is generated; performing identity authentication on the collected person based on the legal user face data set; and when the identity authentication of the collected person is passed, a driving instruction is sent to the card box of the charging card sender, the card box is controlled to pop out a card, and the identity of the card is dynamically bound according to the user identity information of the collected person. An efficient, accurate, safe and intelligent charging card sender identity authentication and card sending process is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of identity authentication for card issuing machines, and more specifically to a method and system for identity authentication of a charging card issuing machine with multi-feature fusion of human faces. Background Art

[0002] In the current era of rapid digital and intelligent development, various places have put forward higher and higher requirements for the accuracy and efficiency of personnel identity authentication and management. Especially in some scenarios where a large number of personnel need to be issued cards, charged, and have their identities recognized, such as enterprise parks, large factories, schools, etc., traditional identity authentication methods have exposed many drawbacks. In the past, relying solely on IC / ID cards for identity recognition, it was easy for cards to be lost or misappropriated, and the security was difficult to guarantee. Moreover, when faced with a large number of personnel flows, the card management work became extremely cumbersome and inefficient.

[0003] Chinese Patent with the authorization announcement number CN108921191B discloses a multi-biometric fusion recognition method based on image quality assessment. First, the face and iris images of the user are collected by the image acquisition subsystem and corresponding preprocessing is performed on them. Secondly, the preprocessed images are respectively sent to the corresponding recognition and authentication subsystems for steps such as quality assessment, feature extraction, and template matching, and their respective matching scores and corresponding matching quality confidence scores are output. Finally, these scores are normalized and sent to the recognition fusion subsystem, and a dynamic weighted fusion algorithm is used to obtain the recognition and authentication result.

[0004] Although the existing technology has a higher recognition accuracy than single-face or iris recognition algorithms and can achieve high-precision personal identity recognition; at the same time, based on the existing mainstream fusion recognition algorithms, useful information on image quality is extracted and applied to the fusion recognition and authentication process to further improve the recognition performance of the system.

[0005] However, it still fails to solve the problems of being difficult to stably collect high-quality face images under complex environmental lighting interference, unable to ensure the initiation of an efficient identity authentication process; the application of multi-feature fusion is not fine enough, and it is difficult to accurately and quickly screen out target users in a large sample legal user database, affecting the authentication efficiency and accuracy. Therefore, in order to overcome these limitations, the present invention proposes a method and system for identity authentication of a charging card issuing machine with multi-feature fusion of human faces. Summary of the Invention

[0006] Aiming at the deficiencies of the existing technology, the purpose of the present invention is to provide a method and system for identity authentication of a charging card issuing machine with multi-feature fusion of human faces, which solves the technical problems of high-quality acquisition of face images in complex environments, fine fusion of multi-features to accurately and quickly authenticate in a large sample legal user database, and dynamic binding of the charging card and user identity security after authentication.

[0007] To achieve the above object, the present invention provides the following technical solutions: A face multi-feature fusion charging card issuing machine identity authentication system, including a face acquisition module, a feature fusion module, an identity authentication module, and a card issuing control module; The face acquisition module is used to obtain the ambient light intensity through the light sensor of the charging card issuing machine when detecting that the person to be collected enters the collection area. After performing adaptive light source filling based on the ambient light intensity, automatically adjust the parameters of the charging card issuing machine camera, and automatically detect the position and angle of the person to be collected to collect a face image; The feature fusion module is used to detect feature points of the collected face image, extract facial features of the face image based on the feature points, including geometric features and texture features, and perform feature fusion on the facial features to generate a facial feature vector; The identity authentication module is used to authenticate the identity of the person to be collected based on the set of legitimate user face data. A preliminary set of legitimate users is quickly screened through the geometric feature vector of the person to be collected, and then an identity authentication set is screened out according to the step-by-step screening ratio of each downsampling scale and the similarity ranking of the texture feature vectors. Then, a key facial feature vector is constructed to screen out the target legitimate user to determine whether the identity authentication of the person to be collected is successful; The card issuing control module is used to send a driving instruction to the card magazine of the charging card issuing machine when the identity authentication of the person to be collected is passed, control the card magazine to eject the card, and dynamically bind the card identity according to the user identity information of the person to be collected.

[0008] Specifically, the steps of collecting a face image include: Continuously monitor the collection area based on the infrared sensor of the charging card issuing machine, obtain the object features in the collection area, and perform human body recognition based on the object features; When a human body is recognized, configure a monitoring time window for human body monitoring. During the monitoring time window, count the number of times a human body is recognized and calculate the human body recognition success rate; Configure a human body recognition threshold. When the human body recognition success rate is greater than the human body recognition threshold, start the face collection process, otherwise do nothing; Configure a collection trigger threshold. Obtain the collection distance between the person to be collected and the camera of the charging card issuing machine through the distance sensor, and obtain the collection distance change rate. When the collection distance change rate is less than the collection trigger threshold, trigger the adjustment of the fill light device parameters and the camera parameters; After adjusting the fill light device parameters and the camera parameters, capture the body contour information of the person to be collected through the camera, and extract the key point coordinates of the human body contour; According to the key point coordinates of the human body contour, calculate the horizontal offset and vertical offset of the head relative to the image center, and calculate the tilt angle of the head of the person to be collected; Configure the position deviation range and the angle deviation range. When the horizontal offset of the head relative to the image center and the vertical offset of the head relative to the image center are both within the position deviation range, and the tilt angle of the head is within the angle deviation range, then collect the face image. Otherwise, give a voice prompt according to the horizontal offset of the head relative to the image center, the vertical offset of the head relative to the image center, and the tilt angle of the head to guide the person being collected to adjust the position and angle.

[0009] Specifically, the steps of collecting the face image further include: Conduct a preliminary quality assessment on the collected face image to obtain the clarity of the face image, the integrity of the face image, and the brightness uniformity of the face image; The clarity of the face image is obtained by calculating the variance of the Laplacian operator of the image; The integrity of the face image is obtained by calculating the integrity and the missing ratio of the key points of the face contour; The brightness uniformity of the face image is obtained by calculating the brightness standard deviation of different regions of the face image; Configure the clarity threshold, the integrity threshold, and the brightness threshold. When the clarity of the face image is less than the clarity threshold, or the integrity of the face image is less than the integrity threshold, or the brightness uniformity of the face image is less than the brightness threshold, re-trigger the image collection. Otherwise, do not perform any operation; Configure the re-collection threshold, monitor the number of times of re-triggering the image collection. When the number of times of re-triggering the image collection is greater than the re-collection threshold, then pause the collection process and give a collection warning.

[0010] Specifically, the steps of adjusting the parameters of the supplementary lighting device include: Configure the light intensity threshold, including the upper limit threshold of the light intensity and the lower limit threshold of the light intensity, and obtain the ambient light intensity through the light sensor of the charging card reader; When the ambient light intensity is less than the lower limit threshold of the light intensity, enhance the supplementary lighting brightness and the supplementary lighting intensity, adjust the initial supplementary lighting brightness according to the ratio of the current ambient light intensity to the lower limit threshold of the light intensity, and optimize the supplementary lighting brightness in combination with the position of the face in the image; When the ambient light intensity is greater than or equal to the upper limit threshold of the light intensity, reduce the supplementary lighting brightness, and on the basis of the initial supplementary lighting intensity, reduce it proportionally as the degree of the light intensity exceeding the upper limit threshold of the light intensity until it reaches the minimum supplementary lighting brightness; After adjusting the supplementary lighting brightness, on the basis of the initial supplementary lighting angle, adjust the supplementary lighting angle according to the offset direction of the face center relative to the image center.

[0011] Specifically, the steps of adjusting the parameters of the camera include: Obtain the ambient light intensity after adjusting the parameters of the fill light device, and adjust the camera parameters, including sensitivity, focal length, aperture, and shutter speed; If, under the reduced fill light brightness of the fill light device, the ambient light intensity obtained by the light sensor of the charging card issuer is still greater than the upper threshold of the light intensity, then adjust the camera exposure parameters, and proportionally reduce the camera sensitivity according to the degree to which the ambient light intensity exceeds the upper threshold; Adjust the camera focal length according to the distance between the person being collected and the camera, and increase or decrease the camera focal length according to the change in distance based on the basic camera focal length; Based on the ambient light intensity, starting from the initial aperture, adjust according to the difference between the current ambient light intensity and the reference light intensity. When the ambient light intensity is higher than the reference light intensity, the camera aperture increases; conversely, the camera aperture decreases; Comprehensively adjust the camera shutter speed based on the ambient light intensity and the acquisition distance. On the basis of the initial shutter speed, when the ambient light intensity is greater than the reference light intensity, the shutter speed decreases proportionally, and the acquisition distance is inversely proportional to the shutter speed.

[0012] Specifically, the steps for generating the facial feature vector include: Obtain a face image, and perform feature point detection on the face image. The feature points include: eye feature points, nose feature points, mouth feature points, and facial contour feature points; Extract the geometric features of the face image based on the feature points of the face image, including: distance features, angle features, and ratio features; Select the diagonal length of the face image as the scale factor, and perform normalization processing on the distance features in the geometric features based on the scale factor to eliminate the influence of the face image size on the distance features; Sort the normalized distance features, angle features, and ratio features respectively in the order from top to bottom and from left to right of the facial image to construct a geometric distance feature vector, a geometric angle feature vector, and a geometric ratio feature vector; Perform feature fusion on the geometric features, and concatenate the geometric distance feature vector, the geometric angle feature vector, and the geometric ratio feature vector in sequence to construct a geometric feature vector; Perform grayscale processing on the face image, and perform multi-scale downsampling on the grayscale-processed face image to obtain the face image at each downsampling scale for texture feature extraction to obtain a texture feature vector; Concatenate the geometric feature vector and the texture feature vector to obtain a facial feature vector.

[0013] Specifically, the steps for obtaining the texture feature vector include: Select key feature points based on the feature points, and perform key feature point detection on the face image at each downsampling scale; Set the neighborhood window size according to the downsampling scale. Taking the detected key feature points as the center, within the neighborhood window, calculate the gradient direction and gradient magnitude of each pixel point relative to the key feature points; Divide the neighborhood window into regions to obtain neighborhood sub-regions. In each neighborhood sub-region, count the gradient direction histogram, and connect the gradient direction histograms of all neighborhood sub-regions to form the texture feature of the key feature points; For each downsampling scale, construct the corresponding sampling scale texture feature vector according to the type of key feature points. Arrange the texture feature vectors of each sampling scale in ascending order of the downsampling scale, and splice the texture feature vectors of each sampling scale to obtain the texture feature vector.

[0014] Specifically, the specific steps of identity authentication include: Let the set of legitimate user face data be , where is the face data of the th legitimate user, including the geometric feature vector of the th legitimate user, the texture feature vector at each downsampling scale, and the facial feature vector; Configure the fast screening ratio, obtain the geometric feature vector of the person being collected, perform fast screening on the set of legitimate user face data, and calculate the similarity between the person being collected and the geometric feature vectors of each legitimate user in the face database set; Sort the similarities between the geometric feature vectors of the collector and the legitimate users in the face database set, and obtain the preliminary set of legitimate users according to the fast screening ratio; Configure the step-by-step screening ratio for each downsampling scale, obtain the texture feature vector of the person being collected at each downsampling scale, perform step-by-step screening on the preliminary set of legitimate users, and calculate the similarity between the person being collected and the texture feature vectors of each legitimate user at the downsampling scale in the preliminary set of legitimate users obtained by fast screening in turn; Sort the similarities of the texture feature vectors at each downsampling scale in turn, and obtain the step-by-step screened set of legitimate users according to the step-by-step screening ratio for each downsampling scale to obtain the identity authentication set.

[0015] Specifically, the specific steps of identity authentication also include: Configure the similarity threshold, obtain the facial feature vector of the person being collected, perform the final identity authentication on the person being collected, and determine the importance score of each facial feature vector of each legitimate user in the identity authentication set and the facial feature vector of the person being collected according to the contribution degree of each vector element in the facial feature vector in identity recognition, and perform importance sorting; Configure the importance threshold , after the importance sorting is completed, according to the importance threshold, select the first Construct a key facial feature vector from the vector elements of a single vector; Calculate the similarity between the key facial feature vector of each legitimate user in the identity authentication set and the key facial feature vector of the person being collected according to the importance score of each vector element in the key facial feature vector; Screen the legitimate user in the identity authentication set with the highest similarity to the key facial feature vector of the person being collected, and mark it as the target legitimate user. If the similarity between the target legitimate user and the key facial feature vector of the person being collected is greater than the similarity threshold, the identity authentication is successful; otherwise, the identity authentication fails. Configure the identity authentication threshold, record the number of times the identity of the person being collected is authenticated. When the number of times of identity authentication is greater than the identity authentication threshold, an identity authentication warning is issued; otherwise, no operation is performed.

[0016] A method for identifying and authenticating identities of a charging card issuer with multi-feature fusion of human faces, including the following steps: When it is detected that the person being collected enters the collection area, obtain the ambient light intensity through the light sensor of the charging card issuer. After performing adaptive light source filling based on the ambient light intensity, automatically adjust the camera parameters of the charging card issuer, and automatically detect the position and angle of the person being collected to collect a face image; Perform feature point detection on the collected face image, extract the facial features of the face image based on the feature points, including geometric features and texture features, and perform feature fusion on the facial features to generate a facial feature vector; Authenticate the identity of the person being collected based on the set of legitimate user face data. Quickly screen out a preliminary set of legitimate users through the geometric feature vector of the person being collected, and then screen out the identity authentication set according to the step-by-step screening ratio of each downsampling scale and the similarity ranking of the texture feature vectors. Then construct a key facial feature vector and screen out the target legitimate user to determine whether the identity authentication of the person being collected is successful; When the identity authentication of the person being collected is passed, send a drive instruction to the card magazine of the charging card issuer to control the card magazine to eject the card, and dynamically bind the card identity according to the user identity information of the person being collected.

[0017] The beneficial effects of the present invention: 1. Through infrared sensors, distance sensors, etc., accurately identify the person being collected entering the collection area, automatically perform adaptive light source filling according to the ambient light intensity, and at the same time automatically detect the position and angle of the person being collected, and automatically adjust the camera parameters of the charging card issuer to ensure that appropriate face images can be collected in various environments and postures of the person being collected, improving the intelligence and accuracy of the collection and reducing the collection failures caused by lighting and position angle problems.

[0018] 2. Extract the geometric features and texture features of the face image, and perform feature fusion to generate a facial feature vector. The identity authentication module performs multi-stage identity authentication based on this, including rapid screening of the geometric feature vector, gradual screening of the texture feature vector, and construction and similarity calculation of the key facial feature vector, etc. It makes full use of various feature information of the face and greatly improves the accuracy and reliability of identity authentication compared with single-feature authentication.

[0019] 3. Adopt a phased screening method. First, quickly narrow down the range, and then gradually refine the screening. While ensuring the accuracy of authentication, it improves the authentication efficiency, reduces unnecessary computational amount and time consumption, can quickly and accurately determine the identity of the person being collected, and enhances the user experience and the overall performance of the charging card issuing machine. Brief Description of the Drawings

[0020] Figure 1 It is a schematic structural diagram of the identity authentication system for the charging card issuing machine with multi-feature fusion of human faces according to the present invention;

[0021] Figure 2 It is a flowchart of the specific steps for collecting a face image according to the present invention;

[0022] Figure 3 It is a schematic diagram of the specific steps for adjusting the parameters of the supplementary light device according to the present invention;

[0023] Figure 4 It is a flowchart of the specific steps for adjusting the parameters of the camera according to the present invention;

[0024] Figure 5 It is a flowchart of the specific steps for generating a facial feature vector according to the present invention;

[0025] Figure 6 It is a flowchart of the specific steps for identity authentication according to the present invention;

[0026] Figure 7 It is a flowchart of the method for identity authentication of the charging card issuing machine with multi-feature fusion of human faces according to the present invention. Detailed Description of the Invention

[0027] Embodiment 1 Please refer to Figure 1 , this embodiment introduces an identity authentication system for the charging card issuing machine with multi-feature fusion of human faces, including a face collection module, a feature fusion module, an identity authentication module, and a card issuing control module; The face collection module is used to obtain the ambient light intensity through the light sensor of the charging card issuing machine when it detects that the person being collected enters the collection area. After performing adaptive light source supplementary lighting based on the ambient light intensity, it automatically adjusts the parameters of the camera of the charging card issuing machine, and collects a face image by automatically detecting the position and angle of the person being collected; In this embodiment, when it is detected that the person to be collected enters the collection area, accurate ambient light intensity data is obtained by means of the light sensor built in the charging card issuing machine. If the light intensity is lower than the set threshold, the supplementary light device is automatically turned on, and the supplementary light brightness and angle are adjusted through an intelligent algorithm to ensure that the light evenly covers the human face and avoid the generation of shadows; if the light is too strong, the supplementary light brightness is appropriately reduced or the exposure parameters of the camera are adjusted to ensure that the brightness and contrast of the collected image are appropriate. In addition, the parameters of the camera of the charging card issuing machine are automatically adjusted, including adjusting the focal length according to the collection distance to ensure that the human face occupies an appropriate proportion in the image. Generally, the height of the human face accounts for about one-third to one-half of the image height; at the same time, according to the ambient light and shooting requirements, the aperture is adjusted to control the depth of field, and an appropriate shutter speed is set to avoid image blurring. It also has the function of automatically detecting the position and angle. The body contour information of the person to be collected is captured by the camera, and the position and angle of the person to be collected are monitored and judged in real time. Once a position or angle deviation is detected, the system will guide the person to be collected to adjust to the best position through voice prompts or guiding marks on the screen. Finally, a clear and complete human face image is collected.

[0028] Please refer to Figure 2 , preferably, the specific steps for collecting the human face image include: Based on the infrared sensor of the charging card issuing machine, the collection area is continuously monitored to obtain the object characteristics in the collection area, including the object shape, object size, object movement speed and object movement trajectory. Human body recognition is performed on the object characteristics through a machine learning algorithm; accurately identify whether the object entering the area is a human body, so as to provide a judgment basis for whether to start the human face collection process subsequently, avoid mis-triggering the collection process, and improve the pertinence of the collection.

[0029] When a human body is recognized, a monitoring time window is configured, and human body monitoring is continuously performed. During the monitoring time window, the number of times the human body is recognized is counted , and the human body recognition success rate is calculated, that is:

[0030] Among them, is the human body recognition success rate within the monitoring time window, is the total number of detections within the monitoring time window; Configure the human body recognition threshold , when the human body recognition success rate is greater than the human body recognition threshold , it indicates that a human body has entered the collection area, and the human face collection process is started. Otherwise, no operation is performed, and the area monitoring continues; avoiding the wrong start of the collection process due to the short-term or accidental appearance of the human body, improving the reliability of the judgment for starting the human face collection process.

[0031] Configure the collection trigger threshold , obtain the acquisition distance between the person being collected and the camera of the charging card issuing machine through a distance sensor, and obtain the acquisition distance change rate, that is:

[0032] wherein, is the acquisition distance change rate at time , is the acquisition distance at time , is the time interval for obtaining the acquisition distance between the person being collected and the camera of the charging card issuing machine. When the acquisition distance change rate is less than the acquisition trigger threshold, trigger the adjustment of the fill light device parameters and the camera parameters; be able to adjust the device parameters in a timely manner according to the movement state of the person being collected to ensure parameter optimization at an appropriate time and prepare for obtaining high-quality face images.

[0033] Please refer to Figure 3 , the steps of adjusting the fill light device parameters include: configuring the light intensity threshold, including the upper limit threshold and the lower limit threshold of the light intensity, and obtaining the ambient light intensity through the light sensor of the charging card issuing machine; When the ambient light intensity is less than the lower limit threshold of the light intensity, turn on the fill light device, and adjust the initial fill light brightness according to the ratio of the current ambient light intensity to the lower limit threshold of the light intensity. The weaker the light intensity, the stronger the fill light brightness, and optimize the fill light brightness in combination with the position of the face in the image. The farther the face deviates from the center of the image, the stronger the fill light brightness, so as to ensure uniform overall illumination, that is:

[0034] wherein, is the enhanced fill light brightness of the fill light device when enhancing the fill light brightness, is the initial fill light brightness, is the ambient light intensity obtained by the light sensor of the charging card issuing machine, is the lower limit threshold of the light intensity, is an adjustment coefficient related to the environment, and its value range is (0.1, 1), and are the offsets of the face center and the image center in the horizontal and vertical directions, is the diagonal length of the image; not only considers the difference in light intensity, but also combines the position of the face in the image to more precisely adjust the fill light brightness.

[0035] When the ambient light intensity is greater than or equal to the upper limit threshold of the light intensity, that is, when the light is too strong, reduce the fill light brightness. On the basis of the initial fill light intensity, reduce it proportionally as the light intensity exceeds the upper limit threshold of the light intensity until it reaches the minimum fill light brightness, that is:

[0036] Among them, is the supplementary lighting brightness after the supplementary lighting device reduces the brightness when reducing the supplementary lighting brightness, is the upper limit threshold of the light intensity, is the minimum supplementary lighting brightness of the supplementary lighting device, is the adjustment coefficient of the supplementary lighting intensity when the light is too strong, and the value range is (0.1, 1); After adjusting the supplementary lighting brightness, on the basis of the initial supplementary lighting angle, adjust the supplementary lighting angle according to the offset direction of the face center relative to the image center, so that the supplementary lighting can more accurately cover the face and avoid generating shadows, that is:

[0037] Among them, is the adjusted supplementary lighting angle, is the initial supplementary lighting angle, and are the offset amounts of the face center and the image center in the horizontal and vertical directions, is the angle adjustment coefficient, and the value range is (0.1, 1); Please refer to Figure 4 , the steps of adjusting the camera parameters include: obtaining the ambient light intensity after adjusting the parameters of the supplementary lighting device to adjust the camera parameters, including sensitivity, focal length, aperture and shutter speed; If, under the supplementary lighting brightness after the supplementary lighting device reduces, the ambient light intensity obtained by the light sensor of the charging card issuer is still greater than the upper limit threshold of the light intensity, then adjust the camera exposure parameters, and reduce the camera sensitivity proportionally according to the degree to which the ambient light intensity exceeds the upper limit threshold, that is:

[0038] Among them, is the adjusted sensitivity, is the initial sensitivity, is the ambient light intensity after adjusting the parameters of the supplementary lighting device, is the sensitivity adjustment coefficient, and the value range is (0.1, 1); Adjust the camera focal length according to the distance between the person being collected and the camera. On the basis of the basic focal length, increase or decrease according to a certain rule according to the change of the distance. The farther the distance, the greater the focal length adjustment amount, so as to ensure that the face can maintain an appropriate proportion in the image, and the face height accounts for about one-third to one-half of the image height, that is:

[0039] Among them, is the adjusted camera focal length, is the initial camera focal length, is the focal length adjustment coefficient, and its value range is (0.1, 1), is the distance between the person being captured and the camera; The aperture is adjusted according to the ambient light intensity. Based on the initial aperture, it is adjusted according to the difference between the current ambient light intensity and the reference light intensity. When the ambient light intensity is higher than the reference light intensity, the camera aperture increases; conversely, the camera aperture decreases, so as to control the depth of field and ensure that the clarity of the face and background in the image meets the requirements, that is:

[0040] Among them, is the adjusted aperture, is the initial aperture, is the reference light intensity corresponding to the initial aperture, is the ambient light intensity after the adjustment of the fill light device parameters, is the aperture adjustment coefficient, and its value range is (0.1, 1), so as to dynamically adjust the aperture size according to the difference between the ambient light intensity and the reference light intensity and control the depth of field.

[0041] The camera shutter speed is adjusted by comprehensively considering the ambient light intensity and the acquisition distance. Based on the initial shutter speed, when the ambient light intensity is greater than the reference light intensity, the shutter speed decreases proportionally; the farther the acquisition distance is, the shutter speed decreases proportionally, so as to avoid image blurring caused by too strong light or the movement of the person being captured, that is:

[0042] Among them, is the adjusted shutter speed, is the initial shutter speed, is the reference light intensity corresponding to the initial shutter speed, is the ambient light intensity after the adjustment of the fill light device parameters, is the distance between the person being captured and the camera, is the shutter speed adjustment coefficient, which comprehensively considers the influence of the light intensity and the acquisition distance on the shutter speed to avoid image blurring.

[0043] After the adjustment of the fill light device parameters and the camera parameters, the camera captures the body contour information of the person being captured, and uses image processing algorithms to monitor and judge the position and angle of the person being captured in real time, and extracts the key point coordinates of the human contour, including the head key points, shoulder key points and waist key points; According to the key point coordinates of the human contour, calculate the horizontal offset and vertical offset of the head relative to the image center, and calculate the tilt angle of the head of the person being captured , that is:

[0044]

[0045]

[0046] Among them, is the horizontal offset of the head relative to the center of the image, is the vertical offset of the head relative to the center of the image, are the coordinates of the head key points, is the coordinate of the center of the image; Configure the position deviation range and the angle deviation range. When the horizontal offset of the head relative to the center of the image and the vertical offset of the head relative to the center of the image are within the position deviation range, and the tilt angle of the head is within the angle deviation range, then face image acquisition is performed. Otherwise, according to the horizontal offset of the head relative to the center of the image, the vertical offset of the head relative to the center of the image, and the tilt angle of the head, voice prompts are given to guide the person being collected to adjust the position and angle. The voice prompts include "Please move a little to the left" and "Please raise your head slightly"; Perform a preliminary quality assessment on the collected face image to obtain the clarity, integrity, and brightness uniformity of the face image. Among them, the clarity of the face image is obtained by calculating the variance of the Laplacian operator of the image, that is:

[0047] Among them, is the face image clarity, is the Laplacian operator, is the variance function; The integrity of the face image is obtained by calculating the integrity and missing ratio of the key points of the face contour, that is:

[0048] Among them, is the face image integrity, is the number of key points detected in the face image of, is the total number of key points of the face contour; The brightness uniformity of the face image is obtained by calculating the brightness standard deviation of different regions of the face image. The face image is divided into multiple regions, and the brightness of each region of the face image and the average brightness of the face image are obtained. The smaller the brightness standard deviation of the face image region, the more uniform the brightness, that is:

[0049] Among them, is a face image brightness uniformity is the number of divided regions of the face image is the brightness of the nth region is the average brightness of the face image; Configure sharpness threshold, integrity threshold, and brightness threshold. When the sharpness of the face image is less than the sharpness threshold, or the integrity of the face image is less than the integrity threshold, or the brightness uniformity of the face image is less than the brightness threshold, re-trigger image acquisition; otherwise, do nothing. The acquired face image meets certain quality standards, improving the reliability and usability of face image data, which is beneficial to subsequent applications such as identity authentication.

[0050] Configure the re-acquisition threshold, monitor the number of times of re-triggering image acquisition. When the number of times of re-triggering image acquisition is greater than the re-acquisition threshold, pause the acquisition process, issue an acquisition warning, and prompt the user to check their own status and environmental conditions. This avoids wasting resources due to continuous acquisition of low-quality images, and at the same time reminds the user to pay attention to factors that may affect the acquisition quality, helping the user to adjust in time to ensure the smooth progress of the acquisition work.

[0051] The feature fusion module is used to detect feature points of the acquired face image, extract facial features of the face image based on the feature points, including geometric features and texture features, and perform feature fusion on the facial features to generate a facial feature vector; In this embodiment, from the acquired face image, the feature fusion module uses a variety of advanced feature extraction algorithms to extract different types of facial features respectively. Measure and record the key geometric features of the face, such as the distance between the two eyes, the length and width of the nose, the contour and position of the mouth, etc. These features constitute the basic shape framework of the face and are one of the important bases for recognition. Using texture analysis technology, extract the texture information on the surface of the face skin, such as details like wrinkles and pores. These texture features are unique to individuals, and even identical twins have differences, providing more detailed information for identity recognition. Fuse the extracted geometric features and texture features, and finally generate a comprehensive facial feature vector. This facial feature vector contains multi-dimensional feature information of the face, greatly improving the accuracy and reliability of identity recognition.

[0052] Please refer to Figure 5 , preferably, the specific steps for generating the facial feature vector include: Obtain a face image, and perform feature point detection on the face image through a deep learning network. The feature points include: eye feature points, nose feature points, mouth feature points, and facial contour feature points; the eye feature points include the eye corner points, pupil center points, pupil edge points, and eyelid edge points; the nose feature points include the nose tip point, nose wing edge points, and nose bridge points; the mouth feature points include the mouth corner points, lip peak points, and lip valley points; the facial contour feature points include the chin vertex point, cheek edge points, and forehead edge points; it provides an accurate position identifier for the geometric feature extraction and texture feature analysis of the face, which is the basis for subsequent in-depth analysis of face features and helps to accurately depict the shape and structure of the face.

[0053] Extract the geometric features of the face image based on the feature points of the face image, including: distance features, angle features, and proportion features; the distance features include the distance between two eyes, the distance between the inner canthi of the eyes, the distance between the outer canthi of the eyes, the height of the eyes, the length of the nose, the width of the nose wings, the width of the mouth, the thickness of the lips, the length of the face, and the width of the face; the angle features include the eye tilt angle, the nose bridge tilt angle, and the mandibular angle; the proportion features include the eye-nose proportion, the mouth-nose proportion, the three-part proportion of the face, and the five-eye proportion; it constructs the basic framework of the face shape and provides an important shape basis for recognition.

[0054] Select the diagonal length of the face image as the scale factor, and perform normalization processing on the distance features in the geometric features based on the scale factor to eliminate the influence of the face image size on the distance features, that is:

[0055] where, is the distance feature of the th face image after normalization processing, is the distance feature of the th face image, The value range of is is the total number of distance features of the face image, is the diagonal length of the face image; it eliminates the influence of the face image size difference caused by factors such as shooting distance and image scaling on the distance features, making the distance features in different images comparable, and improving the stability and reliability of the geometric features in the subsequent fusion and recognition processes.

[0056] Sort the normalized distance features, angle features, and proportion features respectively in the order from top to bottom and from left to right of the facial image to construct a geometric distance feature vector, a geometric angle feature vector, and a geometric proportion feature vector. Perform feature fusion on geometric features, concatenate the geometric distance feature vector, geometric angle feature vector, and geometric ratio feature vector in sequence to construct a geometric feature vector, enabling geometric features to be presented in a structured form, facilitating subsequent fusion with other features and utilization by recognition algorithms, while retaining the information on the geometric relationships of various parts of the human face.

[0057] Perform grayscale processing on the face image, and perform multi-scale downsampling on the grayscale processed face image to obtain face images at each downsampling scale. Images at different scales can capture different levels of detailed information. Small-scale images contain more global features, while large-scale images retain more local details, providing a rich data basis for comprehensively extracting texture features.

[0058] Select key feature points based on feature points, including eye corner points, nose tip points, nose wing edge points, mouth corner points, lip peak points, lip valley points, chin vertex points, cheek edge points, and forehead edge points, and perform key feature point detection on the face image at each downsampling scale. Set the neighborhood window size according to the downsampling scale. Taking the detected key feature points as the center, calculate the gradient direction and gradient magnitude of each pixel point relative to the key feature points within the neighborhood window, highlighting the detailed features of the face texture, such as wrinkles and pores, providing key information for the extraction of texture features.

[0059] Divide the neighborhood window into regions to obtain neighborhood sub-regions. Statistically calculate the gradient direction histogram within each neighborhood sub-region, and connect the gradient direction histograms of all neighborhood sub-regions to form the texture features of the key feature points. For each downsampling scale, construct the corresponding sampling scale texture feature vector according to the type of key feature points. Arrange the texture feature vectors of each sampling scale in ascending order of the downsampling scale and concatenate them to obtain the final texture feature vector. Concatenate the geometric feature vector and the texture feature vector to obtain a facial feature vector. This facial feature vector combines the geometric features and texture features of the human face, contains rich human face information, and provides an important feature basis for subsequent applications such as face recognition and identity authentication.

[0060] The identity authentication module is used to authenticate the person being collected based on the set of legal user face data. A preliminary set of legal users is quickly screened through the geometric feature vector of the person being collected, and then the identity authentication set is screened out according to the step-by-step screening ratio and texture feature vector similarity ranking at each downsampling scale. Then, a key facial feature vector is constructed to screen out the target legal users to determine whether the identity authentication of the person being collected is successful. In this embodiment, the identity authentication module compares the geometric feature vector generated by the feature fusion module, the texture feature vector at each downsampling scale, and the facial feature vector with the set of legitimate user face data of the registered users stored in advance. The feature vectors stored in the database are generated through a strict acquisition and feature extraction process during user registration, and each vector is associated with the identity information of a specific user. Through a multi-level fast authentication process, the similarity between the person being collected and the known users in the set of legitimate user face data is obtained to determine whether the identity authentication is successful.

[0061] Please refer to Figure 6 , preferably, the specific steps of identity authentication include: Let the set of legitimate user face data be , where is the face data of the th legitimate user, including the geometric feature vector, the texture feature vector at each downsampling scale, and the facial feature vector of the th legitimate user, that is: , where is the geometric feature vector of the th legitimate user, is the texture feature vector of the th legitimate user at the downsampling scale , is the number of downsampling scales, is the facial feature vector of the th legitimate user, is the number of legitimate users; clarify the data basis for identity authentication, and structurally store the face data of legitimate users to provide a complete information source for the subsequent authentication process, including multi-dimensional information such as geometric features, texture features, and facial features, to ensure that the authentication process has comprehensive data support.

[0062] Configure the fast screening ratio , obtain the geometric feature vector of the person being collected, quickly screen the set of legitimate user face data, and calculate the similarity between the geometric feature vector of the person being collected and the geometric feature vector of each legitimate user in the face database set. Exemplarily, for the geometric feature vector similarity of the th legitimate user in the set of legitimate user face data, the calculation formula is:

[0063] where is the similarity between the geometric feature vectors of the person being collected and the th legitimate user, is the geometric feature vector of the th legitimate user, is the geometric feature vector of the person being collected, is the th element in the geometric feature vectors of the th legitimate user, is the th element in the geometric feature vector of the person being collected, is the number of elements in the geometric feature vector; By quickly calculating the similarity between the geometric feature vectors of the person being collected and legitimate users, and according to the quick screening ratio, a smaller preliminary set of legitimate users is quickly selected from the large-scale legitimate user face database, significantly narrowing the scope of subsequent processing, improving the preliminary efficiency of authentication, and reducing unnecessary computational effort.

[0064] Sort the similarities between the geometric feature vectors of the collector and legitimate users in the face database set, and according to the quick screening ratio , obtain the preliminary set of legitimate users, that is:

[0065] Among them, is the face data of the th legitimate user in the face database set, ranges from , is to take the preliminary set of legitimate users, is the similarity between the person being collected and the geometric feature vectors of the th legitimate user rank, is the ceiling function; Configure the step-by-step screening ratio for each downsampling scale, that is: Among them, is the step-by-step screening ratio when the downsampling scale is , is the number of downsampling scales, obtain the texture feature vectors of the person being collected at each downsampling scale, and perform step-by-step screening on the preliminary set of legitimate users. Calculate the similarities between the texture feature vectors of the person being collected and each legitimate user in the preliminary set of legitimate users obtained by quick screening at each downsampling scale. Exemplarily, for the texture feature vector similarity of the first legitimate user in the preliminary set of legitimate users at the downsampling scale , the calculation formula is:

[0066] Among them, is the similarity between the person being collected and the texture feature vector of the first legitimate user in the preliminary set of legitimate users at the downsampling scale , is the similarity, is the person being collected at the downsampling scale The texture feature vector under is the first legal user in the preliminary legal user set at the downsampling scale The th element of the texture feature vector, is the th element of the texture feature vector of the person being collected at the downsampling scale ; the value range of is , is the number of elements of the texture feature vector at the downsampling scale ; by using the texture feature vectors of different downsampling scales, calculate the similarity in sequence and perform multiple rounds of screening according to the step-by-step screening ratio, gradually and precisely narrowing the range of legal users, making the final obtained identity authentication set more targeted and accurate, further excluding users with low similarity to the person being collected, and improving the reliability of the authentication result.

[0067] Sort the similarity of the texture feature vectors of each downsampling scale in sequence, and obtain the step-by-step screening legal user set according to the step-by-step screening ratio of each downsampling scale to obtain the identity authentication set, that is:

[0068] Among them, is the identity authentication set, is the th legal user's face data in the step-by-step screening legal user set when the downsampling scale is , ; the value range of is , is the number of legal users in the step-by-step screening legal user set when the downsampling scale is , ; the value range of is , when , , is the similarity between the person being collected and the th legal user in the step-by-step screening legal user set at the downsampling scale of the texture feature vector rank, is the step-by-step screening ratio when the downsampling scale is , gradually narrowing the range of target legal users, is the ceiling function; Configure a similarity threshold, obtain the facial feature vector of the person being collected, perform final identity authentication on the person being collected, compare the facial feature vector of each legitimate user in the identity authentication set with the facial feature vector of the person being collected, determine the importance score according to the contribution degree of each vector element in the facial feature vector to identity recognition, and perform importance ranking. Exemplarily, the importance score of each vector element can be obtained through machine learning algorithms, including random forests, gradient boosting trees, etc., and then sorted according to the score; by determining and ranking the contribution degree of each element in the facial feature vector to identity recognition, the key features are highlighted, making the subsequent similarity calculation more focused on the features that have an important impact on identity recognition, and improving the accuracy and effectiveness of authentication.

[0069] Configure an importance threshold , after the importance ranking is completed, according to the importance threshold, select the first vector elements in each facial feature vector to construct a key facial feature vector; selecting key elements according to the importance threshold to construct a key facial feature vector simplifies the data dimension, reduces the computational complexity, and at the same time retains the most representative feature information, which helps to perform similarity comparison more efficiently.

[0070] Calculate the similarity between the key facial feature vector of each legitimate user in the identity authentication set and the key facial feature vector of the person being collected according to the importance score of each vector element in the key facial feature vector. Exemplarily, the calculation formula for the similarity of the key facial feature vector of the first legitimate user in the identity authentication set is:

[0071] where is the similarity between the person being collected and the key facial feature vector of the first legitimate user in the identity authentication set, is the importance score of the th vector element of the key facial feature vector, ranges from , is the key facial feature vector of the first legitimate user in the identity authentication set, is the key facial feature vector of the person being collected, is the th element of the key facial feature vector of the first legitimate user in the identity authentication set, is the th element of the key facial feature vector of the person being collected; calculating the similarity based on the importance score, screening out the legitimate user most similar to the person being collected as the target legitimate user, and accurately determining the identity authentication result by comparing with the similarity threshold, realizing the accurate identification and confirmation of the identity of the person being collected.

[0072] Screen the legal users in the identity authentication set with the highest similarity to the key facial feature vector of the person being collected, and mark them as the target legal users. If the similarity between the target legal user and the key facial feature vector of the person being collected is greater than the similarity threshold, the identity authentication is successful, and the identity of the person being collected is the target legal user; otherwise, the identity authentication fails, and a warning for authentication failure is issued to remind the person being collected to re-collect the face image. If the authentication fails, timely warning and reminder to the person being collected to re-collect the face image helps to improve the success rate of the next authentication. If the authentication is successful, the identity of the person being collected is clarified, ensuring the normal process of the system and the legitimate rights and interests of users.

[0073] Configure the identity authentication threshold, record the number of identity authentication times of the person being collected. When the number of identity authentication times is greater than the identity authentication threshold, conduct an identity authentication warning, perform operations such as sending a notice to the administrator, temporarily locking the user account, and displaying a more stringent security verification prompt to prevent potential malicious attacks or abnormal situations. At the same time, additional identity verification methods can also be provided for the person being collected to ensure the security and reliability of the system; otherwise, no operation is performed.

[0074] The card issuing control module is used to send a driving instruction to the card magazine of the charging card issuing machine when the identity authentication of the person being collected is passed, control the card magazine to eject the card, and dynamically bind the card identity according to the user identity information of the person being collected; In this embodiment, the card issuing control module performs corresponding operations according to the results of the identity authentication module. When the identity authentication is successful, the card issuing control module immediately starts the card issuing process. It first sends an instruction to the card magazine of the charging card issuing machine to drive the mechanical device in the card magazine to eject a charging card. At the same time, write the user's identity information, such as name, ID number, and relevant usage permission information, into the chip of the charging card. During the card issuing process, display a warm reminder on the display screen to inform the user to take the card, and briefly explain the usage method and precautions of the charging card in the form of animation or text, such as the location of the charging interface, charging operation steps, return time requirements, etc. If the identity authentication fails, the card issuing control module will not perform the card issuing operation. At this time, clear prompt information will be displayed on the display screen. At the same time, it is also responsible for monitoring and maintaining the hardware equipment of the charging card issuing machine to ensure the normal operation of components such as the card issuing device and the card magazine, and timely discover and handle abnormal situations such as paper jams and no cards in the card magazine.

[0075] Preferably, the specific steps for dynamically binding the card identity include: Receive the identity information of the person being collected who has passed the authentication, including name, ID number, user account, user level, and user permissions; Read the unique identifier of the charging card to be issued, including the card number and chip serial number, to ensure that each card has an independent identifier for subsequent binding and management; Send an activation signal to the recharge card to be issued through the radio frequency identification or near field communication interface of the recharge card issuing machine, so that the card enters a readable and writable state and is ready to receive data; Adopt an encryption protocol to establish a secure communication channel between the recharge card issuing machine and the recharge card to be issued, ensuring the confidentiality and integrity during data transmission and preventing information from being stolen or tampered with; In the chip storage area of the recharge card to be issued, according to the preset storage structure, divide a dedicated area for storing user identity information, and perform initialization and formatting operations to ensure the accuracy and consistency of data writing; Write the obtained user identity information byte by byte into the chip of the recharge card according to the preset encoding format and storage location; for example, the name is stored in the first 10 bytes, and the ID number is stored in the next 20 bytes, etc.; After the user identity information is written, read the written information of the recharge card to be issued, compare it with the user identity information, perform data verification, and check for data errors or losses. For example, use methods such as checksum and CRC for data integrity verification; If the data verification passes, send a confirmation instruction to the recharge card to be issued, and the recharge card to be issued returns a confirmation signal, indicating that the identity information and permission information have been successfully bound; Record the relevant information of this binding operation, including user identity, card identifier, binding time, etc., for subsequent query and management; Mark the status of the issued cards in the card magazine of the recharge card issuing machine as bound and issued, and update the quantity and position information of the remaining cards in the card magazine to monitor the inventory of the card magazine in real time.

[0076] Embodiment 2 Please refer to Figure 7 , this embodiment introduces a method for identity authentication of a recharge card issuing machine with multi-feature fusion of human faces, including the following steps: When it is detected that the person to be collected enters the collection area, obtain the ambient light intensity through the light sensor of the recharge card issuing machine. After performing adaptive light source supplementary lighting based on the ambient light intensity, automatically adjust the parameters of the camera of the recharge card issuing machine, and automatically detect the position and angle of the person to be collected to collect a face image; Perform feature point detection on the collected face image, extract facial features of the face image based on the feature points, including geometric features and texture features, and perform feature fusion on the facial features to generate a facial feature vector; Authenticate the identity of the person being collected based on the face data set of legal users. Quickly screen the initial legal user set through the geometric feature vector of the person being collected, and then screen out the identity authentication set according to the step-by-step screening ratio and texture feature vector similarity ranking of each downsampling scale. Then construct the key facial feature vector, screen out the target legal users, and determine whether the identity authentication of the person being collected is successful; When the identity authentication of the person being collected is passed, send a drive instruction to the card magazine of the charging card issuer, control the card magazine to eject the card, and dynamically bind the card identity according to the user identity information of the person being collected.

[0077] Specifically, the steps for generating the facial feature vector include: Obtain the face image, detect the feature points of the face image, and the feature points include: eye feature points, nose feature points, mouth feature points, and facial contour feature points; Extract the geometric features of the face image based on the feature points of the face image, including: distance features, angle features, and ratio features; Select the diagonal length of the face image as the scale factor, and normalize the distance features in the geometric features based on the scale factor to eliminate the influence of the face image size on the distance features; Sort the normalized distance features, angle features, and ratio features respectively in the order from top to bottom and from left to right of the facial image to construct the geometric distance feature vector, geometric angle feature vector, and geometric ratio feature vector; Perform feature fusion on the geometric features, splice the geometric distance feature vector, geometric angle feature vector, and geometric ratio feature vector in sequence to construct the geometric feature vector; Perform grayscale processing on the face image, and perform multi-scale downsampling on the grayscale processed face image to obtain the face image of each downsampling scale for texture feature extraction to obtain the texture feature vector; Splice the geometric feature vector and the texture feature vector to obtain the facial feature vector.

[0078] Specifically, the steps for obtaining the texture feature vector include: Select key feature points based on the feature points, and detect the key feature points of the face image of each downsampling scale; Set the neighborhood window size according to the downsampling scale, and calculate the gradient direction and gradient amplitude of each pixel point relative to the key feature point within the neighborhood window with the detected key feature point as the center; Divide the neighborhood window into regions to obtain neighborhood sub-regions, and count the gradient direction histograms within each neighborhood sub-region, and connect the gradient direction histograms of all neighborhood sub-regions to form the texture features of the key feature points; For each downsampling scale, according to the type of key feature points, construct the corresponding sampling scale texture feature vector. Arrange the texture feature vectors of each sampling scale in ascending order of the downsampling scale and splice them to obtain the texture feature vector.

[0079] Specifically, the specific steps of identity authentication include: Let the set of face data of legitimate users be , where is the face data of the th legitimate user, including the geometric feature vector of the th legitimate user, the texture feature vector at each downsampling scale, and the facial feature vector; Configure the quick screening ratio, obtain the geometric feature vector of the person being collected, quickly screen the set of face data of legitimate users, and calculate the similarity between the person being collected and the geometric feature vectors of each legitimate user in the face database set; Sort the similarities between the geometric feature vectors of the person being collected and the legitimate users in the face database set, and obtain the preliminary set of legitimate users according to the quick screening ratio; Configure the step-by-step screening ratio for each downsampling scale, obtain the texture feature vector of the person being collected at each downsampling scale, perform step-by-step screening on the preliminary set of legitimate users, and calculate the similarity between the person being collected and the texture feature vectors of each legitimate user at the downsampling scale in the quickly screened preliminary set of legitimate users in turn; Sort the similarities of the texture feature vectors at each downsampling scale in turn, and obtain the step-by-step screened set of legitimate users according to the step-by-step screening ratio for each downsampling scale to obtain the identity authentication set.

[0080] Specifically, the specific steps of identity authentication also include: Configure the similarity threshold, obtain the facial feature vector of the person being collected, perform final identity authentication on the person being collected, determine the importance score of each vector element in the facial feature vector of each legitimate user in the identity authentication set and the facial feature vector of the person being collected according to the contribution degree of each vector element in the facial feature vector in identity recognition, and perform importance sorting; Configure the importance threshold , after the importance sorting is completed, according to the importance threshold, select the first vector elements in each facial feature vector to construct the key facial feature vector; According to the importance score of each vector element in the key facial feature vector, calculate the similarity between the key facial feature vectors of each legitimate user in the identity authentication set and the key facial feature vector of the person being collected; Screen for the legitimate user in the identity authentication set with the highest similarity to the key facial feature vector of the person being collected, and mark it as the target legitimate user. If the similarity between the target legitimate user and the key facial feature vector of the person being collected is greater than the similarity threshold, the identity authentication is successful; otherwise, the identity authentication fails. Configure the identity authentication threshold, record the number of times the person being collected has undergone identity authentication. When the number of times of identity authentication is greater than the identity authentication threshold, issue an identity authentication warning; otherwise, do nothing.

[0081] Working principle and its effects: When the face multi-feature fusion charging card issuing machine identity authentication system is working, the face acquisition module monitors the acquisition area with the help of an infrared sensor, and starts acquisition when a human body is recognized and the recognition success rate reaches the standard. According to the acquisition distance change rate of the distance sensor and the ambient light intensity of the light sensor, the parameters of the supplementary light device and the camera are adaptively adjusted. The body contour is captured by the camera, and whether to collect an image is determined based on the head position and angle deviation. After the image is collected, the image quality is evaluated. If it does not meet the standard, it is re-collected; if the number of times exceeds the limit, a warning is issued. The feature fusion module performs feature point detection on the collected face image, extracts feature points such as eyes, nose, mouth, and facial contour, extracts geometric features based on these feature points, normalizes the distance features with the diagonal length of the image, and then constructs geometric distance, angle, and proportion feature vectors and splices them into a geometric feature vector. At the same time, after gray processing of the image, multi-scale downsampling is performed, and the texture feature vector of the face image at each downsampling scale is calculated based on the key feature points. Finally, the geometric and texture feature vectors are spliced to generate a facial feature vector, realizing multi-feature fusion and providing comprehensive and accurate feature data for identity authentication. The identity authentication module is based on the legitimate user face database. First, it quickly screens out a preliminary set of legitimate users according to the geometric feature vector, and then obtains the identity authentication set according to the texture feature vector similarity and the gradual screening ratio. Then, determine the importance of the facial feature vector elements and sort them, construct a key facial feature vector, calculate the similarity to screen for the target legitimate user, compare with the threshold to judge the authentication result, record the number of authentications, and issue a warning if the threshold is exceeded. This multi-stage and refined authentication method effectively improves the accuracy and security of authentication. After the authentication is passed, the card issuing control module receives the user identity information, reads the identifier of the card to be issued, activates the card and establishes an encrypted communication channel, writes and verifies the identity information, records the information and updates the cartridge status after confirming the binding, realizing the dynamic binding of the card and the user identity, and ensuring the accuracy and security of the charging card issuing machine.

[0082] Through the close cooperation of each module, from precise face image acquisition, multi-feature fusion extraction to rigorous identity authentication and secure card binding, the accuracy, security and intelligence of the identity authentication of the charging card issuing machine are effectively improved, providing users with a convenient and reliable service experience.

[0083] If the disclosed technical solution involves personal information, the product using the disclosed technical solution has clearly informed the personal information processing rules and obtained the individual's voluntary consent before processing the personal information. If the disclosed technical solution involves sensitive personal information, the product using the disclosed technical solution has obtained the individual's separate consent before processing the sensitive personal information, and at the same time meets the "explicit consent" requirement. For example, on personal information collection devices such as cameras, clear and prominent signs are set to inform that the personal information collection scope has been entered and personal information will be collected. If the individual voluntarily enters the collection scope, it is deemed that he or she agrees to the collection of his or her personal information; or on the device that processes personal information, when the personal information processing rules are notified by obvious signs / information, the individual's authorization is obtained through pop-up information or by asking the individual to upload his or her personal information; among them, the personal information processing rules may include information such as the personal information processor, the purpose of personal information processing, the processing method, and the type of personal information processed.

[0084] The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A face multi-feature fusion charging hairpin machine identity authentication system, characterized in that It includes a face acquisition module, a feature fusion module, an identity authentication module, and a card issuing control module; The face acquisition module is used to obtain the ambient light intensity through the light sensor of the charging card issuing machine when it detects that the person to be collected enters the collection area. After performing adaptive light source fill light based on the ambient light intensity, it automatically adjusts the parameters of the camera of the charging card issuing machine, and automatically detects the position and angle of the person to be collected to acquire a face image; The feature fusion module is used to detect feature points of the acquired face image, extract facial features of the face image based on the feature points, including geometric features and texture features, and perform feature fusion on the facial features to generate a facial feature vector; The identity authentication module is used to authenticate the identity of the person to be collected based on the set of legitimate user face data. It quickly filters through the geometric feature vector of the person to be collected to obtain a preliminary set of legitimate users, and then screens out the identity authentication set according to the step-by-step screening ratio and texture feature vector similarity ranking at each downsampling scale. Then it constructs a key facial feature vector and screens out the target legitimate user to determine whether the identity authentication of the person to be collected is successful; The card issuing control module is used to send a driving instruction to the card magazine of the charging card issuing machine when the identity authentication of the person to be collected is passed, control the card magazine to eject the card, and dynamically bind the card identity according to the user identity information of the person to be collected.

2. The identity authentication system for a charging hairpin machine with multi-feature fusion of human faces as claimed in claim 1, wherein The steps of acquiring the face image include: Continuously monitor the collection area based on the infrared sensor of the charging card issuing machine, obtain the object features in the collection area, and perform human body recognition based on the object features; When a human body is recognized, configure a monitoring time window for human body monitoring. During the monitoring time window, count the number of times the human body is recognized and calculate the human body recognition success rate; Configure a human body recognition threshold. When the human body recognition success rate is greater than the human body recognition threshold, start the face acquisition process, otherwise do not perform any operation; Configure a collection trigger threshold. Obtain the collection distance between the person to be collected and the camera of the charging card issuing machine through the distance sensor, and obtain the change rate of the collection distance. When the change rate of the collection distance is less than the collection trigger threshold, trigger the adjustment of the fill light device parameters and the camera parameters; After adjusting the fill light device parameters and the camera parameters, capture the body contour information of the person to be collected through the camera, and extract the key point coordinates of the human body contour; According to the key point coordinates of the human body contour, calculate the horizontal offset and vertical offset of the head relative to the image center, and calculate the tilt angle of the head of the person to be collected; Configure a position deviation range and an angle deviation range. When both the horizontal offset of the head relative to the image center and the vertical offset of the head relative to the image center are within the position deviation range, and the tilt angle of the head is within the angle deviation range, then perform face image acquisition. Otherwise, give a voice prompt according to the horizontal offset of the head relative to the image center, the vertical offset of the head relative to the image center, and the tilt angle of the head to guide the person to be collected to adjust the position and angle.

3. The identity authentication system for a charging hairpin machine with multi-feature fusion of human faces as described in claim 2, wherein The steps of acquiring the face image further include: Perform a preliminary quality assessment on the acquired face image to obtain the clarity of the face image, the integrity of the face image, and the brightness uniformity of the face image; The clarity of the face image is obtained by calculating the variance of the Laplacian operator of the image; The integrity of the face image is obtained by calculating the integrity and missing ratio of the key points of the face contour; The brightness uniformity of the face image is obtained by calculating the brightness standard deviation of different regions of the face image; Configure the clarity threshold, integrity threshold, and brightness threshold. When the clarity of the face image is less than the clarity threshold, or the integrity of the face image is less than the integrity threshold, or the brightness uniformity of the face image is less than the brightness threshold, re-trigger the image acquisition; otherwise, do nothing; Configure the re-acquisition threshold, monitor the number of times of re-triggering the image acquisition. When the number of times of re-triggering the image acquisition is greater than the re-acquisition threshold, pause the acquisition process and issue an acquisition warning; 4. The identity authentication system for a charging hairpin machine with multi-feature fusion of human faces according to claim 3, characterized in that The steps for adjusting the parameters of the supplementary lighting device include: Configure the light intensity threshold, including the upper limit threshold and the lower limit threshold of the light intensity. Obtain the ambient light intensity through the light sensor of the charging card issuer; When the ambient light intensity is less than the lower limit threshold of the light intensity, increase the supplementary lighting brightness and intensity. Adjust the initial supplementary lighting brightness according to the ratio of the current ambient light intensity to the lower limit threshold of the light intensity, and optimize the supplementary lighting brightness in combination with the position of the face in the image; When the ambient light intensity is greater than or equal to the upper limit threshold of the light intensity, reduce the supplementary lighting brightness. On the basis of the initial supplementary lighting intensity, reduce it proportionally as the light intensity exceeds the upper limit threshold of the light intensity until the minimum supplementary lighting brightness is reached; After adjusting the supplementary lighting brightness, on the basis of the initial supplementary lighting angle, adjust the supplementary lighting angle according to the offset direction of the face center relative to the image center; 5. The identity authentication system for a charging hairpin machine with multi-feature fusion of human faces as claimed in claim 4, wherein, The steps for adjusting the parameters of the camera include: Obtain the ambient light intensity after adjusting the parameters of the supplementary lighting device. If, under the reduced supplementary lighting brightness of the supplementary lighting device, the ambient light intensity obtained by the light sensor of the charging card issuer is still greater than the upper limit threshold of the light intensity, adjust the camera exposure parameters, and reduce the camera sensitivity proportionally according to the degree to which the ambient light intensity exceeds the upper limit threshold; Adjust the camera focal length according to the distance between the person being captured and the camera; Based on the ambient light intensity, starting from the initial aperture, adjust according to the difference between the current ambient light intensity and the reference light intensity. When the ambient light intensity is higher than the reference light intensity, increase the camera aperture; conversely, decrease the camera aperture; Comprehensively adjust the camera shutter speed according to the ambient light intensity and the acquisition distance. On the basis of the initial shutter speed, when the ambient light intensity is greater than the reference light intensity, the shutter speed decreases proportionally, and the acquisition distance is inversely proportional to the shutter speed; 6. The identity authentication system for a charging hairpin machine with multi-feature fusion of human faces as described in claim 1, wherein The steps for generating the facial feature vector include: Obtain the face image, and perform feature point detection on the face image. The feature points include: eye feature points, nose feature points, mouth feature points, and facial contour feature points; Extract the geometric features of the face image based on the feature points of the face image, including: distance features, angle features, and ratio features; Select the diagonal length of the face image as the scale factor, and perform normalization processing on the distance features in the geometric features based on the scale factor to eliminate the influence of the face image size on the distance features; Sort the normalized distance features, angle features, and scale features respectively in the order from top to bottom and from left to right of the facial image to construct a geometric distance feature vector, a geometric angle feature vector, and a geometric scale feature vector; Concatenate the geometric distance feature vector, the geometric angle feature vector, and the geometric scale feature vector in sequence to construct a geometric feature vector; Perform grayscale processing on the face image, and perform multi-scale downsampling on the grayscale processed face image to obtain the face image at each downsampling scale for texture feature extraction to obtain a texture feature vector; Concatenate the geometric feature vector and the texture feature vector to obtain a facial feature vector.

7. The identity authentication system for a charging hairpin machine with multi-feature fusion of human faces according to claim 6, characterized in that The steps for obtaining the texture feature vector include: Select key feature points based on feature points and perform key feature point detection on the face image at each downsampling scale; Set the neighborhood window size according to the downsampling scale, and calculate the gradient direction and gradient magnitude of each pixel point relative to the key feature point within the neighborhood window with the detected key feature point as the center; Perform regional division on the neighborhood window to obtain neighborhood sub-regions, and statistically calculate the gradient direction histogram within each neighborhood sub-region, and connect the gradient direction histograms of all neighborhood sub-regions to form the texture feature of the key feature point; For each downsampling scale, construct a corresponding sampling scale texture feature vector according to the type of key feature point, and concatenate the texture feature vectors of each sampling scale in the order from smallest to largest downsampling scale to obtain a texture feature vector.

8. The identity authentication system for a charging hairpin machine with multi-feature fusion of human faces as claimed in claim 1, wherein, The specific steps of the identity authentication include: Let the set of legitimate user face data be , where is the face data of the th legitimate user, including the geometric feature vector, texture feature vectors at each downsampling scale, and facial feature vector of the th legitimate user; Configure a quick screening ratio, obtain the geometric feature vector of the person being collected, quickly screen the legal user face data set, and calculate the similarity between the person being collected and the geometric feature vectors of each legal user in the face database set; Sort the similarities between the geometric feature vectors of the collector and the legal users in the face database set, and obtain a preliminary legal user set according to the quick screening ratio; Configure a step-by-step screening ratio for each downsampling scale, obtain the texture feature vector of the person being collected at each downsampling scale, perform step-by-step screening on the preliminary legal user set, and calculate the similarity between the person being collected and the texture feature vectors of each legal user at the downsampling scale in the quickly screened preliminary legal user set in sequence; Sort the similarities of the texture feature vectors at each downsampling scale in sequence, and obtain a step-by-step screened legal user set according to the step-by-step screening ratio of each downsampling scale to obtain an identity authentication set.

9. The identity authentication system for a charging hairpin machine with multi-feature fusion of human faces as claimed in claim 8, wherein The specific steps of the identity authentication further include: Configure a similarity threshold, obtain the facial feature vector of the person being collected, perform final identity authentication on the person being collected, determine the importance score of each vector element in the facial feature vector according to its contribution degree in identity recognition for the facial feature vectors of each legal user in the identity authentication set and the facial feature vector of the person being collected, and perform importance ranking; Configure the importance threshold , after the importance ranking is completed, according to the importance threshold, select the first vector elements in each facial feature vector to construct a key facial feature vector; Calculate the similarity between the key facial feature vectors of each legal user in the identity authentication set and the key facial feature vector of the person being collected according to the importance scores of each vector element in the key facial feature vector; Screen the legal user in the identity authentication set with the highest similarity to the key facial feature vector of the person being collected, and mark it as the target legal user. If the similarity between the target legal user and the key facial feature vector of the person being collected is greater than the similarity threshold, the identity authentication is successful; otherwise, the identity authentication fails. Configure the identity authentication threshold, record the number of times the person being collected has undergone identity authentication. When the number of times of identity authentication is greater than the identity authentication threshold, issue an identity authentication warning; otherwise, do nothing.

10. A method for identity authentication of a charging card issuing machine with multi-feature fusion of human faces, which is implemented based on the identity authentication system of the charging card issuing machine with multi-feature fusion of human faces described in any one of claims 1-9, characterized in that, It includes the following steps: When it is detected that the person being collected enters the collection area, obtain the ambient light intensity through the light sensor of the charging card dispenser. After performing adaptive light source fill light based on the ambient light intensity, automatically adjust the camera parameters of the charging card dispenser, and automatically detect the position and angle of the person being collected to collect a face image. Perform feature point detection on the collected face image, extract the facial features of the face image based on the feature points, including geometric features and texture features, and perform feature fusion on the facial features to generate a facial feature vector. Perform identity authentication on the person being collected based on the legal user face data set. Use the geometric feature vector of the person being collected to quickly screen out a preliminary set of legal users, then screen out the identity authentication set according to the step-by-step screening ratio and texture feature vector similarity ranking at each downsampling scale, and then construct a key facial feature vector to screen out the target legal user to determine whether the identity authentication of the person being collected is successful. When the identity authentication of the person being collected is passed, send a drive command to the card magazine of the charging card dispenser to control the card magazine to eject the card, and dynamically bind the card identity according to the user identity information of the person being collected.

Images