WO2020077822A1

WO2020077822A1 - Image feature configuration and verification method and apparatus, computer device and medium

Info

Publication number: WO2020077822A1
Application number: PCT/CN2018/122731
Authority: WO
Inventors: 胡金丹
Original assignee: 深圳壹账通智能科技有限公司
Priority date: 2018-10-17
Filing date: 2018-12-21
Publication date: 2020-04-23
Also published as: CN109472269A

Abstract

Disclosed are an image feature configuration and verification method and apparatus, a computer device and a medium. The method comprises: acquiring N first images, wherein N is a positive integer greater than or equal to 2; acquiring N human body key points of the first image according to a preset training model; acquiring a first portrait feature according to the N human body key points of the first image; configuring the first portrait feature as a standard image feature; acquiring a second image, and acquiring a human body key point of the second image according to the preset training model; acquiring a second portrait feature according to the human body key point of the second image; matching the second portrait feature with the standard image feature, if the match is successful, outputting a verification pass result. The technical solution provided in the present application can make the accuracy of image feature configuration high , and it is difficult for non-users to impersonate and log in, thereby ensuring the security of user information.

Description

Image feature configuration and verification method, device, computer equipment and medium

This application is based on the Chinese invention patent application with the application number 201811208048.6 filed on October 17, 2018, titled "Image Feature Configuration and Verification Methods, Devices, Computer Equipment, and Media", and claims its priority.

Technical field

The present application belongs to the field of image recognition, and more specifically, to an image feature configuration and verification method, device, computer equipment, and storage medium.

Background technique

At present, the use of mobile phones is becoming more and more common, and the information security of users on mobile phones is also getting more and more attention. In order to protect the user's information security, the mobile phone login method is usually encrypted, and there are many common encryption methods, such as slider, password, voice, portrait or fingerprint. Since the unlocking methods of these encryption methods are single, there are certain cracking methods, and the difficulty of cracking is not very large. Therefore, the user's information security still cannot be well guaranteed.

Summary of the invention

Embodiments of the present application provide an image feature configuration and verification method, device, device, and storage medium, to solve the problem that user login methods are easily cracked.

An image feature configuration and verification method, including:

Acquire N first images, where N is a positive integer greater than or equal to 2;

Acquiring N human key points of the first image according to a preset training model;

Acquiring the first portrait feature according to the N human key points of the first image;

Configuring the first portrait feature as a standard image feature;

Acquiring a second image, acquiring human key points of the second image according to the preset training model;

Acquiring a second portrait feature according to the key points of the human body of the second image;

The second portrait feature is matched with the standard image feature, and if the match is successful, a result of verification is output.

An image feature configuration and verification device, including:

The first image acquisition module is used to acquire N first images, where N is a positive integer greater than or equal to 2;

A first human key point obtaining module, configured to obtain N human key points of the first image according to a preset training model;

A first portrait feature acquisition module, configured to acquire the first portrait feature according to N human key points of the first image;

A standard image feature configuration module, configured to configure the first portrait feature as a standard image feature;

A second human body key point obtaining module, configured to obtain a second image, and obtain human body key points of the second image according to the preset training model;

A second portrait feature acquisition module, configured to acquire a second portrait feature according to the key points of the human body of the second image;

The portrait feature matching verification module is configured to match the second portrait feature with the standard image feature, and if the match is successful, output a result of verification.

A computer device includes a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor. When the processor executes the computer-readable instructions, the following steps are implemented:

Configuring the first portrait feature as a standard image feature;

One or more non-volatile readable storage media storing computer-readable instructions, which when executed by one or more processors, cause the one or more processors to perform the following steps:

Configuring the first portrait feature as a standard image feature;

The details of one or more embodiments of the present application are set forth in the following drawings and description, and other features and advantages of the present application will become apparent from the description, drawings, and claims.

BRIEF DESCRIPTION

In order to more clearly explain the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required in the embodiments or the description of the prior art. Obviously, the drawings in the following description are only for the application In some embodiments, for those of ordinary skill in the art, without paying creative labor, other drawings may be obtained based on these drawings.

1 is a schematic diagram of an application environment of an image feature configuration and verification method in an embodiment of the present application;

2 is a flowchart of an image feature configuration and verification method in an embodiment of the present application;

FIG. 3 is another flowchart of the image feature configuration and verification method in an embodiment of the present application;

4 is another flowchart of an image feature configuration and verification method in an embodiment of the present application;

5 is another flowchart of the image feature configuration and verification method in an embodiment of the present application;

6 is another flowchart of an image feature configuration and verification method in an embodiment of the present application;

7 is a schematic block diagram of an image feature configuration and verification device in an embodiment of the present application;

8 is a schematic block diagram of a first portrait feature acquisition module in an image feature configuration and verification device according to an embodiment of the present application;

9 is another principle block diagram of a first portrait feature acquisition module in an image feature configuration and verification device according to an embodiment of the present application;

10 is a schematic diagram of a computer device in an embodiment of the present application.

detailed description

The technical solutions in the embodiments of the present application will be described clearly and completely in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, but not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the scope of protection of this application.

The image feature configuration and verification method provided in this application can be applied in the application environment as shown in FIG. 1, in which the client communicates with the server through the network, and the server obtains N first images through the client according to the preset Train the model to obtain the human key points of the N first images; obtain the first portrait features based on the N human key points of the first images; configure the first portrait features as standard image features; and then obtain the second images from the server, Acquire the human key points of the second image according to the preset training model; acquire the second human portrait features according to the human key points of the second image; finally match the second human portrait features with the standard image features, and if the matching is successful, output to the client Verify the results. Among them, the client can be, but not limited to, various personal computers, notebook computers, smart phones, tablet computers and portable wearable devices. The server can be implemented with an independent server or a server cluster composed of multiple servers.

In an embodiment, as shown in FIG. 2, an image feature configuration and verification method is provided. The method is applied to the server in FIG. 1 as an example for illustration, including the following steps:

S10: Acquire N first images, where N is a positive integer greater than or equal to 2.

The first image is a user portrait collected when the user sets image features. Optionally, the user's portrait can be collected through the shooting tool of the client, for example, the user's portrait can be collected through the shooting function of the camera of the mobile phone. Optionally, in order to ensure that the user corresponding to the first image is the user himself, before acquiring the first image, the server sends user login verification to the client to enable the user to enter a password or fingerprint to perform login verification; if the login verification is The result is pass, and then send the instruction to collect the first image to the client to collect the first image.

To enhance the representativeness of image features, optionally, the number of first images is N, and N is a positive integer greater than or equal to 2. The first image may be multiple static images, or multiple images acquired by recording video data. In one embodiment, as shown in FIG. 3, step S10 may specifically include:

S11: Acquire the first video data.

Wherein, the first video data is a video recorded to the user himself, for example, a video recording the user blinking.

Specifically, the server sends an instruction to collect the first image to the client. The client opens the shooting tool according to the instruction to collect the first image, records the user's video, and obtains the first video data.

S12: Frame the first video data according to a preset time to obtain N images to be processed.

Among them, the preset time can be specifically set according to the actual situation. Optionally, the total frame number and overall time of the first video data may be obtained, and then the preset time may be obtained by dividing the total frame number by the overall time. The server then frames the first video data according to the obtained preset time to obtain N images to be processed.

S13: Create a normalized image, obtain the height and width information of the normalized image, and obtain the normalized images of N images to be processed based on the height and width information, and replace the normalized images with N pending images The original pixel values of the image are processed to obtain N first images.

Specifically, the server first creates a normalized image, for example, an image of 260 * 260 pixels; then obtains the height and width information of the normalized image; then calculates the height and width information of the image to be processed according to the normalized image After performing the normalized image and replacing the pixel value of the original image to be processed with the normalized image, N first images can be obtained.

In the embodiment corresponding to FIG. 3, by acquiring the first video data; framing the first video data according to a preset time, acquiring N images to be processed; creating a normalized image, acquiring the normalized Information about the height and width of the image, and obtaining the normalized images of the N images to be processed based on the height and width information, and replacing the original images with the normalized images Pixel values to obtain the N first images. The user can input the customized facial movements and behavioral movements as standard image features as required, which makes the image feature configuration more accurate; and does not need to perform multiple shots to collect, which improves the efficiency of acquiring the first image.

S20: Acquire N human key points of the first image according to the preset training model.

The preset training model may be a face detection model, a feature point detection model, a posture detection model, an emotion detection model, and so on. The key points of the human body refer to the points in the first image that reflect human body characteristics, such as eyebrows, eyes, mouth, shoulders, elbow joints, and wrists. Optionally, the preset training model can be trained by inputting sample images marked with key points, and learning to obtain key points of the human body. When the first image is input to the preset training model, the preset training model can identify the key points of the human body in the first image, thereby acquiring the key points of the human body of the first image.

S30: Acquire the first portrait feature according to the key points of the human body of the N first images.

Among them, the first portrait feature refers to a portrait feature composed of the features of the key points of the human body of the N first images, and is used to determine whether it is the user's own credential. For example, the first portrait feature may be a facial feature, an expression feature, a behavior feature, or the like. Optionally, the facial expression feature and the behavioral action feature can be combined as the first portrait feature, that is, the user's facial expression and body movement can be used as the login credential to improve the security of user information. Behavioral action features refer to custom behaviors entered when the user sets up login credentials, such as making a hand-raising action or blinking the left eye to the camera, or a combination of blinking the left eye and raising the right hand. That is, the behavior feature may be a single behavior action, or a combination of a plurality of behavior actions.

Specifically, the first portrait feature can be obtained by extracting, calculating, or recognizing features of key points of the human body by a preset training model. Optionally, a preset training model can be used to extract and recognize the facial expressions formed by key points of the human body, for example, through the features such as the angle of the eyebrow tilt, the downward movement of the mouth and the angle of the eye and the face, the corresponding facial expressions can be identified to obtain Corresponding expression characteristics. Optionally, the position of the key point of the human body can be tracked, and the change of the position of the key point of the human body can be used as the behavior feature. For example, by acquiring video data of a user-defined hand-lifting action, a coordinate system is established for multiple first images obtained from the video data, and the key point of the human body such as the wrist is moved from position A to position B in the multiple first images According to the coordinate change information of the position, according to the coordinate change information, the change information of the position of the wrist, which is a key point of the human body, can be obtained, so as to obtain the user-defined behavior characteristics of the hand-raising action.

S40: Configure the first portrait feature as a standard image feature.

Specifically, the first portrait feature and the user ID are bound in the server database and stored as standard image features, the configuration of the image features is completed, and the standard image features are used as the user login credentials. Among them, the user ID is an identifier used by the server to distinguish different users, and may be the user's mobile phone number, account number, or ID number. Optionally, standard image features and other forms of passwords can be combined as credentials for user login, for example, combined with digital passwords, which can further enhance the security of user information.

S50: Acquire the second image, and acquire key points of the human body of the second image according to the preset training model.

The second image refers to a portrait image obtained when the user performs login verification. Optionally, the number of second images is at least one.

Specifically, when the user logs in, the second image is acquired through the shooting tool of the client. After acquiring the second image, the second image is input into the preset training model, and the key points of the human body of the second image are acquired according to the preset training model. The process of acquiring the key points of the human body of the second image is the same as the process of acquiring the key points of the human body of the first image, and will not be repeated here.

S60: Acquire the second portrait feature according to the key points of the human body in the second image.

Wherein, the process of acquiring the second portrait feature according to the key points of the human body of the second image is the same as the process of acquiring the first portrait feature, which will not be repeated here. It can be understood that the second portrait feature is of the same type as the first portrait feature, for example, all are facial features, facial expression features, or behavioral features.

S70: Match the second portrait feature with the standard image feature, and if the match is successful, output a verified result.

Specifically, the server matches the second portrait feature with the standard image feature, and determines whether the acquired second portrait feature matches the standard image feature. Optionally, when the standard image feature corresponds to a facial feature, each facial feature in the second portrait feature is compared with each facial feature in the standard image feature to determine whether the facial features are the same, such as whether the eyebrows are raised Or whether the corner of the mouth is moving down, etc. If the facial features are the same, it is determined that the matching is successful, otherwise it is determined that the matching fails. When the standard image feature corresponds to the expression feature, the expression corresponding to the second portrait feature is compared with the expression corresponding to the standard image feature to determine whether the result of the expression is the same, for example, whether the result of the expression is happy, sad, or surprised. If the result of the expression is the same, it is determined that the match is successful, otherwise it is determined that the match fails. When the standard image feature corresponds to the behavior feature, the result of the behavior action in the second portrait feature is compared with the result of the behavior action in the standard image feature to determine whether the result of the behavior action is consistent, for example, the behavior action in the standard image feature The result is that the left hand raises the hand, it is judged whether the result of the behavior action of the second portrait feature is also the left hand raise the hand, if the result of the behavior action is the same, it is determined that the match is successful, otherwise it is determined that the match fails.

Specifically, if the server judges that the second portrait feature matches the standard image feature, it outputs the result of the verification and allows the user to log in. If the server judges that the second portrait feature does not match the standard image feature, it outputs a result that the verification fails and rejects the user login. It can be understood that when other users want to log in as fake users, they do not know whether the standard image features correspond to facial expression features, behavioral action features, or a combination of facial expression features and behavioral action features, and do not know the specific facial expression features and behavioral actions Features, so it is difficult to crack.

In the embodiment corresponding to FIG. 2, by acquiring N first images, and then acquiring human key points of N first images according to a preset training model; acquiring first portrait features based on N human key points of first images , The first portrait feature is configured as a standard image feature; then the second image is obtained, the key points of the human body of the second image are obtained according to the preset training model; the second portrait feature is obtained according to the key points of the human body of the second image, and finally the The two portrait features are matched with the standard image features. If the matching is successful, the result of the verification is output. On the one hand, when the standard image features are used as the login credentials, the user does not need to enter a password to log in, which facilitates the user's operation. On the other hand, obtaining portrait features based on key points of the human body and configuring the acquired portrait features as standard image features can make the image feature configuration more representative and improve the accuracy of the image feature configuration. Further, using the standard image features as the user's login credential can enable the user to input a customized facial action or behavioral action as the credential for login verification, and it is difficult for non-users to obtain the login credential, which makes it impossible to crack and impersonate the user Log in to improve the security of user information.

In an embodiment, the first portrait feature can be obtained by establishing a coordinate system for the first image to obtain the coordinates of the key points of the human body, as shown in FIG. Click to obtain the first portrait feature, which may include:

S31: Acquire the coordinates of the human key points of the N first images according to the positions of the human key points of the N first images.

Specifically, a coordinate system may be established in the photo frame of the shooting tool that collects the first image. Optionally, a coordinate system is established using the position of the user's eyebrow center in the photo frame as the origin, and then the coordinates of the key points of the human body of the first image are acquired. When the user enters the photo frame, the preset training model is used to draw the user's portrait; when the user's portrait is drawn, the coordinates of the drawn point are obtained. For example, when drawing points on the eyebrows, the coordinates of the eyebrow drawing points can be obtained through the coordinate system. Optionally, when the user enters the photo frame, the number of human body key points of the first image is counted first, and the coordinates of the human body key points of the first image are acquired after all necessary human body key points have entered the photo frame. Among them, the necessary human key points can be obtained according to the training data. For example, after training, the human key points of the face and hands should all enter the photo frame to obtain the corresponding expression characteristics and hand behavior characteristics, you can set After all the key points of the human face and hands enter the photo frame, the coordinates of the key points of the first image are obtained.

S32: Calculate the coordinates of the key points of the human body of the N first images using a moving index weighted average algorithm to obtain the first feature interval value.

It can be understood that since the first image is N, when the preset training model is used to draw the key points of the human body, the position of the user each time the frame is entered may be different, or the user's position may move during the recording of the video Therefore, the coordinates of the key points of the human body of the N first images obtained each time may be different. Therefore, in order to make the coordinates more representative, it is necessary to further calculate the acquired coordinates, so as to obtain coordinate values that can be used as the first portrait feature. In addition, because the key points of the human body are plotted, a key point of the human body obtains a set of coordinates. For example, when drawing the eyebrows, because the eyebrow drawing points include multiple, then the coordinates of the key point of the human eyebrow is a set of coordinates when acquiring the coordinates, so for a key point of the human body, the range value of the set of coordinates obtained That is, a characteristic interval value.

Specifically, the coordinates of the key points of the human body of the N first images obtained are calculated by using an exponential weighted average algorithm (Exponential Weighted Moving Average, EWMA for short), and the calculated results are formed into feature interval values of the key points of the human body. Among them, the formula for calculating the X coordinate with EWMA can be:

In the formula, X is the weighted average coordinate value; n is the number of the first image (ie N), x _i is the actual value of the ith coordinate; β is the weight of the ith (the sum of the weights is equal to 1); namely The EWMA calculation of the X coordinate value and the EWMA calculation of the Y coordinate are obtained for the same drawing point of the N first images. Then, the obtained EWMA values of the human body key points are combined together to form the first characteristic interval value of the human body key points. Among them, the weight setting can be set to be the same, for example, when the first image is 3, 1/3 is taken as the weight value. You can also set different weights according to the course of the action. For example, in the action of raising your hand, set a larger weight for the start position and end position, and set a smaller weight for the position in the process, that is, focus on the start Whether the coordinates of the position and the end position are in place, regardless of the specific movement route of raising the hand.

S33: Use the first feature interval value as the first portrait feature.

Specifically, the first feature interval values of all the key points of the human body are used as the first portrait feature, which is bound to the user ID and stored in the database on the server side. Among them, the user ID may be a mobile phone number, ID card number and account number used to distinguish different users.

In the embodiment corresponding to FIG. 4, the coordinates of the human key points of the N first images are obtained from the positions of the human key points of the N first images; then, the mobile index weighted average algorithm is used for the human keys of the N first images The coordinates of the points are calculated to obtain the first feature interval value; finally, the first feature interval value is used as the first portrait feature. The user's portrait feature is obtained through the mobile index weighted average algorithm, and the feature data of the user's portrait can be smoothly returned as the first portrait feature according to the first image, thereby improving the accuracy of image feature configuration. Further, the standard image features formed by the first portrait feature obtained according to this embodiment are used as the credentials for login, which can effectively avoid the situation of logging in without user authorization or logging in as a fake user, thereby improving the security of user information Sex.

In an embodiment, the first portrait feature may be obtained after feature extraction or recognition by a preset training model, where the preset training model includes a micro-expression recognition model and a gesture recognition model, specifically, as shown in FIG. 5, In step S30, that is, obtaining the first portrait feature according to the N human body key points of the first image, the method may further include:

S31 ': Acquire the first face image set and the first limb image set according to the N human body key points of the first image.

Specifically, N first images are divided into a first face image set and a first limb image set according to key points of the human body. Optionally, the sample image marked with area division may be input into a preset training model for training so that it can obtain the first face image and the first limb image according to key points of the human body. For example, the first image can be divided into a first face image and a first limb image by using the neck, which is a key point of the human body, as a dividing boundary. And the first face image set is composed of the first face image, and the first limb image set is composed of the first limb image.

S32 ': input the first face image set into the micro-expression recognition model to obtain standard face image features.

Specifically, the first face image set is input into the micro-expression recognition model, and the human face key point characteristics of the first face image are analyzed and recognized according to the micro-expression recognition model, and the expression features of the first face image set are output As a standard face image feature. Optionally, it may also be determined which expression the first face image in the first face image set belongs to. The facial expression features may include head features, eye features, and lip features. For example, facial expressions such as head up, eyebrow up and mouth corner down. It can be understood that, since the first face image set includes multiple face images, the corresponding expression of the user may be changing. Therefore, the standard facial image features can be acquired after the acquired expression features are stable. Among them, the stable expression feature can be set in a continuous preset number of face images to obtain the same expression feature as a stable sign.

Optionally, the expression features of the first face image can be used as the standard face image features, or they can be combined with the results of the expression to form the standard face image features. Among them, the result of expression refers to expressions of happiness, anger or sadness. Optionally, when the first face image is acquired, the international micro expression database can be connected through the server to identify the facial expression expression from the micro expression database. Among them, the international micro-expression database includes 54 kinds of micro-expressions, and specific expressions can be obtained according to subtle changes in key points of the human body.

S33 ': Input the first limb image set into the gesture recognition model to obtain standard limb image features.

Specifically, the first limb image set is input into a gesture recognition model, and the behavior recognition of the key points of the human body of the first limb image is performed according to the gesture recognition model, and the behavior behavior features are output as standard limb image features, for example, output behavior actions Raise the hand for the left hand, then use the left hand as the standard limb image feature. Optionally, a sample set of a series of actions may be input in advance to allow the gesture recognition model to learn, so that the gesture recognition model recognizes the user's behavior. For example, a set of motion raising hand sample sets is input into the gesture recognition model, so that the gesture recognition model can recognize the motion of raising the hand.

S34 ': The standard face image feature and the standard limb image feature are combined into the first portrait feature.

Specifically, the standard face image feature and the standard limb image feature obtained in step S32 'and step S33' are combined into the first portrait feature. For example, a happy facial expression feature and a left-hand raised behavior feature are combined to form a first portrait feature.

In the embodiment corresponding to FIG. 5, the first face image set and the first limb image set are obtained by using the human key points of the N first images; then the first face image set is input into the micro-expression recognition model, respectively To get the standard face image features; input the first limb image set into the gesture recognition model to get the standard limb image features; finally, the standard face image features and the standard limb image features form the first portrait features. By obtaining the facial image feature and the limb image feature respectively according to the micro-expression recognition model and the gesture recognition model, and finally combining the two features to form the first portrait feature, the accuracy of the image feature configuration can be improved. Further, the standard image features formed by the first portrait feature obtained according to this embodiment are used as credentials for user login, which can effectively avoid the situation of logging in without user authorization or impersonating the user, thereby improving the user information. safety.

In an embodiment, in step S60, that is, obtaining the second portrait feature according to the key points of the human body of the second image, as shown in FIG. 6, it may specifically include:

S61: Acquire a second face image and a second limb image according to the key points of the human body of the second image.

Among them, the process of acquiring the second face image and the second limb image according to the key points of the human body of the second image is similar to the process of acquiring the first face image set and the first limb image set according to the key points of the first image, that is, The second image is input into the trained preset training model, and the second face image and the second limb image are obtained according to the divided boundary.

S62: Input the second face image into the micro-expression recognition model to obtain the characteristics of the test face image.

Specifically, the second face image is input into the micro-expression recognition model, and the characteristics of the key points of the human body of the second face image are analyzed and recognized according to the micro-expression recognition model, and the expression features of the second face image are output as the test person Face image features, such as facial expression features such as head up, eyebrow up, and mouth corner down. Among them, the test facial image features are consistent with the standard facial image feature settings. For example, if the facial expression features of the first facial image and the results of the expression are set to form the standard facial image features, then the test facial image features are also The facial expression features of the two-face image are composed of the facial expression results.

S63: Input the second limb image into the gesture recognition model to obtain the test limb image features.

Specifically, the second limb image is input into the gesture recognition model, and the behavior recognition of the key points of the human body of the second limb image is performed according to the gesture recognition model, and the behavior behavior characteristic is output as the test limb image feature.

S64: Combine the test face image feature and the test limb image feature into a second portrait feature.

Specifically, the test face image features and the test limb image features obtained in step S62 and step S63 constitute a second portrait feature.

In the embodiment corresponding to FIG. 6, the second face image and the second limb image are obtained according to the key points of the human body of the second image; then the second face image is input into the micro-expression recognition model to obtain the test face image Feature; then input the second limb image into the posture recognition model to obtain the test limb image feature; finally, the test face image feature and the test limb image feature form a second portrait feature, and the second image's portrait feature can be extracted to be used with Standard image features are compared to enable verification of image features.

It should be understood that the size of the sequence numbers of the steps in the above embodiments does not mean the order of execution, and the execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

In one embodiment, an image feature configuration and verification device is provided, and the image feature configuration and verification device corresponds to the image feature configuration and verification method in the above embodiment in one-to-one correspondence. As shown in FIG. 7, the image feature configuration and verification device includes a first image acquisition module 10, a first human body key point acquisition module 20, a first human portrait feature acquisition module 30, a standard image feature configuration module 40, and a second human body The key point acquisition module 50, the second portrait feature acquisition module 60, and the portrait feature matching verification module 70. The detailed description of each functional module is as follows:

The first image acquisition module 10 is configured to acquire N first images, where N is a positive integer greater than or equal to 2.

Further, the first image acquisition module 10 is also used to:

Get the first video data;

Frame the first video data according to a preset time to obtain N images to be processed;

Create a normalized image, obtain the height and width information of the normalized image, and obtain the normalized image of N images to be processed based on the height and width information, and replace the normalized image with N images to be processed The original pixel value of N to get N first images.

The first human key point obtaining module 20 is configured to obtain N human key points of the first image according to a preset training model.

The first portrait feature acquisition module 30 is configured to acquire the first portrait feature according to the N human body key points of the first image.

The standard image feature configuration module 40 is configured to configure the first portrait feature as a standard image feature.

The second human key point obtaining module 50 is used to obtain a second image and obtain the human key points of the second image according to a preset training model.

The second portrait feature acquisition module 60 is configured to acquire the second portrait feature according to the key points of the human body in the second image.

The portrait feature matching verification module 70 is configured to match the second portrait feature with the standard image feature. If the matching is successful, the verification result is output.

Further, as shown in FIG. 8, the first portrait feature acquisition module 30 includes a coordinate acquisition unit 31, a feature interval value acquisition unit 32, and a first portrait feature setting unit 33.

The coordinate acquiring unit 31 is configured to acquire the coordinates of the human key points of the N first images according to the positions of the human key points of the N first images.

The feature interval value obtaining unit 32 is configured to calculate the coordinates of the key points of the human body of the N first images using a moving index weighted average algorithm to obtain the first feature interval value.

The first portrait feature setting unit 33 is configured to use the first feature interval value as the first portrait feature.

Further, the preset training model includes a micro-expression recognition model and a gesture recognition model; optionally, as shown in FIG. 9, the first portrait feature acquisition module 30 includes an image set acquisition unit 31 ′ and a standard facial feature acquisition unit 32 ', A standard limb feature acquisition unit 33' and a first portrait feature acquisition unit 34 '.

The image set acquisition unit 31 'is used to acquire the first face image set and the first limb image set based on the N human body key points of the first image.

The standard facial feature acquisition unit 32 'is used to input the first facial image set into the micro-expression recognition model to obtain the standard facial image features.

The standard limb feature acquisition unit 33 'is used to input the first limb image set into the posture recognition model to obtain the standard limb image features.

The first portrait feature acquisition unit 34 'is used to form the standard face image feature and the standard limb image feature into the first portrait feature.

Further, the second portrait feature acquisition module 60 is also used to:

Acquire the second face image and the second limb image according to the key points of the human body of the second image;

Input the second face image into the micro-expression recognition model to obtain the test face image features;

Input the second limb image into the gesture recognition model to obtain the test limb image features;

The test face image feature and the test limb image feature are combined to form a second portrait feature.

For the specific limitation of the image feature configuration and verification device, reference may be made to the above limitation on the image feature configuration and verification method, which will not be repeated here. Each module in the above image feature configuration and verification device may be implemented in whole or in part by software, hardware, and a combination thereof. The above modules may be embedded in the hardware or independent of the processor in the computer device, or may be stored in the memory in the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

In one embodiment, a computer device is provided. The computer device may be a server, and an internal structure diagram thereof may be as shown in FIG. The computer device includes a processor, memory, network interface, and database connected by a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, computer-readable instructions, and a database. The internal memory provides an environment for the operation of the operating system and computer-readable instructions in the non-volatile storage medium. The database of the computer device is used to store the first image, the first video data, the preset training model, the standard image features, the moving index weighted average algorithm, and the feature interval values. The network interface of the computer device is used to communicate with external terminals through a network connection. When the computer-readable instructions are executed by the processor to implement an image feature configuration and verification method.

In one embodiment, a computer device is provided, including a memory, a processor, and computer-readable instructions stored on the memory and executable on the processor. The processor implements the computer-readable instructions to implement the following steps:

Acquire N key points of the human body in the first image according to the preset training model;

Obtain the first portrait features according to the key points of the human body of the N first images;

Configure the first portrait feature as a standard image feature;

Obtain the second image, and obtain the key points of the human body of the second image according to the preset training model;

Obtain the second portrait feature according to the key points of the human body in the second image;

The second portrait feature is matched with the standard image feature, and if the match is successful, the result of verification is output.

In one embodiment, one or more non-volatile readable storage media storing computer-readable instructions are provided. When the computer-readable instructions are executed by one or more processors, the one or more Each processor performs the following steps:

Configure the first portrait feature as a standard image feature;

A person of ordinary skill in the art may understand that all or part of the process in the method of the foregoing embodiments may be completed by instructing relevant hardware through computer-readable instructions, and the computer-readable instructions may be stored in a non-volatile computer In the readable storage medium, when the computer-readable instructions are executed, they may include the processes of the foregoing method embodiments. Wherein, any reference to the memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous chain (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

Those skilled in the art can clearly understand that, for convenience and conciseness of description, only the above-mentioned division of each functional unit and module is used as an example for illustration. In practical applications, the above-mentioned functions can be allocated by different functional units, Module completion means that the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above.

The above-mentioned embodiments are only used to illustrate the technical solutions of the present application, not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still implement the foregoing The technical solutions described in the examples are modified, or some of the technical features are equivalently replaced; and these modifications or replacements do not deviate from the spirit and scope of the technical solutions of the embodiments of the present application. Within the scope of protection of this application.

Claims

An image feature configuration and verification method, which includes:

Acquire N first images, where N is a positive integer greater than or equal to 2;

Acquiring N human key points of the first image according to a preset training model;

Acquiring the first portrait feature according to the N human key points of the first image;

Configuring the first portrait feature as a standard image feature;

Acquiring a second image, acquiring human key points of the second image according to the preset training model;

Acquiring a second portrait feature according to the key points of the human body of the second image;

The second portrait feature is matched with the standard image feature, and if the match is successful, a result of verification is output.
The image feature configuration and verification method according to claim 1, wherein the acquiring the first portrait feature according to the N human body key points of the first image includes:

Acquiring the coordinates of the N human key points of the first image according to the positions of the N human key points of the first image;

Use the moving index weighted average algorithm to calculate the coordinates of the key points of the human body of the N first images to obtain the first feature interval value;

Use the first feature interval value as the first portrait feature.
The image feature configuration and verification method according to claim 1, wherein the preset training model includes a micro-expression recognition model and a gesture recognition model;

The obtaining the first portrait feature according to the N human key points of the first image includes:

Acquiring a first face image set and a first limb image set according to the N human body key points of the first image;

Input the first face image set into the micro-expression recognition model to obtain standard face image features;

Input the first limb image set into the gesture recognition model to obtain standard limb image features;

The standard face image feature and the standard limb image feature are combined into the first portrait feature.
The image feature configuration and verification method according to claim 3, wherein the acquiring the second portrait feature according to the key points of the human body of the second image includes:

Acquiring a second face image and a second limb image according to the key points of the human body of the second image;

Input the second face image into the micro-expression recognition model to obtain the characteristics of the test face image;

Input the second limb image into the posture recognition model to obtain test limb image characteristics;

The test face image feature and the test limb image feature are combined to form the second portrait feature.
The image feature configuration and verification method according to claim 1, wherein the acquiring N first images includes:

Get the first video data;

Frame the first video data according to a preset time to obtain N images to be processed;

Create a normalized image, obtain height and width information of the normalized image, and obtain a normalized image of the N images to be processed based on the height and width information, and normalize the normalized image Replace the original pixel values of the N images to be processed to obtain the N first images.
An image feature configuration and verification device, characterized in that it includes:

The first image acquisition module is used to acquire N first images, where N is a positive integer greater than or equal to 2;

A first human key point obtaining module, configured to obtain N human key points of the first image according to a preset training model;

A first portrait feature acquisition module, configured to acquire the first portrait feature according to N human key points of the first image;

A standard image feature configuration module, configured to configure the first portrait feature as a standard image feature;

A second human body key point obtaining module, configured to obtain a second image, and obtain human body key points of the second image according to the preset training model;

A second portrait feature acquisition module, configured to acquire a second portrait feature according to the key points of the human body of the second image;

The portrait feature matching verification module is configured to match the second portrait feature with the standard image feature, and if the match is successful, output a result of verification.
The image feature configuration and verification device according to claim 6, wherein the first portrait feature acquisition module includes a coordinate acquisition unit, a feature interval value acquisition unit, and a first portrait feature setting unit;

The coordinate acquiring unit is configured to acquire coordinates of N human key points of the first image according to positions of N human key points of the first image;

The characteristic interval value obtaining unit is used to calculate the coordinates of the key points of the human body of the N first images using a moving index weighted average algorithm to obtain the first characteristic interval value;

The first portrait feature setting unit is configured to use the first feature interval value as a first portrait feature.
The image feature configuration and verification device according to claim 6, wherein the preset training model includes a micro-expression recognition model and a gesture recognition model, and the first portrait feature acquisition module includes an image set acquisition unit, A standard facial feature acquisition unit, a standard limb image feature acquisition unit, and a first portrait feature acquisition unit;

The image set obtaining unit is configured to obtain a first face image set and a first limb image set according to N human body key points of the first image;

The standard facial feature acquisition unit is configured to input the first facial image set into the micro-expression recognition model to obtain standard facial image features;

The standard limb feature acquisition unit is configured to input the first limb image set into the gesture recognition model to obtain standard limb image features;

The first portrait feature acquisition unit is configured to compose the standard face image feature and the standard limb image feature into the first portrait feature.
The image feature configuration and verification device according to claim 8, wherein the second portrait feature acquisition module is further used to:

Acquiring a second face image and a second limb image according to the key points of the human body of the second image;

Input the second face image into the micro-expression recognition model to obtain the characteristics of the test face image;

Input the second limb image into the posture recognition model to obtain test limb image characteristics;

The test face image feature and the test limb image feature are combined to form the second portrait feature.
The image feature configuration and verification device according to claim 6, wherein the first image acquisition module is further used to:

Get the first video data;

Frame the first video data according to a preset time to obtain N images to be processed;

Create a normalized image, obtain height and width information of the normalized image, and obtain a normalized image of the N images to be processed based on the height and width information, and normalize the normalized image Replace the original pixel values of the N images to be processed to obtain the N first images.
A computer device, including a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor, characterized in that, when the processor executes the computer-readable instructions, it is implemented as follows step:

Acquire N first images, where N is a positive integer greater than or equal to 2;

Acquiring N human key points of the first image according to a preset training model;

Acquiring the first portrait feature according to the N human key points of the first image;

Configuring the first portrait feature as a standard image feature;

Acquiring a second image, acquiring human key points of the second image according to the preset training model;

Acquiring a second portrait feature according to the key points of the human body of the second image;

The second portrait feature is matched with the standard image feature, and if the match is successful, a result of verification is output.
The computer device according to claim 11, wherein the acquiring the first portrait feature based on the N human key points of the first image includes:

Acquiring the coordinates of the N human key points of the first image according to the positions of the N human key points of the first image;

Use the moving index weighted average algorithm to calculate the coordinates of the key points of the human body of the N first images to obtain the first feature interval value;

Use the first feature interval value as the first portrait feature.
The computer device according to claim 11, wherein the preset training model includes a micro-expression recognition model and a gesture recognition model;

The obtaining the first portrait feature according to the N human key points of the first image includes:

Acquiring a first face image set and a first limb image set according to the N human body key points of the first image;

Input the first face image set into the micro-expression recognition model to obtain standard face image features;

Input the first limb image set into the gesture recognition model to obtain standard limb image features;

The standard face image feature and the standard limb image feature are combined into the first portrait feature.
The computer device according to claim 13, wherein the acquiring the second portrait feature according to the key points of the human body of the second image comprises:

Acquiring a second face image and a second limb image according to the key points of the human body of the second image;

Input the second face image into the micro-expression recognition model to obtain the characteristics of the test face image;

Input the second limb image into the posture recognition model to obtain test limb image characteristics;

The test face image feature and the test limb image feature are combined to form the second portrait feature.
The computer device according to claim 11, wherein the acquiring N first images includes:

Get the first video data;

Frame the first video data according to a preset time to obtain N images to be processed;

Create a normalized image, obtain height and width information of the normalized image, and obtain a normalized image of the N images to be processed based on the height and width information, and normalize the normalized image Replace the original pixel values of the N images to be processed to obtain the N first images.
One or more non-volatile readable storage media storing computer-readable instructions, which when executed by one or more processors, cause the one or more processors to perform the following steps:

Acquire N first images, where N is a positive integer greater than or equal to 2;

Acquiring N human key points of the first image according to a preset training model;

Acquiring the first portrait feature according to the N human key points of the first image;

Configuring the first portrait feature as a standard image feature;

Acquiring a second image, acquiring human key points of the second image according to the preset training model;

Acquiring a second portrait feature according to the key points of the human body of the second image;

The second portrait feature is matched with the standard image feature, and if the match is successful, the result of verification is output.
The non-volatile readable storage medium according to claim 16, wherein the acquiring the first portrait feature based on the N human key points of the first image includes:

Acquiring the coordinates of the N human key points of the first image according to the positions of the N human key points of the first image;

Use the moving index weighted average algorithm to calculate the coordinates of the key points of the human body of the N first images to obtain the first feature interval value;

Use the first feature interval value as the first portrait feature.
The non-volatile readable storage medium of claim 16, wherein the preset training model includes a micro-expression recognition model and a gesture recognition model;

The obtaining the first portrait feature according to the N human key points of the first image includes:

Acquiring a first face image set and a first limb image set according to the N human body key points of the first image;

Input the first face image set into the micro-expression recognition model to obtain standard face image features;

Input the first limb image set into the gesture recognition model to obtain standard limb image features;

The standard face image feature and the standard limb image feature are combined into the first portrait feature.
The non-volatile readable storage medium of claim 18, wherein the acquiring the second portrait feature according to the key points of the human body of the second image includes:

Acquiring a second face image and a second limb image according to the key points of the human body of the second image;

Input the second face image into the micro-expression recognition model to obtain the characteristics of the test face image;

Input the second limb image into the posture recognition model to obtain test limb image characteristics;

The test face image feature and the test limb image feature are combined to form the second portrait feature.
The non-volatile readable storage medium of claim 16, wherein the acquiring N first images includes:

Get the first video data;

Frame the first video data according to a preset time to obtain N images to be processed;

Create a normalized image, obtain height and width information of the normalized image, and obtain a normalized image of the N images to be processed based on the height and width information, and normalize the normalized image Replace the original pixel values of the N images to be processed to obtain the N first images.