WO2020258119A1 - Face recognition method and apparatus, and electronic device - Google Patents

Abstract

A face recognition method and apparatus, and an electronic device. It can be recognized whether a face is authentic, thereby improving the security of face recognition. The face recognition method comprises: obtaining a first target image and a first eye image of a first recognition target; performing iris-based face anti-spoofing determination according to the first eye image to determine whether the first recognition target is a living face, and outputting a liveness determination result; performing feature template matching according to the first target image, and outputting a matching result; and outputting a face recognition result according to the liveness determination result and the matching result.

Description

Method, device and electronic equipment for face recognition Technical field

This application relates to the field of biometric recognition technology, and more specifically, to a method, device and electronic device for face recognition.

Background technique

Face recognition is a kind of biometric recognition technology based on human facial feature information. Use a camera or camera to collect images or video streams containing human faces, and automatically detect and track human faces in the images, and then perform facial image preprocessing, image feature extraction, and matching and recognition of the detected human faces. Related technology is usually called face recognition or facial recognition. With the rapid development of computer and network technology, face recognition technology has been widely used in many industries and fields such as smart access control, mobile terminals, public safety, entertainment, and military.

Currently, face recognition generally uses two-dimensional (Two Dimensional, 2D) image recognition based on the face, to determine whether the 2D image is a specific user face, and does not determine whether the 2D image is from a living human face. In other words, In some technologies, 2D face recognition based on 2D images has no anti-counterfeiting function and poor security performance.

Summary of the invention

The embodiments of the present application provide a face recognition method, device, and electronic equipment, which can recognize the authenticity of a human face, thereby improving the security of face recognition.

In the first aspect, a face recognition method is provided, including:

Acquiring a first target image and a first eye image of the first recognition target;

Perform iris-based face anti-counterfeiting judgment according to the first eye image to determine whether the first recognition target is a living human face and output a living judgment result;

Performing feature template matching according to the first target image, and outputting a matching result;

The face recognition result is output according to the living body judgment result and the matching result.

The present application provides a face recognition solution with anti-counterfeiting function. By acquiring the first target image and the first eye image of the first recognition target, the face anti-counterfeiting is performed based on the iris features in the first eye image. Based on whether the eye image including the iris is from a living human face, the feature template matching is performed according to the first target image to determine whether it is a user, thereby greatly improving the security of the face recognition device and electronic equipment.

In a possible implementation manner, the outputting the face recognition result according to the living body judgment result and the matching result includes:

When the matching result is successful, output the face recognition result according to the living body judgment result; or, when the living body judgment result is a living body, output the face recognition result according to the matching result; or, in the matching When the result is failure or the living body judgment result is non-living body, the face recognition result is output.

In a possible implementation manner, the performing feature template matching according to the first target image and outputting the matching result includes:

Performing face detection based on the first target image;

When the face detection is successful, acquiring a first face image based on the first target image;

Matching the first face image with a plurality of pre-stored first feature templates;

When the first face image is successfully matched with any one of the plurality of first feature templates, the output matching result is successful; or,

When the first face image fails to match the multiple first feature templates, output a matching result as failure;

Or, when the face detection fails, the output matching result is failed.

In a possible implementation manner, the acquiring the first target image and the first eye image of the first recognition target includes:

A first target image of the first recognition target is acquired, and the first eye image is acquired based on the first target image.

In a possible implementation manner, the first eye image is a two-dimensional infrared image.

In a possible implementation manner, the first eye image is a human eye area image or an iris area image including the iris.

In a possible implementation manner, the performing an iris-based face anti-counterfeiting judgment according to the first eye image includes:

Processing the first eye image by using a histogram equalization method to obtain a first optimized eye image;

According to the first optimized eye image, the iris-based face anti-counterfeiting judgment is performed.

In a possible implementation manner, the performing an iris-based face anti-counterfeiting judgment according to the first optimized eye image includes:

Perform classification processing on the first optimized eye image through a neural network to determine whether the first recognition target is a living human face.

In a possible implementation manner, the first eye image includes a first left eye image and/or a first right eye image, and the histogram equalization method is used to perform a calculation on the first eye image. Image processing to obtain the first optimized eye image includes:

Processing the first left-eye eye image by using the histogram equalization method to obtain a first optimized left-eye eye image; and/or

Using the histogram equalization method to process the first right eye image to obtain a first optimized right eye image.

In a possible implementation manner, the first eye image includes: the first left eye image or the first right eye image;

The neural network includes: a first flattened layer, at least one first fully connected layer, and at least one first excitation layer.

In a possible implementation manner, the classification processing of the first optimized eye image through a neural network includes:

Processing the first optimized left eye image or the first optimized right eye image through the first flattening layer to obtain multiple eye pixel values;

Using the at least one first fully connected layer to fully connect the multiple eye pixel values to obtain multiple characteristic constants;

Through the at least one first excitation layer, nonlinearization processing or classification processing is performed on the plurality of characteristic constants.

In a possible implementation manner, the neural network includes: the first flattened layer, two first fully connected layers, and two first excitation layers.

In a possible implementation manner, the excitation functions in the two first excitation layers are the modified linear unit ReLU function and the Sigmoid function, respectively.

In a possible implementation manner, the first eye image includes: the first left eye image and the first right eye image;

The neural network includes a first network, a second network, and a third network;

The first network includes: a second flattening layer, at least one second fully connected layer, and at least one second excitation layer;

The second network includes: a third flattening layer, at least one third fully connected layer, and at least one third excitation layer;

The third network includes: at least one fourth fully connected layer and at least one fourth excitation layer.

In a possible implementation manner, the classification processing of the first optimized eye image through a neural network includes:

Processing the first optimized left-eye eye image through the first network to obtain a left-eye classification feature value;

Processing the first optimized right eye image through the second network to obtain right eye classification feature values;

Fully connect the left-eye classification feature value and the right-eye classification feature value through the third network.

In a possible implementation manner, the first network includes: the second flattened layer, two second fully connected layers, and two second excitation layers;

The second network includes: the third flattening layer, two third fully connected layers, and two third excitation layers;

The third network includes: a fourth fully connected layer and a fourth excitation layer.

In a possible implementation manner, the excitation functions in the two second excitation layers are the modified linear unit ReLU function and the Sigmoid function respectively; and/or,

The excitation functions in the two third excitation layers are the modified linear unit ReLU function and the Sigmoid function respectively; and/or,

One of the excitation functions in the fourth excitation layer is a modified linear unit ReLU function.

In a possible implementation manner, the method further includes:

Acquiring a second eye image of the second recognition target;

According to the second eye image, the iris-based face anti-counterfeiting judgment is performed to determine whether the second recognition target is a living human face, wherein the result of the face anti-counterfeiting judgment is used to establish a face feature template.

In a possible implementation manner, the second eye image is a second eye infrared image.

In a possible implementation manner, the method further includes:

A second target image of the second recognition target is acquired, the second eye image is acquired based on the second target image, and the face feature template is established based on the second target image.

In a possible implementation manner, the method further includes:

Performing face detection based on the second target image;

Wherein, the establishment of a facial feature template based on the second target image includes:

When the face detection is successful, a second face image is acquired based on the second target image, and the face feature template is established based on the second face image.

In a possible implementation, the establishing the face feature template based on the second face image includes:

Judging whether the second face image belongs to a face feature template library;

When the second face image belongs to the face feature template library, the second face image is matched with multiple face feature templates in the face feature template library.

When the second face image does not belong to the face feature template library, the iris-based face anti-counterfeiting judgment is performed according to the second eye image, and when it is determined that the second recognition target is a living face, The second face image is established as a face feature template.

In a possible implementation, the matching the second face image with multiple face feature templates in the face feature template library includes:

When the matching is successful, perform iris-based face anti-counterfeiting judgment according to the second eye image;

When it is determined that the second recognition target is a living face, the second face image is established as a face feature template.

In a possible implementation manner, when the matching is successful, the iris-based face anti-counterfeiting judgment according to the second eye image includes:

When the matching is successful, obtain the 3D point cloud data of the second recognition target;

When the 3D point cloud data is a valid point cloud, iris-based face anti-counterfeiting judgment is performed according to the second eye image.

In a possible implementation manner, the acquiring the second eye image based on the second target image includes:

Acquiring a face area image based on the second target image;

Acquiring the second eye image based on the face region image.

In a possible implementation manner, the second eye image is a human eye area image or an iris area image including the iris.

In a possible implementation manner, the performing iris-based face anti-counterfeiting judgment according to the second eye image includes:

Using a histogram equalization method to process the second eye image to obtain a second optimized eye image;

According to the second optimized eye image, the iris-based face anti-counterfeiting judgment is performed.

In a possible implementation manner, the performing the iris-based anti-counterfeiting judgment of the human face according to the second optimized eye image includes:

Perform classification processing on the second optimized eye image through a neural network to determine whether the second recognition target is a living human face.

In a possible implementation manner, the second eye image includes a second left eye image and/or a second right eye image, and the second optimized eye image is performed on the neural network. Classification processing includes:

Perform classification processing on the second left eye image and/or the second right eye image through a neural network.

In a possible implementation manner, the neural network includes:

At least one flattening layer, at least one fully connected layer and at least one excitation layer.

In a second aspect, a face recognition device is provided, including a processor, configured to execute the face recognition method in the first aspect or any possible implementation of the first aspect.

In a third aspect, an electronic device is provided, including the face recognition device in the second aspect or any possible implementation of the second aspect.

In a fourth aspect, a chip is provided. The chip includes an input and output interface, at least one processor, at least one memory, and a bus. The at least one memory is used to store instructions, and the at least one processor is used to call Instructions to execute the method in the first aspect or any possible implementation of the first aspect.

In a fifth aspect, a computer-readable medium is provided for storing a computer program, and the computer program includes instructions for executing the above-mentioned first aspect or any possible implementation of the first aspect.

In a sixth aspect, a computer program product including instructions is provided. When the computer runs the instructions of the computer program product, the computer executes the first aspect or any of the possible implementations of the first aspect. Methods of face recognition.

Specifically, the computer program product can run on the electronic device of the third aspect.

Description of the drawings

Fig. 1(a) is a schematic block diagram of a face recognition device according to an embodiment of the present application.

Figure 1(b) is a schematic flowchart of a face recognition method according to an embodiment of the present application.

Fig. 1(c) is a schematic block diagram of a convolutional neural network according to an embodiment of the present application.

Fig. 2 is a schematic flowchart of a face recognition method according to an embodiment of the present application.

Figure 3 (a) is an infrared image of a human face of a three-dimensional model according to an embodiment of the present application.

Figure 3 (b) is an infrared image of a human face of a user according to an embodiment of the present application.

Fig. 4 is a schematic flowchart of another method for face recognition according to an embodiment of the present application.

Fig. 5 is a schematic flowchart of another face recognition method according to an embodiment of the present application.

Fig. 6 is a schematic flowchart of another face recognition method according to an embodiment of the present application.

Fig. 7 is a schematic flowchart of another face recognition method according to an embodiment of the present application.

Fig. 8 is a schematic flowchart of a method for anti-counterfeiting discrimination of a face in a face recognition method according to an embodiment of the present application.

Fig. 9 is a schematic flow chart of another method for anti-counterfeiting discrimination of a face in a face recognition method according to an embodiment of the present application.

Fig. 10 is a schematic block diagram of a convolutional neural network according to an embodiment of the present application.

Fig. 11 is a schematic diagram of a fully connected layer according to an embodiment of the present application.

Fig. 12 is a schematic block diagram of another convolutional neural network according to an embodiment of the present application.

FIG. 13 is a schematic flow chart of another method for anti-counterfeiting discrimination in a face recognition method according to an embodiment of the present application.

Fig. 14 is a schematic block diagram of another convolutional neural network according to an embodiment of the present application.

Fig. 15 is a schematic flowchart of a face registration method in a face recognition method according to an embodiment of the present application.

Fig. 16 is a schematic flowchart of another face registration method in a face recognition method according to an embodiment of the present application.

Fig. 17 is a schematic flowchart of another face registration method in a face recognition method according to an embodiment of the present application.

Fig. 18 is a schematic block diagram of a face recognition device according to an embodiment of the present application.

Fig. 19 is a schematic block diagram of an electronic device according to an embodiment of the present application.

Detailed ways

The technical solutions in the embodiments of the present application will be described below in conjunction with the drawings.

The embodiments of the present application may be applicable to optical face recognition systems, including but not limited to products based on optical face imaging. The optical face recognition system can be applied to various electronic devices with image acquisition devices (such as cameras). The electronic devices can be mobile phones, tablet computers, smart wearable devices, smart door locks, etc. The embodiments of the present disclosure are to this Not limited.

It should be understood that the specific examples in this document are only to help those skilled in the art to better understand the embodiments of the present application, rather than limiting the scope of the embodiments of the present application.

It should also be understood that the formulas in the embodiments of the present application are only examples, and do not limit the scope of the embodiments of the present application. Each formula can be modified, and these modifications should also fall within the protection scope of the present application.

It should also be understood that, in the various embodiments of the present application, the size of the sequence number of each process 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 correspond to the embodiments of the present application. The implementation process constitutes any limitation.

It should also be understood that the various implementation manners described in this specification can be implemented individually or in combination, which is not limited in the embodiments of the present application.

Unless otherwise specified, all technical and scientific terms used in the embodiments of the present application have the same meaning as commonly understood by those skilled in the technical field of the present application. The terminology used in this application is only for the purpose of describing specific embodiments, and is not intended to limit the scope of this application. The term "and/or" as used in this application includes any and all combinations of one or more related listed items.

In order to facilitate understanding, first, in conjunction with Figure 1 (a), Figure 1 (b) and Figure 1 (c), the process of unlocking an electronic device based on 2D image face recognition is briefly introduced.

As shown in FIG. 1( a ), the face recognition device 10 includes an infrared light emitting module 110, an infrared image acquisition module 120 and a processor 130. Wherein, the infrared light emitting module 110 is used to emit infrared light signals, which may be an infrared light emitting diode (Light Emitting Diode, LED), or may also be a vertical cavity surface emitting laser (Vertical Cavity Surface Emitting Laser, VCSEL), etc. The infrared light emitting light source is not limited in the embodiment of the present application. The infrared image acquisition module 120 may be an infrared camera, which includes an infrared image sensor, which is used to receive infrared light signals and convert the received infrared light signals into corresponding electrical signals, thereby generating infrared images. The processor 130 may be a microprocessor unit (MPU), which may control the infrared light emitting module 110 and the infrared image acquisition module 120 to collect facial images and perform facial image recognition.

Specifically, as shown in Figure 1(b), when face recognition is required, the specific 2D recognition process is as follows:

S110: Collect 2D infrared images of the identified target. Specifically, the infrared light-emitting module 110 emits infrared light, and the infrared light is irradiated on a recognition target. The recognition target may be a user's face, or may be a photo, a 3D model or any other object. The infrared reflected light reflected by the surface of the identified target is received by the infrared image sensor 120 and converted into a 2D infrared image, and the infrared image sensor 120 transmits the 2D infrared image to the processor 130.

S120: face detection (face detection). That is, it receives 2D infrared images and detects whether there are human faces on the 2D infrared images. For example, a single convolutional neural network (Convolutional Neural Networks, CNN) is used to perform face detection on 2D infrared images. First, train a face detection convolutional neural network to determine whether a human face is not a face, and input the data of the 2D infrared image into the face detection convolutional neural network. Through the steps of convolution calculation, the characteristics of the 2D infrared image data After extraction, classification is performed to determine whether there is a human face on the 2D infrared image.

Specifically, as shown in Figure 1(c), the convolutional neural network mainly includes a convolutional layer 101 (convolutional layer), an activation layer 102 (activation layer), a pooling layer 103 (pooling layer), and a fully connected layer 104 ( fully-connected layer). Among them, each convolutional layer in the convolutional neural network is composed of several convolutional kernels, and the parameters of each convolutional kernel are optimized through the back propagation algorithm. The purpose of the convolution operation is to extract different features of the input. Different convolution kernels extract different feature maps. More layers of convolutional networks can iteratively extract more complex features from low-level features such as edge features and line features. feature. The activation layer uses activation functions to introduce nonlinearity to the convolutional neural network. Commonly used activation functions include sigmoid, tanh, and ReLU functions. Usually after the convolutional layer, a feature with a large dimension is obtained. The pooling layer cuts the feature into several regions, and takes the maximum value (max pooling) or average value (average pooling) to obtain new features with smaller dimensions Figure. The fully connected layer combines all local features into global features, which are used to calculate the final score of each category to determine the category of the input data.

S121: If there is a face on the 2D infrared image, perform face cut on the 2D infrared image. Specifically, the fully connected layer of the above-mentioned face detection convolutional neural network is changed to a convolutional layer, so that the network becomes a fully convolutional network, and the 2D infrared image passes through the fully convolutional network to obtain a feature map. Each feature map The "point" corresponds to the probability that the location mapped to the original image area belongs to the face, and the face whose probability is greater than the set threshold is regarded as the face candidate frame. Cut the image in the face candidate frame in the 2D infrared image to form a new face 2D infrared image.

S122: If there is no human face in the 2D infrared image, add 1 to the restart parameter.

If there is no face in the 2D infrared image, the face detection fails. In other words, the recognition target is not the user, and the matching fails.

Optionally, face detection can also be performed by cascading CNN, Dlib, OpenCV and other methods, and a new 2D infrared image of the face can be cut. This is not limited in the embodiments of this application.

S130: 2D face recognition (face recognition). That is, the 2D infrared image of the human face formed in S131 is recognized, and it is determined whether the 2D infrared image of the human face is the user's face. For example, using a convolutional neural network method for face recognition, specifically, first train a face recognition convolutional neural network to determine whether it is a user's face, and the face recognition convolutional neural network is based on multiple templates in the template library. Feature template classification. Input the data of the 2D infrared image of the face into the face recognition convolutional neural network. After the features of the 2D infrared image of the face are extracted through steps such as convolution calculation, the classification and judgment are performed to determine the 2D infrared image of the face Whether it matches with multiple feature templates in the template library.

S131: If the matching is successful, the 2D infrared image of the face is the face image of the user, and the 2D recognition is successful. Further, the electronic device where the face recognition apparatus 10 is located can be unlocked, and the application on the electronic device can also be unlocked.

S132: If the matching fails, the 2D infrared image of the face is not the user's face image, and the 2D recognition fails, and the restart parameter is increased by 1.

S140: Determine whether the restart parameter is less than the first threshold.

S141: If the restart parameter is less than the first threshold, enter S110;

S142: If the restart parameter is greater than or equal to the first threshold, the recognition fails.

In Figure 1(b), the face recognition device 10 performs face recognition by collecting 2D infrared images of human faces and judging whether the 2D images of the human faces match the characteristic faces in the characteristic face template library, thereby identifying the electronic device Unlock with an application (APP) on the electronic device. Since during the unlocking process, the face recognition device 10 only performs face recognition based on the two-dimensional features on the 2D image, it cannot identify whether the collected 2D infrared image is derived from a living human face or other photos, videos, and other non-living human faces. Objects, in other words, the face recognition device 10 does not have an anti-counterfeiting function. It can unlock electronic equipment and applications by stealing photos, videos and other information with the user’s face, so the face recognition device and electronic equipment are safe Performance has been greatly affected.

Since the iris of living human eyes and non-living human eyes (human eye images in photos or videos, human eye models in three-dimensional models) have significant differences in infrared light reflection, the resulting infrared iris of living human eyes and non-living human eyes The images are therefore very different. Based on this, the embodiment of the present application provides a face recognition solution with anti-counterfeiting function. By acquiring an infrared image of the eye of the recognition target, and performing face anti-counterfeiting on the infrared image of the eye, it is determined whether it is from a living person of the user Eye iris, thereby judging whether the recognition target is a living human face, thus greatly improving the safety of face recognition devices and electronic equipment.

Hereinafter, the face recognition method provided by the embodiment of the present application will be introduced in detail with reference to FIGS. 2-9.

FIG. 2 is a face recognition method 200 provided by an embodiment of this application, including:

S210: Obtain a target image and an eye image of the recognition target;

S220: Perform iris-based face anti-counterfeiting judgment according to the eye image to determine whether the recognition target is a living human face and output a living judgment result;

S230: Perform feature template matching according to the target image, and output a matching result;

S240: Output a face recognition result according to the living body judgment result and the matching result.

It should be understood that the recognition target is also called the first recognition target, the second recognition target, etc., which can be used to distinguish different target objects. Accordingly, the target image and the eye image of the recognition target can also be called the first recognition target. Target image or second target image, first eye image or second eye image, etc. The recognition target includes, but is not limited to, any objects such as human faces, photos, videos, and three-dimensional models. For example, the recognition target may be a user's face, other people's faces, a user's photo, a curved surface model with photos, etc.

Optionally, the eye image may be an eye area image of a living human face, and the human eye structure is composed of parts such as sclera, iris, pupil lens, and retina. The iris is a circular part located between the black pupil and the white sclera. It contains many intertwined spots, filaments, crowns, stripes, crypts and other detailed features. These characteristics determine the uniqueness of iris characteristics, and also determine the uniqueness of identification. Because it has the unique characteristics of identifying the human body, it is also difficult to copy and forge, so it can be used for face anti-counterfeiting and face recognition.

Optionally, when the iris image in the eye image can be used to distinguish between a living human face and a non-living human face, the eye image can be a color image generated by visible light, or an infrared image generated by infrared light or other Image, this embodiment of the application does not limit this.

Preferably, in the embodiment of the present application, the eye image is an infrared image. The following takes the eye image as an infrared image as an example for detailed description. Specifically, the infrared (Infrared Radation, IR) image is an image formed based on the infrared light signal reflected from the surface of the identification target, which is expressed as a gray scale image, and the identification target is expressed by the gray scale of the image pixels. Appearance shape. The eye image of the recognition target is an eye infrared image including the iris area, for example, an infrared image based on a living human eye of the user's face and reflected infrared light reflected by the iris, or an infrared image based on the eye and The infrared image formed by the reflected infrared light reflected by the iris area and so on.

Due to the special shape, composition and structure of the iris of a living human eye, the reflected infrared light reflected by it is quite different from the reflected infrared light reflected by objects such as photos and models. Therefore, different recognitions can be distinguished in the eye image including the iris The iris information of the target is used to distinguish between a living human face and a non-living human face. In other words, the eye image of the living human face including the iris is different from the eye image of the non-living human face including the iris, and the difference is large. The embodiment of the present application utilizes this difference to perform the anti-counterfeiting judgment of the human face based on the eye image including the iris. Wherein, the non-living human face includes but is not limited to: user face photos, user face videos, user face photos placed on a three-dimensional curved surface, user face models, and so on.

For example, as shown in Figure 3, Figure 3 (a) is an infrared image of the face of the three-dimensional model. The infrared image of the eye is only the human eye model, and the "iris" area of the eye model is only a simulated iris shape The schematic picture of, does not contain the iris information of a living human eye. The picture (b) in FIG. 3 is an infrared image of the face of the user's living body, and the characteristic information of the iris of the human eye of a real person can be reflected in the picture, which is completely different from the picture (a) in FIG. 3.

After obtaining the eye image including the iris, based on the feature information of the eye image, the face anti-counterfeiting judgment is performed to determine whether the iris of the recognition target is the iris of a living human face, so as to determine whether the recognition target is a living human Face, achieve the effect of anti-counterfeiting of human face.

Specifically, in the process of face recognition, in addition to judging whether the recognition target is a living body face, it is also necessary to perform feature template matching, which combines the feature template matching and the result of living body judgment to perform face recognition. The feature template matching is to match the target image with the feature template of at least one user, and it can be determined whether the target image belongs to the user's image. Optionally, the feature template is feature data of multiple faces or partial face images of the user under different conditions such as different angles and different environments. The feature template is stored in a face recognition device, in particular, it can be stored in a memory in the device.

Combining face anti-counterfeiting judgment and feature template matching judgment can enhance the reliability of the face recognition process and improve safety performance.

Currently, there are different security levels for face anti-counterfeiting, as shown in Table 1 below. Different levels represent different face anti-counterfeiting requirements. That is, for example: when the anti-counterfeiting level is level 1, the 2D printed static flat face can be recognized.

Table 1

The face recognition device and face recognition method in Figure 1(a) and Figure 1(b) cannot determine whether the collected 2D image is from a photo or a real face, so it does not have the anti-counterfeiting function and cannot reach the face anti-counterfeiting level in Table 1. Level 1. However, in the embodiments of the present application, since the characteristic information of the human iris can be obtained from the eye image including the iris, the living human face and the non-living human face can be identified, so that the face anti-counterfeiting level 5 can be reached, and the anti-counterfeiting and recognition Safety performance has been greatly improved.

Optionally, in the embodiment of the present application, the infrared image of the recognition target may be obtained by the infrared image acquisition device, and then the infrared image of the eyes of the recognition target may be obtained from the infrared image of the recognition target.

In a possible implementation manner, a rough eye area is initially detected and cut out from the infrared image of the identified target, and then it is determined whether there is an iris in the rough eye area, and the iris infrared image is obtained or includes Eye infrared image with iris infrared image, specifically, obtaining left eye iris infrared image and/or right eye iris infrared image, or left eye infrared image including left eye iris infrared image and/or right eye iris infrared image Infrared image of right eye.

For example: according to the left eye area and the right eye area symmetrically located on the front face image, the recognition of the symmetrical area in the infrared image can detect the approximate eye area of the recognition target. Or according to the geometric features of the "three courts and five eyes" of the human face, a rough eye area is cut out from the face image obtained by face detection and cropping to form an infrared image of the eye. Optionally, the human face is divided into three equal parts longitudinally to form the upper court, the atrium and the lower court, and the approximate eye area is located at 1/5 of the bottom of the upper court and 3/5 of the upper atrium. Optionally, it is detected whether there is an iris in the approximate eye area according to the gray value change in the approximate eye area or other methods.

In another possible implementation manner, the infrared image of the iris or the infrared image of the eye including the infrared image of the iris may be directly detected and cut out from the infrared image of the recognition target.

For example, in an infrared grayscale image of a human face, the gray value of the eye area is compared with other areas of the face, the gray value of the iris area of the eye is small, and the gray value of the scleral area is large, and the iris area The gray value change gradient between and the sclera area is obvious. Therefore, the eye area and the iris area can be detected by detecting the gray change feature in the infrared image, and the coordinates of the eye area and the iris area in the image can be obtained, and the result can be cut An infrared image of the iris or an infrared image of the eye including an infrared image of the iris.

Specifically, in the embodiments of the present application, when the infrared image of the eye includes an iris image, it is an effective infrared image of the eye, which can be directly used for face recognition in vivo, or the iris can be cut out from the infrared image of the eye. Infrared image, used for live face recognition. When the infrared image of the eye does not include the iris image, that is, when the user closes his eyes or there is no iris or iris pattern in the other recognition target, the infrared image of the eye is an invalid infrared image of the eye and cannot be used for live face recognition . For the convenience of description, the infrared image of the eye including the iris image and the infrared image of the iris are both referred to as the infrared image of the eye hereinafter. Specifically, the infrared image of the eye includes: an infrared image of the left eye and/or an infrared image of the right eye.

It should be understood that in the embodiments of the present application, any other algorithms or methods that can recognize the iris of the eye can also be used to obtain an eye image from the infrared image of the recognition target and determine whether there is an iris in the eye image. This is not limited.

Specifically, in the embodiments of the present application, the feature template matching of 2D recognition can be performed based on the acquired 2D target image of the recognition target, and the face recognition can be performed and output based on the feature template matching result of 2D recognition and the result of face anti-counterfeiting judgment. Face recognition result.

In the embodiments of the present application, when the feature template is a 2D image, feature template matching is a main step and implementation in 2D recognition. In the following, 2D recognition can also be understood as feature template matching in 2D recognition.

Optionally, 2D recognition can be performed first, and on the basis of 2D recognition, the iris-based face anti-counterfeiting is performed again based on the eye infrared image according to the 2D recognition result, so that the recognition process is more safe and effective. For example, as shown in FIG. 4, another method 300 for face recognition provided by an embodiment of the present application includes:

S310: Acquire a target image of the recognition target;

S340: Perform 2D recognition based on the target image;

When the target image is successfully matched with any one of the multiple characteristic templates, the 2D recognition is successful, indicating that the target image includes the user's face image. When the target image fails to match with multiple feature templates, the 2D recognition fails, which means that the target image does not include the user's face image.

Optionally, in this embodiment of the present application, the 2D recognition may be the same or similar to the 2D recognition process in FIG. 1(b).

S351: When the 2D recognition is successful, obtain an eye image of the recognition target based on the target image;

S352: When the 2D recognition fails, it is determined that the face recognition fails, and the first face recognition result is output;

Optionally, the first face recognition result may include, but is not limited to, specific information such as failures and non-authenticated users.

S360: Perform iris-based face anti-counterfeiting judgment according to the eye image to determine whether the recognition target is a living human face;

S371: When the recognition target is a living human face, determine that the face recognition is successful, and output a second face recognition result;

Optionally, the second face recognition result may include, but is not limited to, specific information such as success and biometric authentication of the user.

S372: When the recognition target is a non-living face, determine that the face recognition fails, and output a third face recognition result.

Optionally, the third face recognition result may include, but is not limited to, specific information such as failures and non-living authenticated users.

Optionally, in the embodiment of the present application, the target image may be an infrared image, a visible light image, or other images. When the target image is an infrared image, an eye infrared image of the recognition target is acquired based on the infrared image, and an iris-based human face anti-counterfeiting judgment is performed according to the eye infrared image.

Optionally, face anti-counterfeiting can be performed first, and on the basis of face anti-counterfeiting, 2D recognition can be performed according to the result of face anti-counterfeiting, which can exclude non-living human faces in advance and improve the efficiency of recognition. For example, as shown in FIG. 5, another face recognition method 400 provided by an embodiment of the present application includes:

S410: Obtain a target image of the recognition target;

S440: Acquire an eye image of the recognition target based on the target image;

S450: Perform iris-based face anti-counterfeiting judgment according to the eye image to determine whether the recognition target is a living human face;

S461: When the recognition target is a living human face, perform 2D recognition based on the target image;

Optionally, the 2D recognition in this step may be the same as step S340 in FIG. 4, and the specific implementation manner may refer to the foregoing solution, which will not be repeated here.

S462: When the recognition target is a non-living human face, determine that the face recognition fails, and output a fourth face recognition result;

Optionally, the fourth face recognition result may include, but is not limited to, specific information such as failure and non-living body.

S471: When the 2D recognition is successful, it is determined that the face recognition is successful, and the fifth face recognition result is output.

Optionally, the fifth face recognition result may include, but is not limited to, specific information such as success and biometric authentication of the user.

S472: When the 2D recognition fails, it is determined that the face recognition fails, and the sixth face recognition result is output.

Optionally, the sixth face recognition result may include, but is not limited to, specific information such as failure and a living non-authenticated user.

Optionally, in step S310 and step S410, the target image of the recognition target may be acquired through the image acquisition module. The image acquisition module may be the infrared image acquisition module 120 in FIG. 1(a).

Optionally, the infrared image acquisition module may include an infrared photoelectric sensor, wherein the infrared photoelectric sensor includes a plurality of pixel units, and each pixel unit is used to collect the reflected infrared light signal after the infrared light is reflected on the surface of the identification target , And convert the reflected infrared light signal into a pixel electrical signal corresponding to its light intensity. The value of the electrical signal of each pixel corresponds to a pixel of the infrared image, and its size is expressed as the gray value of the infrared image. Therefore, an infrared image formed by a pixel matrix composed of multiple pixel units can also be expressed as a numerical matrix composed of gray values of multiple pixel points. Optionally, the gray value range of each pixel is between 0 and 255, the gray value 0 represents black, and the gray value 255 represents white.

Optionally, step S351 may specifically include: 3D face reconstruction. That is, when the 2D recognition is successful, the 3D data of the recognition target is obtained, and the 3D face reconstruction is performed based on the 3D data. If the 3D face reconstruction is successful, the eye image of the recognition target is obtained based on the target image, and the The eye image is subjected to the iris-based face anti-counterfeiting judgment. If the 3D face reconstruction fails, the face anti-counterfeiting judgment is not performed. Specifically, the reconstructed face image reflects the feature information of the face from the three-dimensional space, and on the basis of the success of the 3D face, the face anti-counterfeiting judgment is performed.

Optionally, as shown in FIG. 6, the face recognition method 300 further includes:

S320: face detection, specifically, perform face detection based on the target image;

S331: There is a face, that is, when the face detection is successful, cut the target image to obtain a face image;

S332: There is no face, that is, when the face detection fails, the restart parameter is increased by 1;

S340 specifically includes S341: performing 2D face recognition based on the face image.

S351 specifically includes S353: when the 2D recognition is successful, crop the face image to obtain the eye image of the recognition target;

S352: 2D recognition fails. If it is determined that the face recognition fails, add 1 to the restart parameter;

S373: When the recognition target is not a living human face, add 1 to the restart parameter;

Optionally, as shown in FIG. 7, the face recognition method 400 further includes:

S420: face detection, specifically, perform face detection based on the target image;

S431: There is a face, that is, when the face detection is successful, cut the target image to obtain a face image;

S432: There is no face, that is, when the face detection fails, the restart parameter is increased by 1;

S464: When the recognition target is a non-living human face, add 1 to the restart parameter;

S463: When the recognition target is a living human face, proceed to step S465: Perform 2D face recognition based on the face image.

Optionally, the steps S320 to S332 and steps S420 to S432 may be the same as steps S120 to S122 in FIG. 1(b). For specific implementations, refer to the relevant description in FIG. 1(b), which will not be omitted here. Repeat.

Optionally, in the embodiment of FIG. 6 and FIG. 7, the method further includes: judging the size of the restart parameter, and when the restart parameter is less than the second threshold, enter S310 or enter S410; when the restart parameter is greater than or equal to At the second threshold, it is determined that the recognition fails.

The following describes in detail the process of performing iris-based face anti-counterfeiting judgment based on the eye image in S360 and S450 in conjunction with FIGS. 8 to 14 to determine whether the recognition target is a living face, that is, the detailed process of face anti-counterfeiting.

Optionally, as shown in FIG. 8, a face anti-counterfeiting discrimination method 500 is specifically used to perform face anti-counterfeiting discrimination based on the infrared eye image in step S220 to determine whether the recognition target is a living human face. Specifically, after preprocessing the infrared image of the eye, it is input into a neural network for classification, so as to obtain a face anti-counterfeiting discrimination result.

Optionally, as shown in FIG. 8, the face anti-counterfeiting discrimination method 500 includes:

S510: Preprocess the eye image to obtain an optimized eye image; after the eye image is preprocessed, the contrast of the eye image is increased, the image quality of the eye image is improved, and it is more conducive to the processing and processing of the neural network. classification.

Specifically, the eye image includes a left eye image and a right eye image. Optionally, preprocessing the left eye image and/or the right eye image to obtain the left eye optimized eye image and/or the right eye optimized eye image.

Optionally, the preprocessing process includes S511: eye image equalization. Specifically, the left-eye optimized eye image and/or the right-eye optimized eye image are image equalized.

Optionally, when the eye image is an infrared gray-scale image, a histogram equalization method is used for image equalization processing, which can improve the contrast of the eye infrared image, and can also transform the eye infrared image The grayscale values are almost uniformly distributed images.

Specifically, the histogram equalization step includes:

1) Calculate the probability p(i) of pixels in each gray value of the infrared image of the eye according to the following formula:

Where n is the total number of pixels, n _i is the number of pixels whose gray value is i, and L is the total number of gray values.

2) Calculate the cumulative probability function c(i) of p according to the following formula:

The calculated c is the cumulative normalized histogram of the image.

3) Scale c(i) to y(i) in the range of 0～255 according to the following formula:

y(i)=255*c(i)

Specifically, the gray value of the pixel with the gray value of i in the original infrared image of the eye is changed to y(i), thereby achieving the balance of the infrared image of the eye, and obtaining an optimized infrared image of the eye.

It should be understood that the preprocessing process may also include, but is not limited to, local binary pattern (Local Binary Pattern, LBP) feature processing, normalization, correction, image enhancement, and other processing processes, which are not limited in the embodiment of the present application.

Optionally, in a possible implementation manner, after preprocessing the eye image, a deep learning network is used to classify the preprocessed optimized eye image to determine whether the recognition target is a living body human face. In the embodiments of the present application, the deep learning network includes but is not limited to a neural network, and may also be other deep learning networks. The embodiment of the present application does not limit this. The following uses a neural network as an example to illustrate the classification in the embodiment of the present application Approach.

Optionally, as shown in FIG. 8, the face anti-counterfeiting discrimination method 500 further includes:

S520: Perform classification processing on the optimized eye image through a neural network to determine whether the recognition target is a living human face.

First, construct a neural network structure, for example, a two-layer neural network or a more-layer network structure can be used, and the structure of each layer of the network structure can also be adjusted according to the face information to be extracted, which is not limited in the embodiment of the application.

Second, set the initial training parameters and convergence conditions of the neural network.

Optionally, in the embodiment of the present application, the initial training parameters may be randomly generated, or obtained based on empirical values, or may be parameters of a neural network model pre-trained based on a large amount of true and false face data. The embodiments of this application do not limit this.

Then, input a large number of optimized eye images of the user's live and non-living faces to the neural network. The neural network can process the optimized eye images based on the initial training parameters, and determine the optimal eye image for each optimized eye image. The determination result, further, adjust the structure of the neural network and/or the training parameters of each layer according to the determination result until the determination result meets the convergence condition.

Optionally, in this embodiment of the application, the convergence condition may include at least one of the following:

1. The probability of judging the optimized eye image of the living body face as the optimized eye image of the living body face is greater than the first probability, for example, 98%;

2. The probability of judging the optimized eye image of the non-living human face as the optimized eye image of the non-living human face is greater than the second probability, for example, 95%;

3. The probability of judging the optimized eye image of the living human face as the optimized eye image of the non-living human face is less than the third probability, for example, 2%;

4. The probability of determining the optimized eye image of the non-living human face as the optimized eye image of the living human face is less than the fourth probability, for example, 3%.

After completing the training of the neural network to determine whether it is a living human face, in the process of face recognition, the processed optimized eye image of the current recognition target is input into the neural network, so that the neural network can use the trained The parameters process the optimized eye image of the recognition target to determine whether the recognition target is a living human face.

Optionally, in a possible implementation manner, the left-eye optimized eye image or the right-eye optimized eye image in the optimized eye image is classified by the neural network 50 to determine whether the recognition target It is a living human face.

Optionally, when the optimized eye image is a left eye optimized eye image or a right eye optimized eye image, as shown in FIG. 9, the face anti-counterfeiting determination method 501 includes:

S511: Using a histogram equalization method to perform image equalization processing on the left eye image or the right eye image to obtain an optimized left eye image or an optimized right eye image;

S521: Perform classification processing on the optimized left eye image or the optimized right eye image through the neural network to determine whether the recognition target is a living human face.

As shown in FIG. 10, the neural network 50 includes a flattened layer 510, a fully connected layer 520, and an excitation layer 530.

Wherein, the flattened layer 510 is used to one-dimensionalize the two-dimensional data of the left-eye optimized eye image input to the neural network, that is, to form a one-dimensional array. For example, the left-eye optimized eye image is represented as a two-dimensional matrix of 20*20 pixels, each of which represents a gray value. After flattening, a one-dimensional matrix of 400*1 is formed. That is, 400 pixel values are output. That is, in the embodiment of the present application, the two-dimensional image data is flattened into one-dimensional data through the flattening layer 510, and then the one-dimensional data is input to the fully connected layer to be fully connected.

Specifically, each node in the fully connected layer 520 is connected to each node in the upper layer, and is used to synthesize the features extracted from the previous neural network, and act as a "classifier" in the entire neural network. . For example, as shown in Figure 11, x ₁ to x _n are the output nodes of the previous layer, the fully connected layer 520 includes m fully connected nodes c ₁ to c _n , and m characteristic constants are output, which is convenient for m The characteristic constant classification is used for judgment classification. Specifically, each of the m fully connected nodes includes multiple parameters obtained by the above training convergence, and is used for weighted connection of x ₁ to x _n to obtain a characteristic constant result.

In the following, in the embodiment of the present application, x ₁ to x _n are one-dimensional data output by the flattening layer 510 as an example, the fully connected layer will be described.

Is one-dimensional data x ₁ to x _n, m one-dimensional data is subjected to full output full connection layer connected to the node data of m constants a ₁ to a _m, the formula wherein a ₁ to a _m are as follows:

a ₁ =W ₁₁ *x ₁ +W ₁₂ *x ₂ +W ₁₃ *x ₃ +…+W _1n *x _n +b ₁ ;

a ₂ =W ₂₁ *x ₁ +W ₂₂ *x ₂ +W ₂₃ *x ₃ +…+W _2n *x _n +b ₂ ;

...

a _m =W _m1 *x ₁ +W _m2 *x ₂ +W _m3 *x ₃ +…+W _mn *x _n +b _m ;

Among them, W and b are the weighting parameters and bias parameters in the nodes of the fully connected layer 520, both of which can be obtained by the above-mentioned process of training a convergent neural network.

Optionally, the fully connected layer 520 includes at least one fully connected layer. For example, in the embodiment of the present application, as shown in FIG. 12, the fully connected layer 520 includes a first fully connected layer 521 and a second fully connected layer 522. Specifically, the calculation principles of the two fully connected layers are the same, and both are weighted fully connected to the input one-dimensional array.

Optionally, as shown in FIG. 12, the excitation layer 530 includes a first excitation layer 531 and a second excitation layer 532, and the first excitation layer 531 includes an excitation function for nonlinearizing a one-dimensional array deal with. Optionally, the activation function includes, but is not limited to, a corrected linear unit (Rectified Linear Unit, ReLU) function, an exponential linear unit (ELU) function, and several variants of the ReLU function, such as: linear with leakage correction Unit (Leaky ReLU, LReLU), parametric correction linear unit (Parametric ReLU, PReLU), random correction linear unit (Randomized ReLU, RReLU), etc.

Preferably, in the embodiment of the present application, the excitation function used is the modified linear unit ReLU function. Specifically, the formula of the ReLU function is as follows:

After ReLU processing, the value less than or equal to 0 becomes 0, and the value greater than 0 remains unchanged, which makes the output one-dimensional array sparse. The neural network structure after ReLU achieves sparseness can better mine relevant features and fit Training data.

Optionally, the second excitation layer 532 includes a classification function Sigmoid to classify and discriminate the constants output by the fully connected layer.

Among them, the formula of the Sigmoid function is as follows:

In the Sigmoid function, when the input tends to positive or negative infinity, the function approaches a smooth state. Because the output range of the Sigmoid function is from 0 to 1, this function is often used for the probability of two classifications. The multiple probability values obtained by the Sigmoid function processing are judged to obtain the final face anti-counterfeiting judgment result to determine whether the recognition target is a living face.

It should be understood that, in the embodiment of the present application, the neural network 50 may further include: one or more fully connected layers 520 and/or one or more excitation layers 530. For example: the structure of flattened layer-fully connected layer-excited layer, or flattened layer-fully connected layer-excited layer-fully connected layer-excited layer-fully connected layer-excited layer structure, the embodiment of this application does not have this Make a limit.

It should also be understood that the excitation functions adopted by the multiple excitation layers 530 may be different, and/or the fully connected parameters in the multiple fully connected layers 520 may also be different. The embodiment of the present application does not limit this.

Preferably, in another possible implementation manner, a deep learning algorithm is used to perform comprehensive calculation on the left-eye optimized eye image and the right-eye optimized eye image of the recognition target, and perform classification processing together to It is determined whether the recognition target is a living human face. Using this method, the iris characteristics of the left-eye optimized eye image and the right-eye optimized eye image can be integrated, and the accuracy of anti-counterfeiting judgment can be improved.

Specifically, as shown in FIG. 13, a face anti-counterfeiting discrimination method 600 includes:

S611: Perform image equalization processing on the left eye image by using a histogram equalization method to obtain an optimized left eye image;

S612: Using a histogram equalization method to perform image equalization processing on the right eye image to obtain an optimized right eye image;

S620: Perform classification processing on the optimized left eye image and the optimized right eye image through a neural network to determine whether the recognition target is a living human face.

Optionally, wherein the optimized left eye image and the optimized right eye image have the same size.

Optionally, in this embodiment of the present application, the optimized left eye image and the optimized right eye image are classified by the neural network 60 to determine whether the recognition target is a living human face.

Specifically, as shown in FIG. 14, the neural network 60 is used to comprehensively classify the left-eye optimized eye image and the right-eye optimized eye image of the recognition target. Specifically, the neural network 60 includes a first network 610, The second network 620 and the third network 630. The first network 610 includes: a second flattened layer, at least one second fully connected layer, and at least one second excitation layer; the second network 620 includes: a third flattened layer, at least one third fully connected layer And at least one third incentive layer; the third network 630 includes: at least one fourth fully connected layer and at least one fourth incentive layer.

Preferably, in the embodiment of the present application, the first network 610 includes: a second flattened layer 611, a second upper fully connected layer 612, a second upper excitation layer 613, a second lower fully connected layer 614, and a second The lower excitation layer 615 is used to flatten and fully connect the input left-eye optimized eye image, and output a left-eye one-dimensional feature array, which is also called a left-eye classification feature value. The second network 620 includes: a third flattened layer 621, a third upper fully connected layer 622, a third upper excitation layer 623, a third lower fully connected layer 624, and a third lower excitation layer 625, which are used for the right eye of the input The optimized eye image is flattened and fully connected, and the output is a one-dimensional feature array of the right eye, also known as the right eye classification feature value.

The third network 630 includes: a fourth fully connected layer 631 and a fourth excitation layer 632, which are used to fully connect and classify the left-eye one-dimensional feature array and the right-eye one-dimensional feature array. For example, if the first network 610 outputs a left-eye one-dimensional feature array that includes 10 feature constants, and the second network 620 outputs a right-eye one-dimensional feature array that also includes 10 feature constants, the left-eye one-dimensional feature array and the right-eye one-dimensional feature array are combined A total of 20 feature constants in the feature array are input to the third network 630 together for full connection and classification processing.

It should be understood that the fully connected layer activation functions or classification functions in the first network, the second network, and the third network may be the same or different, which is not limited in the embodiment of the present application.

Preferably, the second upper fully connected layer 612 and the third upper fully connected layer 622 both use ReLU activation functions, and the second lower fully connected layer 613 and the third lower fully connected layer 623 both use the Sigmoid classification function. .

Optionally, the third network 630 may use the ReLU excitation function to perform non-linearization processing on the output characteristic constant again, to correct the classification result, and to improve the accuracy of recognition and judgment.

In the embodiment of the present application, the neural network 30 and the neural network 40 have a simple network structure and a fast running speed, and can be run on an Advanced RISC Machine (ARM).

In the above-mentioned application embodiment, the iris-based face anti-counterfeiting judgment is performed according to the eye image to determine whether the recognition target is a living human face, wherein the result of the face anti-counterfeiting judgment is used for face recognition.

Optionally, the result of the face anti-counterfeiting discrimination can also be used for face registration, that is, to generate a face feature template in the 2D face recognition process. Specifically, face anti-counterfeiting is added in the face registration process to prevent photos collected based on face photos or other non-living face models from being used as templates for face recognition matching, which can improve the accuracy of 2D recognition.

Specifically, as shown in FIG. 15, the face registration method 700 includes:

S710: Acquire an eye image of the recognition target.

S720: Perform iris-based face anti-counterfeiting judgment according to the eye image to determine whether the recognition target is a living human face, wherein the result of the face anti-counterfeiting judgment is used to establish a face feature template.

It should be understood that the process of the face registration method in this embodiment of the application and the process of the aforementioned face recognition method are two independent stages, and only the face feature template established in the process of the registration method is used for the judgment of 2D recognition in the face recognition process . After the face feature template is established through the face registration method, face recognition is performed through the above-mentioned face recognition method and face anti-counterfeiting discrimination method.

It should also be understood that the recognition target in the embodiment of the present application may be the same as or different from the recognition target in the above-mentioned face recognition process. For example, it may be both the user's live face, and the user's live face is registered and recognized; or The recognition target in the registration process is the user's live face, but the recognition target in the recognition process is other non-living faces. The embodiments of this application do not limit this.

Optionally, the step S710 may be the same as the step S210 described above, and the eye image of the recognition target is acquired through the image acquisition device. Optionally, the eye image is an infrared image or a visible light color image.

Optionally, in the step S720, the iris-based face anti-counterfeiting judgment is performed according to the eye image to determine whether the recognition target is a living human face. The aforementioned face recognition anti-counterfeiting judgment method 500 and face recognition Any one of the anti-counterfeiting discrimination method 501 and the face recognition anti-counterfeiting discrimination method 600 performs discrimination. For a specific description, please refer to the above-mentioned application embodiment, which will not be repeated here.

Optionally, in the embodiment of the present application, the face registration method further includes: acquiring a target image of the recognition target, acquiring the eye image based on the target image, and establishing a face feature template based on the target image .

In a possible implementation manner, when the target image is an infrared image, the infrared image of the identified target is acquired first, the template matching is performed based on the infrared image, and anti-counterfeiting is performed on the basis of successful matching.

For example, FIG. 16 shows a face registration method 800, which includes:

S810: Obtain an infrared image of the identified target;

S850: Perform template matching based on the infrared image;

S851: When the template matching is successful, obtain the eye image based on the infrared image;

S852: When the template matching fails, the face feature template is not established;

S860: Perform iris-based face anti-counterfeiting judgment according to the eye image to determine whether the recognition target is a living face;

S871: When the recognition target is a living human face, store an infrared image as a face feature template;

S872: When the recognition target is not a living human face, not storing the infrared image as a facial feature template.

Wherein, optionally, step S810 may be the same as step S310. Step S851 may be the same as step S351. Step S860 may be the same as step S360.

Optionally, step S850 may be similar to step S340 for performing 2D recognition based on the target image. The infrared image is matched with multiple facial feature templates in the facial feature template library. If the matching is successful, the face target image is the user. If the matching fails, the target face image is not the user’s face image.

Optionally, in step S871, when the recognition target is a living human face, the data of the infrared image is stored in the storage unit as a new face feature template in the face feature template library. The storage unit may be the execution person. The storage unit in the processor of the face registration method may also be the memory in the electronic device that executes the face registration method.

Optionally, as shown in FIG. 17, the face registration method 800 may further include:

S820: face detection;

S821: When the human face detects that there is a human face on the infrared image, perform face cropping on the infrared image to obtain a human face image;

S822: When the human face detects that there is no human face in the infrared image, the restart parameter increases by 1;

Optionally, step S820 to step S822 may be the same as step S320 to step S332.

S830: 3D face reconstruction;

Specifically, by emitting structured light or light pulses, after reflecting on the recognition target surface, the reflected structured light or reflected light pulses carrying the information of the recognition target surface are received, so as to obtain the 3D data of the recognition target. The 3D data contains the recognition target. The depth information can indicate the surface shape of the recognition target. The 3D data can be expressed in various forms such as a depth image (Depth Image), a 3D point cloud (Point Cloud), and a geometric model. In the embodiment of the present application, 3D face reconstruction can be performed based on the 3D data, that is, a 3D shape image representing the recognition target can be obtained.

S831: When the 3D face reconstruction is successful, that is, when the 3D shape image of the recognition target is obtained according to the 3D data, go to S840.

Optionally, when the 3D face reconstruction is successful, the 3D data is stored in the storage unit, for example, the 3D point cloud data is stored in the storage unit as a 3D point cloud data template to form a 3D point cloud data template library.

S832: When the 3D face reconstruction fails, that is, when the 3D shape image of the recognition target cannot be obtained according to the 3D data, the restart parameter is increased by 1.

S840: Determine whether the face image cut in step S821 belongs to the face feature template library. Optionally, by acquiring user identification (ID) information of the target image, it is determined whether there is a face feature template library of the user ID, and when there is a face feature template library of the user ID, enter S842: the person The face image belongs to the face feature template library. When there is no face feature template library of the user ID, enter S841: the face image does not belong to the face feature template library.

S8411: When the face image does not belong to the face feature template library, obtain an eye image based on the infrared image, and proceed to step S860.

Optionally, a new user facial feature template library can be established according to the acquired user ID information of the target image.

S8501: When the face image belongs to the face feature template library, perform template matching based on the face image cut in step S821. The specific matching method may be the same as step S850.

S851: When the template matching is successful, obtain an eye image based on the infrared image, and proceed to step S860.

S852: When the template matching fails, the face feature template is not established, and the restart parameter is increased by 1.

S860: Perform iris-based face anti-counterfeiting judgment according to the eye image to determine whether the recognition target is a living face.

S8711: When the recognition target is a living human face, go to S8712: Determine whether it is a valid point cloud.

Optionally, the 3D point cloud data collected by face reconstruction in S830 is matched with multiple 3D point cloud data templates in the 3D point cloud data template library to determine whether it is a valid point cloud. When the matching is successful, it is an invalid point cloud, and when the matching fails, it is a valid point cloud. Specifically, point cloud matching is used to determine whether the face angle of the recognized target in the collected 3D point cloud data is the same as the face angle in the 3D point cloud data template. When the angle is the same, the matching is successful, indicating that there is a template library The 3D point cloud data with the same face angle is an invalid point cloud; when the angles are different, the matching fails, which means that there is no 3D point cloud data with the same face angle in the template library, and it is a valid point cloud.

Optionally, in this process, multiple pieces of 3D point cloud data of the recognized target can be collected, and point cloud splicing and point cloud fusion can be performed to form 3D data and 3D images of the face in all directions and angles, according to the 3D image Can perform 3D face recognition.

S8713: When judging that the 3D point cloud data is a valid point cloud, store the face image as a face feature template. Specifically, the data of the face image is stored in the storage unit as a new face feature template in the face feature template library.

S8714: When it is judged that the 3D point cloud data is invalid, the restart parameter is increased by 1.

Optionally, after determining that the 3D point cloud data is a valid point cloud, it can also be determined whether the face feature templates in the face feature template library are full.

Specifically, it is determined whether the number of face feature templates in the face feature template library is equal to a preset value. If it is equal to the preset value, the face feature template is full, and no new face feature template is stored.

For example, the preset value is 8, when the number of face feature templates in the face feature template library is 8, no more face feature templates are added.

When the face feature template is not full, the face image is stored as the face feature template. Specifically, the data of the face image is stored in the storage unit as a new face feature template in the face feature template library.

Optionally, the face registration method 800 further includes:

Determine whether the restart parameter is less than the second threshold. If the restart parameter is less than the second threshold, enter S810; if the restart parameter is greater than or equal to the second threshold, the recognition fails.

The face recognition method embodiment of the present application is described in detail above with reference to Figs. 2 to 17, and the face recognition device embodiment of the present application is described in detail below with reference to Fig. 18. It should be understood that the device embodiment and the method embodiment are mutually Correspondingly, similar descriptions can refer to the method embodiments.

FIG. 18 is a schematic block diagram of a face recognition device 20 according to an embodiment of the present application, including: a processor 210;

The processor 210 is configured to: obtain a first eye image of a first recognition target;

According to the first eye image, the iris-based face anti-counterfeiting judgment is performed to determine whether the first recognition target is a living human face, wherein the result of the face anti-counterfeiting judgment is used for face recognition.

Optionally, the processor 220 may be a processor of the face recognition device 20, or a processor of an electronic device including the face recognition device 20, which is not limited in the embodiment of the present application.

Optionally, the first eye image is a first eye infrared image.

Optionally, the face recognition device 20 further includes: an image acquisition device 220, configured to obtain a first target image of the first recognition target;

The processor 210 is further configured to: perform two-dimensional recognition based on the first target image; when the two-dimensional recognition is successful, obtain the first eye image based on the first target image;

The processor 210 is further configured to: when the first recognition target is a living face, determine that the face recognition is successful; or, when the first recognition target is a non-living face, determine that the face recognition fails.

Optionally, the processor 210 is specifically configured to: acquire a first target image of the first recognition target, and acquire the first eye image based on the first target image;

The processor 210 is further configured to: when the first recognition target is a living human face, perform two-dimensional recognition based on the first target image;

When the two-dimensional recognition is successful, it is determined that the face recognition is successful, or when the two-dimensional recognition fails, it is determined that the face recognition has failed;

Or, when the first recognition target is a non-living face, it is determined that the face recognition fails.

Optionally, the processor 210 is specifically configured to: acquire a first face image based on the first target image;

The first face image is matched with multiple feature templates, and when the matching is successful, the two-dimensional recognition is successful, or when the matching fails, the two-dimensional recognition fails.

Optionally, the processor 210 is specifically configured to: acquire a face region image based on the first target image; acquire the first eye image based on the face region image.

Optionally, the first eye image is a human eye area image or an iris area image including the iris.

Optionally, the processor 210 is specifically configured to: use a histogram equalization method to process the first eye image to obtain a first optimized eye image;

According to the first optimized eye image, the iris-based face anti-counterfeiting judgment is performed.

Optionally, the processor 210 is specifically configured to: perform classification processing on the first optimized eye image through a neural network to determine whether the first recognition target is a living human face.

Optionally, the first eye image includes a first left eye image and/or a first right eye image, and the processor 210 is specifically configured to:

Processing the first left-eye eye image by using the histogram equalization method to obtain a first optimized left-eye eye image; and/or

Using the histogram equalization method to process the first right eye image to obtain a first optimized right eye image.

Optionally, the first eye image includes: the first left eye image or the first right eye image;

The neural network includes: a first flattened layer, at least one first fully connected layer, and at least one first excitation layer.

The processor 210 is specifically configured to: use the first flattening layer to process the first optimized left-eye eye image or the first optimized right-eye image to obtain multiple eye pixel values;

Using the at least one first fully connected layer to fully connect the multiple eye pixel values to obtain multiple characteristic constants;

Through the at least one first excitation layer, nonlinearization processing or classification processing is performed on the plurality of characteristic constants.

Optionally, the neural network includes: the first flattened layer, two first fully connected layers, and two first excitation layers.

Optionally, the excitation functions in the two first excitation layers are respectively a modified linear unit ReLU function and a Sigmoid function.

Optionally, the first eye image includes: the first left eye image and the first right eye image;

The neural network includes a first network, a second network, and a third network;

The first network includes: a second flattening layer, at least one second fully connected layer, and at least one second excitation layer;

The second network includes: a third flattening layer, at least one third fully connected layer, and at least one third excitation layer;

The third network includes: at least one fourth fully connected layer and at least one fourth excitation layer.

The processor 210 is specifically configured to: process the first optimized left-eye eye image through the first network to obtain a left-eye classification feature value;

Processing the first optimized right eye image through the second network to obtain right eye classification feature values;

Fully connect the left-eye classification feature value and the right-eye classification feature value through the third network.

Optionally, the first network includes: the second flattened layer, two second fully connected layers, and two second excitation layers;

The second network includes: the third flattening layer, two third fully connected layers, and two third excitation layers;

The third network includes: a fourth fully connected layer and a fourth excitation layer.

Optionally, the excitation functions in the two second excitation layers are respectively a modified linear unit ReLU function and a Sigmoid function; and/or,

The excitation functions in the two third excitation layers are the modified linear unit ReLU function and the Sigmoid function respectively; and/or,

One of the excitation functions in the fourth excitation layer is a modified linear unit ReLU function.

Optionally, the processor 210 is further configured to: obtain a second eye image of the second recognition target;

According to the second eye image, the iris-based face anti-counterfeiting judgment is performed to determine whether the second recognition target is a living human face, wherein the result of the face anti-counterfeiting judgment is used to establish a face feature template.

Optionally, the second eye image is a second eye infrared image.

Optionally, the processor 210 is further configured to: acquire a second target image of the second recognition target, acquire the second eye image based on the second target image, and based on the second target image Establish the face feature template.

Optionally, the processor 210 is further configured to: perform face detection based on the second target image;

Wherein, the establishment of a facial feature template based on the second target image includes:

When the face detection is successful, a second face image is acquired based on the second target image, and the face feature template is established based on the second face image.

Optionally, the processor 210 is specifically configured to: determine whether the second face image belongs to a face feature template library;

When the second face image belongs to the face feature template library, the second face image is matched with multiple face feature templates in the face feature template library.

When the second face image does not belong to the face feature template library, the iris-based face anti-counterfeiting judgment is performed according to the second eye image, and when it is determined that the second recognition target is a living face, The second face image is established as a face feature template.

Optionally, the processor 210 is specifically configured to: when the matching is successful, perform an iris-based face anti-counterfeiting judgment according to the second eye image;

When it is determined that the second recognition target is a living face, the second face image is established as a face feature template.

Optionally, the processor 210 is specifically configured to obtain 3D point cloud data of the second recognition target when the matching is successful;

When the 3D point cloud data is a valid point cloud, iris-based face anti-counterfeiting judgment is performed according to the second eye image.

Optionally, the processor 210 is specifically configured to: acquire a face region image based on the second target image;

Acquiring the second eye image based on the face region image.

Optionally, the second eye image is a human eye area image or an iris area image including the iris.

Optionally, the processor 210 is specifically configured to: use a histogram equalization method to process the second eye image to obtain a second optimized eye image;

According to the second optimized eye image, the iris-based face anti-counterfeiting judgment is performed.

Optionally, the processor 210 is specifically configured to: perform classification processing on the second optimized eye image through a neural network to determine whether the second recognition target is a living human face.

Optionally, the second eye image includes a second left eye image and/or a second right eye image, and the processor 210 is specifically configured to: compare the second left eye image to the second left eye through a neural network. Performing classification processing on the second image and/or the second right eye image.

Optionally, the neural network includes: at least one flattened layer, at least one fully connected layer, and at least one excitation layer.

As shown in FIG. 19, an embodiment of the present application further provides an electronic device 2, which may include the face recognition apparatus 20 of the foregoing application embodiment.

For example, the electronic device 2 is a smart door lock, a mobile phone, a computer, an access control system and other devices that need to apply face recognition. The face recognition device 20 includes software and hardware devices for face recognition in the electronic device 2.

It should be understood that the processor of the embodiment of the present application may be an integrated circuit chip with signal processing capability. In the implementation process, the steps of the foregoing method embodiments can be completed by hardware integrated logic circuits in the processor or instructions in the form of software. The aforementioned processor may be a general-purpose processor, a digital signal processor (Digital Signal Processor, DSP), an application specific integrated circuit (ASIC), a ready-made programmable gate array (Field Programmable Gate Array, FPGA) or other Programming logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps, and logical block diagrams disclosed in the embodiments of the present application can be implemented or executed. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor or the like. The steps of the method disclosed in the embodiments of the present application may be directly embodied as being executed and completed by a hardware decoding processor, or executed and completed by a combination of hardware and software modules in the decoding processor. The software module can be located in a mature storage medium in the field such as random access memory, flash memory, read-only memory, programmable read-only memory, or electrically erasable programmable memory, registers. The storage medium is located in the memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware.

It can be understood that the face recognition in the embodiments of the present application may further include a memory, and the memory may be a volatile memory or a nonvolatile memory, or may include both volatile and nonvolatile memory. Among them, the non-volatile memory can be read-only memory (Read-Only Memory, ROM), programmable read-only memory (Programmable ROM, PROM), erasable programmable read-only memory (Erasable PROM, EPROM), and electrically available Erase programmable read-only memory (Electrically EPROM, EEPROM) or flash memory. The volatile memory may be a random access memory (Random Access Memory, RAM), which is used as an external cache. By way of exemplary but not restrictive description, many forms of RAM are available, such as static random access memory (Static RAM, SRAM), dynamic random access memory (Dynamic RAM, DRAM), synchronous dynamic random access memory (Synchronous DRAM, SDRAM), double data rate synchronous dynamic random access memory (Double Data Rate SDRAM, DDR SDRAM), enhanced synchronous dynamic random access memory (Enhanced SDRAM, ESDRAM), synchronous connection dynamic random access memory (Synchlink DRAM, SLDRAM) ) And Direct Rambus RAM (DR RAM). It should be noted that the memories of the systems and methods described herein are intended to include, but are not limited to, these and any other suitable types of memories.

The embodiment of the present application also proposes a computer-readable storage medium that stores one or more programs, and the one or more programs include instructions. When the instructions are included in a portable electronic device that includes multiple application programs When executed, the portable electronic device can be made to execute the method of the embodiment shown in Figs. 1-17.

The embodiment of the present application also proposes a computer program, the computer program includes instructions, when the computer program is executed by the computer, the computer can execute the method of the embodiment shown in FIG. 1-17.

An embodiment of the present application also provides a chip that includes an input and output interface, at least one processor, at least one memory, and a bus. The at least one memory is used to store instructions, and the at least one processor is used to call the at least one memory. To execute the method of the embodiment shown in Figure 1-17.

A person of ordinary skill in the art may be aware that the units and algorithm steps of the examples described in combination with the embodiments disclosed herein can be implemented by electronic hardware or a combination of computer software and electronic hardware. Whether these functions are executed by hardware or software depends on the specific application and design constraint conditions of the technical solution. Professionals and technicians can use different methods for each specific application to implement the described functions, but such implementation should not be considered beyond the scope of this application.

Those skilled in the art can clearly understand that, for the convenience and conciseness of description, the specific working process of the above-described system, device, and unit can refer to the corresponding process in the foregoing method embodiment, which will not be repeated here.

In the several embodiments provided in this application, it should be understood that the disclosed system, device, and method may be implemented in other ways. For example, the device embodiments described above are only illustrative. For example, the division of the units is only a logical function division, and there may be other divisions in actual implementation, for example, multiple units or components can be combined or It can be integrated into another system, or some features can be ignored or not implemented. In addition, the displayed or discussed mutual coupling or direct coupling or communication connection may be indirect coupling or communication connection through some interfaces, devices or units, and may be in electrical, mechanical or other forms.

The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the objectives of the solutions of the embodiments.

In addition, the functional units in each embodiment of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units may be integrated into one unit.

If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer readable storage medium. Based on this understanding, the technical solution of the present application essentially or the part that contributes to the prior art or the part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium , Including several instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage media include: U disk, mobile hard disk, read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), magnetic disk or optical disk and other media that can store program code .

The above are only specific implementations of this application, but the protection scope of this application is not limited to this. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed in this application. Should be covered within the scope of protection of this application. Therefore, the protection scope of this application shall be subject to the protection scope of the claims.

Claims (31)

A method for face recognition, characterized in that it includes: Acquiring a first target image and a first eye image of the first recognition target; Perform iris-based face anti-counterfeiting judgment according to the first eye image to determine whether the first recognition target is a living human face and output a living judgment result; Performing feature template matching according to the first target image, and outputting a matching result; The face recognition result is output according to the living body judgment result and the matching result. The method according to claim 1, wherein said outputting a face recognition result according to said living body judgment result and said matching result comprises: When the matching result is successful, output the face recognition result according to the living body judgment result; or, when the living body judgment result is a living body, output the face recognition result according to the matching result; or, in the matching When the result is failure or the living body judgment result is non-living body, the face recognition result is output. The method according to claim 1 or 2, wherein the performing feature template matching according to the first target image and outputting the matching result comprises: Performing face detection based on the first target image; When the face detection is successful, acquiring a first face image based on the first target image; Matching the first face image with a plurality of pre-stored first feature templates; When the first face image is successfully matched with any one of the plurality of first feature templates, the output matching result is successful; or, When the first face image fails to match the multiple first feature templates, output a matching result as failure; Or, when the face detection fails, the output matching result is failed. The method according to any one of claims 1 to 3, wherein the acquiring the first target image and the first eye image of the first recognition target comprises: A first target image of the first recognition target is acquired, and the first eye image is acquired based on the first target image. The method according to any one of claims 1 to 4, wherein the first eye image is a two-dimensional infrared image. The method according to any one of claims 1 to 5, wherein the first eye image is a human eye area image or an iris area image including an iris. The method according to claim 6, wherein said performing an iris-based face anti-counterfeiting judgment according to the first eye image comprises: Processing the first eye image by using a histogram equalization method to obtain a first optimized eye image; According to the first optimized eye image, the iris-based face anti-counterfeiting judgment is performed. 8. The method according to claim 7, wherein said performing an iris-based face anti-counterfeiting judgment according to the first optimized eye image comprises: Perform classification processing on the first optimized eye image through a neural network to determine whether the first recognition target is a living human face. The method according to claim 7 or 8, wherein the first eye image includes a first left eye image and/or a first right eye image, and the histogram equalization method is used to The processing of the first eye image to obtain the first optimized eye image includes: Processing the first left-eye eye image by using the histogram equalization method to obtain a first optimized left-eye eye image; and/or Using the histogram equalization method to process the first right eye image to obtain a first optimized right eye image. The method according to claim 9, wherein the first eye image comprises: the first left eye image or the first right eye image; The neural network includes: a first flattened layer, at least one first fully connected layer, and at least one first excitation layer. The method according to claim 10, wherein said classifying said first optimized eye image through a neural network comprises: Processing the first optimized left eye image or the first optimized right eye image through the first flattening layer to obtain multiple eye pixel values; Using the at least one first fully connected layer to fully connect the multiple eye pixel values to obtain multiple characteristic constants; Through the at least one first excitation layer, nonlinearization processing or classification processing is performed on the plurality of characteristic constants. The method according to claim 10 or 11, wherein the neural network comprises: the first flattened layer, two first fully connected layers, and two first excitation layers. The method according to claim 12, wherein the excitation functions in the two first excitation layers are respectively a modified linear unit ReLU function and a Sigmoid function. The method according to claim 9, wherein the first eye image comprises: the first left eye image and the first right eye image; The neural network includes a first network, a second network, and a third network; The first network includes: a second flattening layer, at least one second fully connected layer, and at least one second excitation layer; The second network includes: a third flattening layer, at least one third fully connected layer, and at least one third excitation layer; The third network includes: at least one fourth fully connected layer and at least one fourth excitation layer. The method according to claim 14, wherein the classification processing of the first optimized eye image through a neural network comprises: Processing the first optimized left-eye eye image through the first network to obtain a left-eye classification feature value; Processing the first optimized right eye image through the second network to obtain right eye classification feature values; Fully connect the left-eye classification feature value and the right-eye classification feature value through the third network. The method according to claim 14 or 15, wherein the first network comprises: the second flattening layer, two second fully connected layers, and two second excitation layers; The second network includes: the third flattening layer, two third fully connected layers, and two third excitation layers; The third network includes: a fourth fully connected layer and a fourth excitation layer. The method according to claim 16, wherein the excitation functions in the two second excitation layers are respectively a modified linear unit ReLU function and a Sigmoid function; and/or, The excitation functions in the two third excitation layers are the modified linear unit ReLU function and the Sigmoid function respectively; and/or, One of the excitation functions in the fourth excitation layer is a modified linear unit ReLU function. The method according to any one of claims 1-17, wherein the method further comprises: Acquiring a second eye image of the second recognition target; According to the second eye image, perform iris-based face anti-counterfeiting judgment to determine whether the second recognition target is a living human face, wherein the result of the face anti-counterfeiting judgment is used to establish a face feature template. The method of claim 18, wherein the second eye image is a two-dimensional infrared image. The method according to claim 18 or 19, wherein the method further comprises: A second target image of the second recognition target is acquired, the second eye image is acquired based on the second target image, and the face feature template is established based on the second target image. The method of claim 20, wherein the method further comprises: Performing face detection based on the second target image; Wherein, the establishment of a facial feature template based on the second target image includes: When the face detection is successful, a second face image is acquired based on the second target image, and the face feature template is established based on the second face image. 22. The method of claim 21, wherein the establishing the face feature template based on the second face image comprises: Judging whether the second face image belongs to a face feature template library; When the second face image belongs to the face feature template library, matching the second face image with multiple face feature templates in the face feature template library; When the second face image does not belong to the face feature template library, the iris-based face anti-counterfeiting judgment is performed according to the second eye image, and when it is determined that the second recognition target is a living face, The second face image is established as a face feature template. The method according to claim 22, wherein the matching the second face image with multiple face feature templates in the face feature template library comprises: When the matching is successful, perform iris-based face anti-counterfeiting judgment according to the second eye image; When it is determined that the second recognition target is a living face, the second face image is established as a face feature template. 22. The method according to claim 23, wherein, when the matching is successful, performing an iris-based face anti-counterfeiting judgment according to the second eye image comprises: When the matching is successful, obtain the 3D point cloud data of the second recognition target; When the 3D point cloud data is a valid point cloud, iris-based face anti-counterfeiting judgment is performed according to the second eye image. The method according to any one of claims 18-24, wherein the second eye image is a human eye area image or an iris area image including the iris. The method according to claim 25, wherein said performing an iris-based face anti-counterfeiting judgment according to the second eye image comprises: Using a histogram equalization method to process the second eye image to obtain a second optimized eye image; According to the second optimized eye image, an iris-based face anti-counterfeiting judgment is performed. The method according to claim 26, wherein said performing an iris-based anti-counterfeiting judgment on the face according to the second optimized eye image comprises: Perform classification processing on the second optimized eye image through a neural network to determine whether the second recognition target is a living human face. 28. The method according to claim 27, wherein the second eye image comprises a second left eye image and/or a second right eye image, and the second optimization is performed on the second eye image through a neural network. The classification of eye images includes: Perform classification processing on the second left eye image and/or the second right eye image through a neural network. The method according to claim 27 or 28, wherein the neural network comprises: At least one flattening layer, at least one fully connected layer and at least one excitation layer. A face recognition device, which is characterized by comprising: a processor; The processor is configured to execute: the method for face recognition according to any one of claims 1-29. An electronic device, characterized in that it comprises: The device for face recognition according to claim 30.

Images