CN112070022A - Face image recognition method and device, electronic equipment and computer readable medium - Google Patents

Abstract

The embodiment of the disclosure discloses a face image recognition method, a face image recognition device, electronic equipment and a computer readable medium. One embodiment of the method comprises: detecting the definition of a face image to be processed; responding to the definition smaller than a set definition threshold value, and generating a face simulation image based on the face image to be processed, wherein the definition of the face simulation image is larger than the definition threshold value; and identifying the face simulation image to obtain face identification information corresponding to the face image to be processed. The embodiment can generate the human face simulation image with higher definition than the human face image to be processed, and improves the identification accuracy of the human face image to be processed.

Description

Face image recognition method, apparatus, electronic device and computer readable medium

technical field

Embodiments of the present disclosure relate to the technical field of face recognition, and in particular, to a face image recognition method, apparatus, electronic device, and computer-readable medium.

Background technique

Face recognition is a kind of biometric identification technology based on human facial feature information. A series of related technologies that use cameras or cameras to collect images or video streams containing faces, and automatically detect and track faces in the images, and then perform face recognition on the detected faces. .

In practice, due to factors such as shooting distance, lens pixels, and ambient light brightness, the clarity of face images may not be high. Correspondingly, the accuracy of face recognition is not high.

SUMMARY OF THE INVENTION

This summary of the disclosure serves to introduce concepts in a simplified form that are described in detail in the detailed description that follows. The content section of this disclosure is not intended to identify key features or essential features of the claimed technical solution, nor is it intended to be used to limit the scope of the claimed technical solution.

Some embodiments of the present disclosure propose a face image recognition method, apparatus, electronic device, and computer-readable medium to solve the technical problems mentioned in the above background art section.

In a first aspect, some embodiments of the present disclosure provide a face image recognition method, the method includes: detecting the sharpness of the face image to be processed; in response to the above-mentioned sharpness being less than a set sharpness threshold, based on the above-mentioned Process the face image to generate a face simulation image, the definition of the face simulation image is greater than the definition threshold; identify the face simulation image to obtain the face recognition information corresponding to the to-be-processed face image.

In a second aspect, some embodiments of the present disclosure provide a face image recognition apparatus, the apparatus includes: an image sharpness detection unit configured to detect the sharpness of a to-be-processed face image; a face simulation image generation unit is configured to generate a simulated face image based on the above-mentioned face image to be processed in response to the above-mentioned sharpness being less than the set sharpness threshold, and the sharpness of the above-mentioned simulated face image is greater than the above-mentioned sharpness threshold; the face recognition unit is configured to The above-mentioned face simulation images are identified in pairs to obtain face recognition information corresponding to the above-mentioned face images to be processed.

In a third aspect, some embodiments of the present disclosure provide an electronic device, comprising: one or more processors; and a memory on which one or more programs are stored, when the one or more programs are stored by the one or more programs described above When executed by each processor, the above-mentioned one or more processors are caused to execute the above-mentioned face image recognition method of the first aspect.

In a fourth aspect, some embodiments of the present disclosure provide a computer-readable medium on which a computer program is stored, characterized in that, when the program is executed by a processor, the face image recognition method of the first aspect above is implemented.

One of the above-mentioned embodiments of the present disclosure has the following beneficial effects: firstly, the sharpness of the face image to be processed is detected, so as to realize the preprocessing of the image quality. Then, when the sharpness is less than the set sharpness threshold, a simulated face image corresponding to the face image to be processed is generated. The definition of the face simulation image is greater than the definition threshold, and the face simulation image retains the face features of the face image to be processed. In this way, it is beneficial to improve the accuracy of face image recognition. Finally, the face simulation image is recognized, and the face recognition information is obtained. Thus, the recognition accuracy of the face image to be processed is improved.

Description of drawings

The above and other features, advantages and aspects of various embodiments of the present disclosure will become more apparent when taken in conjunction with the accompanying drawings and with reference to the following detailed description. Throughout the drawings, the same or similar reference numbers refer to the same or similar elements. It should be understood that the drawings are schematic and that elements and elements are not necessarily drawn to scale.

1 is a schematic diagram of an application scenario of a face image recognition method according to some embodiments of the present disclosure;

2 is a flowchart of some embodiments of a face image recognition method according to the present disclosure;

3 is a flow chart of other embodiments of the face image recognition method according to the present disclosure;

4 is a flow chart of other embodiments of a face image recognition method according to the present disclosure;

5 is a schematic structural diagram of some embodiments of a face image recognition apparatus according to the present disclosure;

6 is a schematic structural diagram of an electronic device suitable for implementing some embodiments of the present disclosure.

Detailed ways

Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While certain embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided for a thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are only for exemplary purposes, and are not intended to limit the protection scope of the present disclosure.

In addition, it should be noted that, for the convenience of description, only the parts related to the related invention are shown in the drawings. The embodiments of this disclosure and features of the embodiments may be combined with each other without conflict.

It should be noted that concepts such as "first" and "second" mentioned in the present disclosure are only used to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or interdependence.

It should be noted that the modifications of "a" and "a plurality" mentioned in the present disclosure are illustrative rather than restrictive, and those skilled in the art should understand that unless the context clearly indicates otherwise, they should be understood as "one or a plurality of". multiple".

The names of messages or information exchanged between multiple devices in the embodiments of the present disclosure are only for illustrative purposes, and are not intended to limit the scope of these messages or information.

The present disclosure will be described in detail below with reference to the accompanying drawings and in conjunction with embodiments.

FIG. 1 is a schematic diagram of an application scenario of a face image recognition method according to some embodiments of the present disclosure.

As shown in FIG. 1 , after the electronic device 100 receives the face image 101 to be processed, in order to ensure the accuracy and validity of the face recognition, the clarity 102 of the face image 101 to be processed may be detected first. When the sharpness 102 is greater than or equal to the set sharpness threshold, it means that accurate face recognition can be performed on the face image 101 to be processed. When the sharpness 102 is smaller than the set sharpness threshold, it means that when face recognition is performed directly on the face image 101 to be processed, accurate and effective face recognition information cannot be obtained. At this time, the electronic device 100 may generate a face simulation image 103 based on the face image 101 to be processed. Among them, the face simulation image 103 is an image with the same image content as the face image 101 to be processed, but the definition is higher than the set definition threshold. On this basis, the electronic device 100 directly performs face recognition on the face simulation image 103, and the obtained face recognition information has higher accuracy and validity.

It should be understood that the number of electronic devices 100 in FIG. 1 is merely illustrative. There may be any number of electronic devices 100 according to implementation needs.

Continuing to refer to FIG. 2 , FIG. 2 illustrates a process 200 of some embodiments of a face image recognition method according to the present disclosure. The face image recognition method includes the following steps:

Step 201, detecting the sharpness of the face image to be processed.

In some embodiments, the execution body of the face image recognition method (eg, the electronic device 100 shown in FIG. 1 ) may receive the face image to be processed through a wired connection or a wireless connection. It should be pointed out that the above wireless connection methods may include but are not limited to 3G/4G connection, WiFi connection, Bluetooth connection, WiMAX connection, Zigbee connection, UWB (ultra wideband) connection, and other wireless connection methods currently known or developed in the future .

The execution subject can detect the sharpness of the face image to be processed. For example, the execution subject can convert the face image to be processed into a grayscale image, and perform noise reduction processing on the grayscale image to obtain a noise reduction image. Then, get the edge points within the denoised image. Among them, the edge points can represent various lines related to the face image in the denoised image. The blurred image can be obtained after the execution subject performs low-pass filtering on the denoised image, and then calculates the sharpness feature quantity of each edge point in the denoised image and the blurred image, and then calculates the sharpness value of each edge point. Finally, the average of the sharpness values of all edge points is taken as the sharpness of the face image to be processed. The execution subject can also detect the contrast of pixels in the face image to be processed. When the contrast is less than the set threshold, it can be considered that the pixel difference in the face image to be processed is too small, and it is difficult to perform face recognition on the face image to be processed. Conversely, when the contrast is greater than or equal to the set threshold, face recognition can be performed on the face image to be processed.

Step 202, in response to the above-mentioned definition being less than the set definition threshold, generate a face simulation image based on the above-mentioned to-be-processed face image.

In some embodiments, when the sharpness is less than the set sharpness threshold, the execution subject can generate a face simulation image corresponding to the face image to be processed through various methods. For example, the execution subject may perform image processing on the face image to be processed, increase the brightness of the face image to be processed, increase the contrast, etc., to obtain the face simulation image. Wherein, the above-mentioned face simulation image may be an image with the same image content as the to-be-processed face image, and the definition is greater than the above-mentioned definition threshold.

Step 203: Recognize the above-mentioned face simulation image to obtain face recognition information corresponding to the above-mentioned face image to be processed.

After obtaining the face simulation image, the execution subject may perform face recognition on the face simulation image to obtain face recognition information corresponding to the face image to be processed. In this way, the recognition accuracy and effectiveness of the face image to be processed can be improved.

In the face image recognition method disclosed by some embodiments of the present disclosure, firstly, the sharpness of the face image to be processed is detected, so as to realize the preprocessing of the image quality. Then, when the sharpness is less than the set sharpness threshold, a simulated face image corresponding to the face image to be processed is generated. The definition of the face simulation image is greater than the definition threshold, and the face simulation image retains the face features of the face image to be processed. In this way, it is beneficial to improve the accuracy of face image recognition. Finally, the face simulation image is recognized, and the face recognition information is obtained. Thus, the recognition accuracy of the face image to be processed is improved.

Continuing to refer to FIG. 3 , FIG. 3 illustrates a process 300 of some embodiments of a face image recognition method according to the present disclosure. The face image recognition method includes the following steps:

Step 301, detecting the sharpness of the face image to be processed.

The content of step 301 is the same as that of step 201, and details are not repeated here.

Step 302: Import the above-mentioned face image to be processed into a face image generation model to obtain the above-mentioned face simulation image.

The execution subject can import the face image to be processed into the face image generation model to obtain the above-mentioned face simulation image. Among them, the face image generation model may be a deep learning model, a genetic algorithm model and other models, which will not be repeated here.

In some optional implementations of some embodiments, the above-mentioned face image generation model is obtained through the following steps:

In the first step, a plurality of sample face input images and a sample face target image corresponding to each sample face input image in the above-mentioned plurality of sample face input images are obtained.

The execution body can acquire multiple sample face input images and multiple sample face target images. Wherein, the clarity of the sample face input image is less than or equal to the clarity threshold, the sample face target image has the same image content as the sample face input image, and the clarity is greater than the clarity threshold. The sample face target images can be manually annotated images containing multiple face key points. The sample face input image and the corresponding sample face target image may be face images under various angles and distances.

Optionally, the execution subject may also first obtain a clear sample face target image, and then process the sample face target image to obtain a sample face input image whose clarity is less than the clarity threshold. The execution subject may also obtain the sample face input image and the sample face target image in other ways, which will not be repeated here.

In the second step, a pre-built generative adversarial network is extracted.

In this embodiment, the above-mentioned execution subject may extract a pre-established Generative Adversarial Nets (GAN). The generative adversarial network includes a generative network and a discriminant network, the generative network is used to generate a face target image by using the face input image, and the discriminant network is used to determine whether the image input to the above-mentioned discriminant network is an image output by the generative network.

It should be noted that the above-mentioned generation network may be a convolutional neural network for image processing (eg, various convolutional neural network structures including convolutional layers, pooling layers, de-pooling layers, and deconvolutional layers). The above-mentioned discriminant network may be a convolutional neural network (for example, various convolutional neural network structures including a fully-connected layer, wherein the above-mentioned fully-connected layer can implement a classification function). In addition, the above-mentioned discriminant network may also be other model structures for implementing the classification function, such as a support vector machine (Support Vector Machine, SVM). Exemplarily, if the discriminant network determines that the input image is an image output by the above-mentioned generation network (from the generation data), it can output 1; if it is determined that the input image is not an image output by the above-mentioned generation network, it can output 0. It should be noted that the discriminant network can also output other values, such as a value between 0 and 1, which represents the probability that the image input to the discriminant network comes from real data.

In the third step, using the machine learning method, each sample face input image in the above-mentioned multiple sample face input images is used as the input of the generation network, and the image output by the generation network corresponds to the sample face input image input into the generation network. The sample face target image is used as the input of the discriminant network, the generation network and the discriminant network are trained, and the trained generation network is determined as the face image generation model.

In this embodiment, based on the above-mentioned generative adversarial network, the execution subject can use the machine learning method to use each sample face input image in the multiple sample face input images as the input of the generation network, and use the output of the generation network. The image and the sample face input image of the input generation network The sample face target image corresponding to the input image is used as the input of the discriminant network, and the generation network and the discriminant network are trained. Then, the trained generative network is determined as a face image generative model.

In some optional implementations of some embodiments, the above-mentioned training of the generation network and the discrimination network is performed, and the trained generation network is determined as a face image generation model, and the following training steps can be performed:

The first step is to fix the parameters of the above-mentioned generation network, use each sample face input image in the above-mentioned multiple sample face input images as the input of the generation network, and input the image output by the generation network and the sample face input of the input generation network. The sample face target image corresponding to the image is used as the input of the discriminant network, and the discriminant network is trained by the machine learning method. Machine learning methods can be supervised learning, unsupervised learning, semi-supervised learning, reinforcement learning, etc.

In the second step, the parameters of the discriminant network after training are fixed, and each sample face input image in the above-mentioned multiple sample face input images is used as the input of the generation network, and the generation network is trained by the machine learning method. The executive body can also combine machine learning methods with back-propagation algorithms, gradient descent algorithms, etc. to train the generative network.

The third step is to determine the accuracy of the discrimination result output by the trained discriminant network, and in response to determining that the accuracy is greater than an accuracy threshold (for example, it can be 80%), determine the most recently trained generation network as a face image generation model.

In some embodiments, in response to determining that the accuracy is less than or equal to the above-mentioned accuracy threshold, the above-mentioned training steps are re-performed using the most recently trained generative network and discriminant network. Therefore, the parameters of the face image generation model trained by the generative adversarial network can not only be obtained based on the training samples, but also determined based on the back-propagation of the discriminant network, and the generation model can be realized without relying on a large number of labeled samples. After training, a face image generation model is obtained, which reduces labor costs and improves the accuracy and effectiveness of generating face simulation images.

In some optional implementation manners of some embodiments, the above-mentioned sample face target image may be obtained through the following steps:

The first step is to obtain the face features of the above-mentioned sample face input image.

In order to obtain an accurate and effective sample face target image, the execution subject may first acquire the face features of the above-mentioned sample face input image. The facial features may be round faces, long faces, square faces, and the like.

In the second step, the key points of face prediction are determined based on the above-mentioned face features.

Each facial organ on a human face (for example, eyes, eyebrows, nose, mouth, etc.) has a relatively fixed position. The executive body can determine the face prediction key points according to the face features. In this way, the corresponding relationship established between the face prediction key points and the face features is beneficial to make the subsequent sample face target images and the sample face input images have the same face features.

The third step is to construct a sample face target image through the above face prediction key points.

After obtaining the face prediction key points, the execution subject can construct a sample face target image based on the face prediction key points. That is, the sample face target image may be a fake face image constructed by the execution subject and including the face prediction key points of the sample face input image.

In some optional implementations of some embodiments, the above-mentioned determination of face prediction key points based on the above-mentioned face features may include the following steps:

The first step is to determine the position information of the face prediction key point and the initial image of the face prediction key point based on the above-mentioned face structure information.

In practice, faces can face in all directions, and faces vary. Therefore, the above-mentioned face features may include face spatial inclination and face structure information. The executing subject may determine the position information of the face prediction key points based on the face structure information. Then, the executive body may further determine the initial image of the face prediction key point according to the position information of the face prediction key point. The initial image of the face prediction key point may be an image of the face organ corresponding to the face prediction key point.

The second step is to adjust the position information of the face prediction key point and the initial image of the face prediction key point based on the face space inclination to determine the face prediction key point.

Different facial parts can present different images under different spatial inclinations. The execution subject may adjust the above-mentioned position information of the face prediction key point and the initial image of the face prediction key point according to the inclination of the face space, and then determine the face prediction key point. In this way, the accuracy and effectiveness of face prediction key points are greatly improved.

In some optional implementations of some embodiments, the above-mentioned construction of a sample face target image by using the above-mentioned face prediction key points may include the following steps:

The first step is to construct an initial face image based on the above face prediction key points.

After obtaining the face prediction key points, the execution subject can refer to the above-mentioned face structure information to construct an initial face image. At this time, the initial face image only contains face prediction key points and face structure information, and is not affected by factors such as ambient light. Therefore, the definition of the initial face image can be high.

In the second step, the above-mentioned initial face image is rendered to obtain a sample face target image.

Finally, the execution subject may render the above-mentioned initial face image to obtain a sample face target image. At this time, the obtained sample face target image may be an image that includes the face key points of the sample face input image and has a high definition (higher than the above-mentioned definition threshold).

Step 303: Recognize the above-mentioned face simulation image to obtain face recognition information corresponding to the above-mentioned face image to be processed.

The content of step 303 is the same as that of step 203, and details are not repeated here.

Referring further to FIG. 4 , it shows a schematic diagram of another application scenario of the face image recognition method.

In FIG. 4 , the electronic device first receives the face image 401 to be processed, and in the case that the clarity of the face image 401 to be processed is less than the set clarity threshold, it determines how many faces according to the face features in the face image 401 to be processed. Face prediction key points 402 . After that, a face simulation image 403 is generated according to the face prediction key points 402 . Then, the face simulation image 403 may be a fake face image that includes the face features in the face image 401 to be processed, and whose clarity is greater than the clarity threshold. After that, the electronic device performs face recognition on the face simulation image 403 , and the obtained face recognition information 404 is the face recognition information of the face image 401 to be processed.

Further referring to FIG. 5 , as an implementation of the methods shown in the above figures, the present disclosure provides some embodiments of a face image recognition apparatus, these apparatus embodiments correspond to those method embodiments shown in FIG. 2 , the The device can be specifically applied to various electronic devices.

As shown in FIG. 5 , a face image recognition apparatus 500 in some embodiments includes: an image clarity detection unit 501 , a face simulation image generation unit 502 , and a face recognition unit 503 . The image sharpness detection unit 501 is configured to detect the sharpness of the face image to be processed; the face simulation image generation unit 502, in response to the above-mentioned sharpness being less than the set sharpness threshold, is configured to be based on the above-mentioned to-be-processed image The face image generates a face simulation image, and the clarity of the above-mentioned face simulation image is greater than the above-mentioned sharpness threshold; the face recognition unit 503 is configured to identify the above-mentioned face simulation image, and obtains corresponding to the above-mentioned face image to be processed. face recognition information.

In an optional implementation manner of some embodiments, the above-mentioned face simulation image generation unit 502 may include: a face simulation image generation subunit (not shown in the figure), configured to import the above-mentioned face image to be processed into a human face An image generation model is used to obtain the above-mentioned face simulation image.

In an optional implementation manner of some embodiments, the above-mentioned face simulation image generation subunit may include a face image generation model module (not shown in the figure), which is configured to train a face image generation model, and the above-mentioned face image generation The model module may include: a sample acquisition sub-module (not shown in the figure), a network extraction sub-module (not shown in the figure), and a face image generation model training sub-module (not shown in the figure). The sample acquisition sub-module is configured to acquire a plurality of sample face input images and a sample face target image corresponding to each sample face input image in the above-mentioned plurality of sample face input images, wherein the above-mentioned sample face The clarity of the face input image is less than or equal to the above-mentioned clarity threshold, the above-mentioned sample face target image is the same as the image content of the above-mentioned sample face input image, and the image with the clarity greater than the above-mentioned clarity threshold; the network extraction sub-module is configured to Extract a pre-established generative adversarial network, wherein the above-mentioned generative adversarial network includes a generative network and a discriminant network, the generative network is used to use the face input image to generate a face target image, and the discriminant network is used to determine whether the image input to the above-mentioned discrimination network is not is the image output by the generation network; the face image generation model training sub-module is configured to use the machine learning method to use each sample face input image in the above-mentioned multiple sample face input images as the input of the generation network, and generate The image output by the network and the sample face target image corresponding to the sample face input image input to the generation network are used as the input of the discriminant network, the generation network and the discriminant network are trained, and the trained generation network is determined as the face image generation model.

In an optional implementation manner of some embodiments, the above-mentioned face image generation model training sub-module may include: a face image generation model training module (not shown in the figure), configured to fix the parameters of the above-mentioned generation network, Each sample face input image in the above-mentioned multiple sample face input images is used as the input of the generation network, and the sample face target image corresponding to the image output by the generation network and the sample face input image input to the generation network is used as the target image of the discriminant network. Input, use the machine learning method to train the discriminant network; fix the parameters of the discriminant network after training, take each sample face input image in the above multiple sample face input images as the input of the generation network, and use the machine learning method. The generative network is trained; the accuracy of the discrimination result output by the discriminant network after training is determined, and in response to determining that the accuracy is greater than the accuracy threshold, the most recently trained generative network is determined as the face image generation model.

In an optional implementation manner of some embodiments, the above-mentioned face image generation model training sub-module further includes: an adjustment module (not shown in the figure), configured to use an adjustment module (not shown in the figure), configured to use The most recently trained generative and discriminative networks are returned to the above face image generative model training module.

In an optional implementation manner of some embodiments, the above-mentioned sample acquisition sub-module includes a sample face target image construction module (not shown in the figure), and the above-mentioned sample face target image construction module may include: a face feature acquisition sub-module A module (not shown in the figure), a face prediction key point determination sub-module (not shown in the figure) and a sample face target image construction sub-module (not shown in the figure). Wherein, the face feature acquisition sub-module is configured to obtain the face features of the above-mentioned sample face input image; the face prediction key point determination sub-module is configured to determine the face prediction key points based on the above-mentioned face features; The sample face target image construction sub-module is configured to construct the sample face target image through the above-mentioned face prediction key points.

In an optional implementation manner of some embodiments, the above-mentioned face features include face spatial inclination and face structure information, and the above-mentioned face prediction key point determination sub-module may include: a prediction component (not shown in the figure) ) and a face prediction keypoint determination component (not shown in the figure). Wherein, the prediction component is configured to determine the position information of face prediction key points and the initial image of face prediction key points based on the above-mentioned face structure information; the face prediction key point determination component is configured to The above face prediction key point position information and the face prediction key point initial image are adjusted to determine the face prediction key point.

In an optional implementation manner of some embodiments, the above-mentioned sample face target image construction sub-module may include: an initial face image construction component (not shown in the figure) and a sample face target image generation component (not shown in the figure) out). Wherein, the initial face image construction component is configured to construct an initial face image based on the above-mentioned face prediction key points; the sample face target image generation component is configured to render the above-mentioned initial face image to obtain a sample face target image .

It can be understood that the units recorded in the apparatus 500 correspond to the respective steps in the method described with reference to FIG. 2 . Therefore, the operations, features and beneficial effects described above with respect to the method are also applicable to the apparatus 500 and the units included therein, and details are not described herein again.

As shown in FIG. 6, an electronic device 600 may include a processing device (eg, a central processing unit, a graphics processor, etc.) 601 that may be loaded into random access according to a program stored in a read only memory (ROM) 602 or from a storage device 608 Various appropriate actions and processes are executed by the programs in the memory (RAM) 603 . In the RAM 603, various programs and data necessary for the operation of the electronic device 600 are also stored. The processing device 601 , the ROM 602 , and the RAM 603 are connected to each other through a bus 604 . An input/output (I/O) interface 605 is also connected to bus 604 .

Typically, the following devices can be connected to the I/O interface 605: input devices 606 including, for example, a touch screen, touchpad, keyboard, mouse, camera, microphone, accelerometer, gyroscope, etc.; including, for example, a liquid crystal display (LCD), speakers, vibration An output device 607 of a computer, etc.; a storage device 608 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 609. Communication means 609 may allow electronic device 600 to communicate wirelessly or by wire with other devices to exchange data. While FIG. 6 shows electronic device 600 having various means, it should be understood that not all of the illustrated means are required to be implemented or provided. More or fewer devices may alternatively be implemented or provided. Each block shown in FIG. 6 may represent one device, or may represent multiple devices as required.

In particular, according to some embodiments of the present disclosure, the processes described above with reference to the flowcharts may be implemented as computer software programs. For example, some embodiments of the present disclosure include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the method illustrated in the flowchart. In some such embodiments, the computer program may be downloaded and installed from the network via the communication device 609 , or from the storage device 608 , or from the ROM 602 . When the computer program is executed by the processing device 601, the above-mentioned functions defined in the methods of some embodiments of the present disclosure are performed.

It should be noted that, in some embodiments of the present disclosure, the computer-readable medium described above may be a computer-readable signal medium or a computer-readable storage medium, or any combination of the foregoing two. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus or device, or a combination of any of the above. More specific examples of computer readable storage media may include, but are not limited to, electrical connections with one or more wires, portable computer disks, hard disks, random access memory (RAM), read only memory (ROM), erasable Programmable read only memory (EPROM or flash memory), fiber optics, portable compact disk read only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing. In some embodiments of the present disclosure, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. Rather, in some embodiments of the present disclosure, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code therein. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the foregoing. A computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device . Program code embodied on a computer readable medium may be transmitted using any suitable medium including, but not limited to, electrical wire, optical fiber cable, RF (radio frequency), etc., or any suitable combination of the foregoing.

In some embodiments, the client and server can communicate using any currently known or future developed network protocol such as HTTP (HyperText Transfer Protocol), and can communicate with digital data in any form or medium (eg, a communications network) interconnected. Examples of communication networks include local area networks ("LAN"), wide area networks ("WAN"), the Internet (eg, the Internet), and peer-to-peer networks (eg, ad hoc peer-to-peer networks), as well as any currently known or future development network of.

The above-mentioned computer-readable medium may be included in the above-mentioned electronic device; or may exist alone without being assembled into the electronic device. The above-mentioned computer-readable medium carries one or more programs, and when the above-mentioned one or more programs are executed by the electronic device, the electronic device: detects the sharpness of the face image to be processed; Determine a sharpness threshold, generate a simulated face image based on the above-mentioned face image to be processed, and the definition of the above-mentioned simulated face image is greater than the above-mentioned sharpness threshold; face recognition information.

Computer program code for carrying out operations of some embodiments of the present disclosure may be written in one or more programming languages, including object-oriented programming languages—such as Java, Smalltalk, C++, or a combination thereof, Also included are conventional procedural programming languages - such as the "C" language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer, or entirely on the remote computer or server. Where a remote computer is involved, the remote computer may be connected to the user's computer through any kind of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (eg, using an Internet service provider to via Internet connection).

The flowchart and block diagrams in the Figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code that contains one or more logical functions for implementing the specified functions executable instructions. It should also be noted that, in some alternative implementations, the functions noted in the blocks may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It is also noted that each block of the block diagrams and/or flowchart illustrations, and combinations of blocks in the block diagrams and/or flowchart illustrations, can be implemented in dedicated hardware-based systems that perform the specified functions or operations , or can be implemented in a combination of dedicated hardware and computer instructions.

The units described in some embodiments of the present disclosure may be implemented by means of software, and may also be implemented by means of hardware. The described unit can also be provided in the processor, for example, it can be described as: a processor includes an image sharpness detection unit, a face simulation image generation unit and a face recognition unit. Wherein, the names of these units do not constitute a limitation of the unit itself under certain circumstances, for example, the face simulation image generation unit can also be described as "a unit that generates a fake face image corresponding to the face image to be processed".

The functions described herein above may be performed, at least in part, by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: Field Programmable Gate Arrays (FPGAs), Application Specific Integrated Circuits (ASICs), Application Specific Standard Products (ASSPs), Systems on Chips (SOCs), Complex Programmable Logical Devices (CPLDs) and more.

According to one or more embodiments of the present disclosure, there is provided a method for recognizing a face image, comprising: detecting the sharpness of the face image to be processed; in response to the above-mentioned sharpness being less than a set sharpness threshold, based on the above-mentioned to-be-processed face image The face image generates a face simulation image, and the definition of the face simulation image is greater than the definition threshold; the face simulation image is recognized to obtain the face recognition information corresponding to the to-be-processed face image.

According to one or more embodiments of the present disclosure, generating the face simulation image based on the to-be-processed face image includes: importing the to-be-processed face image into a face image generation model to obtain the face simulation image.

According to one or more embodiments of the present disclosure, the above-mentioned face image generation model is obtained by the following steps: obtaining a plurality of sample face input images and corresponding to each sample face input image in the above-mentioned plurality of sample face input images The sample face target image, wherein the clarity of the above-mentioned sample face input image is less than or equal to the above-mentioned sharpness threshold, the above-mentioned sample face target image and the above-mentioned sample face input image have the same image content, and the clarity is greater than the above-mentioned clarity Threshold image; extracting a pre-established generative adversarial network, wherein the above-mentioned generative adversarial network includes a generative network and a discriminant network, the generative network is used to generate a face target image by using the face input image, and the discriminant network is used to determine the input of the above-mentioned discrimination. Whether the image of the network is an image output by the generation network; using the machine learning method, each sample face input image in the above-mentioned multiple sample face input images is used as the input of the generation network, and the image output by the generation network and the input generated by the network are generated. The sample face target image corresponding to the sample face input image is used as the input of the discriminant network, the generation network and the discriminant network are trained, and the trained generation network is determined as the face image generation model.

According to one or more embodiments of the present disclosure, the above-mentioned training of the generation network and the discrimination network, and determining the trained generation network as a face image generation model, includes: performing the following training steps: fixing the parameters of the generation network, Each sample face input image in the multiple sample face input images is used as the input of the generation network, and the output image of the generation network and the sample face target image corresponding to the sample face input image input to the generation network are used as the input of the discriminant network. , using the machine learning method to train the discriminant network; fixing the parameters of the discriminant network after training, using each sample face input image in the above-mentioned multiple sample face input images as the input of the generation network, and using the machine learning method to generate The network is trained; the accuracy of the discriminant result output by the discriminant network after training is determined, and in response to determining that the accuracy is greater than the accuracy threshold, the most recently trained generation network is determined as the face image generation model.

According to one or more embodiments of the present disclosure, the above-mentioned training of the generation network and the discrimination network, and determining the trained generation network as a face image generation model, further includes: in response to determining that the accuracy rate is less than or equal to the accuracy rate threshold, using The most recently trained generative and discriminative networks re-execute the above training steps.

According to one or more embodiments of the present disclosure, the above-mentioned sample face target image is obtained through the following steps: obtaining face features of the above-mentioned sample face input image; determining key points for face prediction based on the above-mentioned face characteristics; Predict keypoints to construct a sample face target image.

According to one or more embodiments of the present disclosure, the above-mentioned face features include face spatial inclination and face structure information, and the above-mentioned determination of face prediction key points based on the above-mentioned face characteristics includes: based on the above-mentioned face structure information Determine the position information of the face prediction key point and the initial image of the face prediction key point; adjust the position information of the face prediction key point and the initial image of the face prediction key point based on the above-mentioned face space inclination, and determine the face prediction key point. .

According to one or more embodiments of the present disclosure, constructing a sample face target image by using the above-mentioned face prediction key points includes: constructing an initial face image based on the above-mentioned face prediction key points; rendering the above-mentioned initial face image to obtain Sample face target image.

According to one or more embodiments of the present disclosure, there is provided a face image recognition device, comprising: an image clarity detection unit configured to detect the clarity of the face image to be processed; a face simulation image generation unit, In response to the above-mentioned sharpness being less than the set sharpness threshold, it is configured to generate a face simulation image based on the above-mentioned face image to be processed, and the sharpness of the above-mentioned face simulation image is greater than the above-mentioned sharpness threshold; the face recognition unit is configured to The above-mentioned face simulation image is recognized to obtain the face recognition information corresponding to the above-mentioned to-be-processed face image.

According to one or more embodiments of the present disclosure, the above-mentioned face simulation image generation unit includes: a face simulation image generation subunit, configured to import the above-mentioned face image to be processed into a face image generation model to obtain the above-mentioned face simulation image.

According to one or more embodiments of the present disclosure, the above-mentioned face simulation image generation subunit includes a face image generation model module configured to train a face image generation model, and the above-mentioned face image generation model module includes: a sample acquisition submodule , is configured to obtain a plurality of sample face input images and a sample face target image corresponding to each sample face input image in the above-mentioned plurality of sample face input images, wherein the clarity of the above-mentioned sample face input images less than or equal to the above-mentioned sharpness threshold, the above-mentioned sample face target image is the same as the image content of the above-mentioned sample face input image, and the image whose sharpness is greater than the above-mentioned sharpness threshold; the network extraction sub-module is configured to extract the pre-established generative formula Adversarial network, wherein, the above-mentioned generative confrontation network includes a generation network and a discriminant network, the generation network is used to generate a face target image using a face input image, and the discriminant network is used to determine whether the image input to the above-mentioned discrimination network is an image output by the generation network The face image generation model training sub-module is configured to utilize the machine learning method to use each sample face input image in the above-mentioned multiple sample face input images as the input of the generation network, and the image and the input output of the generation network will be generated. The sample face target image corresponding to the sample face input image of the generation network is used as the input of the discriminant network, the generation network and the discriminant network are trained, and the trained generation network is determined as the face image generation model.

According to one or more embodiments of the present disclosure, the above-mentioned face image generation model training sub-module includes: a face image generation model training module, configured to fix the parameters of the above-mentioned generation network, and input the above-mentioned multiple sample faces into the image Each sample face input image in the network is used as the input of the generation network, and the sample face target image corresponding to the output image of the generation network and the sample face input image input to the generation network is used as the input of the discriminant network, and the machine learning method is used to discriminate. The network is trained; the parameters of the discriminant network after training are fixed, each sample face input image in the above-mentioned multiple sample face input images is used as the input of the generation network, and the generation network is trained by the machine learning method; The accuracy rate of the discrimination result output by the discriminant network, in response to determining that the accuracy rate is greater than the accuracy rate threshold, the most recent training generation network is determined as the face image generation model.

According to one or more embodiments of the present disclosure, the above-mentioned face image generation model training sub-module further includes: an adjustment module, configured to use the most recently trained generation network and The discriminant network returns to the above face image generation model training module.

According to one or more embodiments of the present disclosure, the above-mentioned sample acquisition sub-module includes a sample face target image construction module, and the above-mentioned sample face target image construction module includes: a face feature acquisition sub-module configured to acquire the above-mentioned face features of the sample face input image; face prediction key point determination sub-module, configured to determine face prediction key points based on the above-mentioned face features; sample face target image construction sub-module, configured to pass the above Face prediction keypoints construct sample face target images.

According to one or more embodiments of the present disclosure, the above-mentioned face features include face spatial inclination and face structure information, and the above-mentioned face prediction key point determination sub-module includes:

The prediction component is configured to determine the position information of the face prediction key points and the initial image of the face prediction key point based on the above-mentioned face structure information; The face prediction key point position information and the face prediction key point initial image are adjusted to determine the face prediction key point.

According to one or more embodiments of the present disclosure, the above-mentioned sample face target image construction sub-module includes: an initial face image construction component configured to construct an initial face image based on the above-mentioned face prediction key points; the sample face target image The image generation component is configured to render the above-mentioned initial face image to obtain a sample face target image.

The above descriptions are merely some preferred embodiments of the present disclosure and illustrations of the applied technical principles. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present disclosure is not limited to the technical solution formed by the specific combination of the above technical features, and should also cover, without departing from the above inventive concept, the above Other technical solutions formed by any combination of technical features or their equivalent features. For example, a technical solution is formed by replacing the above-mentioned features with the technical features disclosed in the embodiments of the present disclosure (but not limited to) with similar functions.

Claims (18)

1. A face image recognition method, comprising: Detect the sharpness of the face image to be processed; In response to the sharpness being less than a set sharpness threshold, generating a face simulation image based on the to-be-processed face image, the sharpness of the face simulation image being greater than the sharpness threshold; The face simulation image is identified to obtain face identification information corresponding to the to-be-processed face image. 2. The method according to claim 1, wherein the generating a face simulation image based on the to-be-processed face image comprises: The face image to be processed is imported into a face image generation model to obtain the face simulation image. 3. The method according to claim 2, wherein the face image generation model is obtained by the following steps: Obtain a plurality of sample face input images and a sample face target image corresponding to each sample face input image in the plurality of sample face input images, wherein the clarity of the sample face input images is less than or equal to the definition threshold, the image content of the sample face target image is the same as that of the sample face input image, and the clarity is greater than the image with the clarity threshold; Extract a pre-established generative adversarial network, wherein the generative adversarial network includes a generative network and a discriminant network, the generative network is used to generate a face target image using the face input image, and the discriminant network is used to determine the input to the discriminant network. Whether the image is the image that generated the output of the network; Using the machine learning method, each sample face input image in the multiple sample face input images is used as the input of the generation network, and the sample face input image corresponding to the image output by the generation network and the sample face input image input into the generation network The face target image is used as the input of the discriminant network, the generation network and the discriminant network are trained, and the trained generation network is determined as the face image generation model. 4. The method according to claim 3, wherein the generating network and the discriminating network are trained, and the trained generating network is determined as a face image generation model, comprising: Perform the following training steps: fixing the parameters of the generation network, using each sample face input image in the multiple sample face input images as the input of the generation network, and generating the image output from the network and the input to generate the sample of the network The sample face target image corresponding to the face input image is used as the input of the discriminant network, and the discriminant network is trained by the machine learning method; the parameters of the discriminant network after training are fixed, and each of the multiple sample face input images is The sample face input image is used as the input of the generation network, and the generation network is trained by machine learning method; the accuracy of the discrimination result output by the discriminant network after training is determined, and in response to the determination that the accuracy is greater than the accuracy threshold, the most recent training The generative network determines to generate models for face images. 5. The method according to claim 4, wherein the generating network and the discriminating network are trained, and the trained generating network is determined as a face image generation model, further comprising: In response to determining that the accuracy is less than or equal to the accuracy threshold, the training step is re-performed using the most recently trained generative network and discriminant network. 6. The method according to claim 3, wherein the sample face target image is obtained by the following steps: obtaining face features of the sample face input image; Determine a face prediction key point based on the face feature; A sample face target image is constructed through the face prediction key points. 7. The method of claim 6, wherein the face features include face spatial inclination, face structure information, and Described determining the face prediction key point based on the described face feature, including: Determine face prediction key point position information and face prediction key point initial image based on the face structure information; The position information of the face prediction key point and the initial image of the face prediction key point are adjusted based on the face space inclination to determine the face prediction key point. 8. The method according to claim 6, wherein said constructing a sample face target image through the face prediction key points comprises: constructing an initial face image based on the face prediction key points; A sample face target image is obtained by rendering the initial face image. 9. A face image recognition device, comprising: an image sharpness detection unit, configured to detect the sharpness of the face image to be processed; A face simulation image generation unit, configured to generate a face simulation image based on the to-be-processed face image, in response to the definition being less than a set definition threshold, the face simulation image having a definition greater than the clarity degree threshold; The face recognition unit is configured to recognize the face simulation image to obtain face recognition information corresponding to the to-be-processed face image. 10. The apparatus according to claim 9, wherein the face simulation image generation unit comprises: The face simulation image generation subunit is configured to import the to-be-processed face image into a face image generation model to obtain the face simulation image. 11. The apparatus according to claim 10, wherein the face simulation image generation subunit comprises a face image generation model module configured to train a face image generation model, and the face image generation model module comprises: a sample acquisition sub-module configured to acquire a plurality of sample face input images and a sample face target image corresponding to each sample face input image in the plurality of sample face input images, wherein the sample face The clarity of the face input image is less than or equal to the clarity threshold, the image content of the sample face target image and the sample face input image is the same, and the clarity is greater than the clarity threshold image; The network extraction sub-module is configured to extract a pre-established generative adversarial network, wherein the generative adversarial network includes a generative network and a discriminant network, the generative network is used to generate a face target image using the face input image, and the discriminant network uses for determining whether the image input to the discriminant network is an image output by the generation network; The face image generation model training sub-module is configured to use the machine learning method to use each sample face input image in the plurality of sample face input images as the input of the generation network, and the image and the input output of the generation network will be generated. The sample face target image corresponding to the sample face input image of the generation network is used as the input of the discriminant network, the generation network and the discriminant network are trained, and the trained generation network is determined as the face image generation model. 12. The apparatus according to claim 11, wherein the face image generation model training submodule comprises: The face image generation model training module is configured to fix the parameters of the generation network, and each sample face input image in the multiple sample face input images is used as the input of the generation network, and the output of the generation network is generated. The image and the sample face target image corresponding to the input image of the input generation network are used as the input of the discriminant network, and the discriminant network is trained by the machine learning method; the parameters of the discriminant network after training are fixed, and the multiple sample people are Each sample face input image in the face input image is used as the input of the generation network, and the generation network is trained by the machine learning method; the accuracy rate of the discrimination result output by the discriminant network after training is determined, in response to determining that the accuracy rate is greater than the accuracy rate Threshold to determine the most recently trained generative network as a face image generative model. 13. The apparatus according to claim 12, wherein the face image generation model training submodule further comprises: The adjustment module, in response to determining that the accuracy rate is less than or equal to the accuracy rate threshold, is configured to return the face image generation model training module using the most recently trained generation network and discriminant network. 14. The apparatus according to claim 11, wherein the sample acquisition submodule comprises a sample face target image construction module, and the sample face target image construction module comprises: a face feature acquisition sub-module, configured to obtain the face feature of the sample face input image; a face prediction key point determination submodule, configured to determine a face prediction key point based on the face feature; The sample face target image construction sub-module is configured to construct the sample face target image through the face prediction key points. 15. The apparatus of claim 14, wherein the face features include face spatial inclination, face structure information, and The face prediction key point determination submodule includes: a prediction component, configured to determine face prediction key point position information and face prediction key point initial image based on the face structure information; The face prediction key point determination component is configured to adjust the position information of the face prediction key point and the initial image of the face prediction key point based on the inclination of the face space to determine the face prediction key point. 16. The apparatus according to claim 14, wherein the sample face target image construction sub-module comprises: an initial face image construction component, configured to construct an initial face image based on the face prediction key points; The sample face target image generation component is configured to render the initial face image to obtain the sample face target image. 17. An electronic device comprising: one or more processors; a storage device on which one or more programs are stored, The one or more programs, when executed by the one or more processors, cause the one or more processors to implement a method as claimed in any one of claims 1 to 8. 18. A computer-readable medium having stored thereon a computer program, wherein the program, when executed by a processor, implements the method of any one of claims 1 to 8.

Images