WO2020248387A1

WO2020248387A1 - Face recognition method and apparatus based on multiple cameras, and terminal and storage medium

Info

Publication number: WO2020248387A1
Application number: PCT/CN2019/103389
Authority: WO
Inventors: 戴磊
Original assignee: 平安科技（深圳）有限公司
Priority date: 2019-06-11
Filing date: 2019-08-29
Publication date: 2020-12-17
Also published as: CN110443110B; CN110443110A

Abstract

A face recognition method based on multiple cameras, comprising: establishing a sample database for each camera; obtaining a face image to be detected photographed by each camera and recognizing a photographing angle of said face image; calculating an angle difference between the photographing angle of said face image and the photographing angle of each camera; screening out the minimum angle difference to serve as a target angle difference; determining the sample database of the camera corresponding to the target angle difference as a target sample database; and recognizing said face image photographed by the camera corresponding to the target angle difference according to the face image in the target sample database. The present application further provides a face recognition apparatus based on multiple cameras, a terminal and a storage medium. According to the present application, when face recognition is carried out, said face image obtained by each camera is only compared with the face image in the corresponding target sample database, the face recognition accuracy is improved, and the face recognition rate is improved.

Description

Face recognition method, device, terminal and storage medium based on multi-channel camera

This application claims the priority of the Chinese patent application filed with the Chinese Patent Office on June 11, 2019. The application number is 201910502775.1. The invention title is "Multi-channel camera-based face recognition method, device, terminal and storage medium". The content is incorporated in this application by reference.

Technical field

This application relates to the field of image recognition technology, and in particular to a face recognition method based on multiple cameras, a face recognition device based on multiple cameras, a terminal and a storage medium.

Background technique

With the development and progress of image processing technology, face recognition technology, as a high-end and reliable means of identity detection, has greatly improved the user experience in modern production and life applications, so it can be widely used in access management, Monitoring management, computer security protection, photo search, ATM machine intelligent video alarm, railway security inspection and identification and other applications.

In the prior art, when performing face recognition, a multi-channel camera device is generally used to capture the image of the face to be recognized, and then through a series of image preprocessing and image recognition processing, the face in the captured image is processed Recognition, obtain the identity recognition result of the face to be recognized. The inventor found that the current general practice is that all camera devices share the same sample database, but the placement of each camera device, the environment, and the angle at which the face is captured may be different, resulting in the same sample database. The adaptability is different. For example, the shooting angle, for a sample database where the face images in the sample database are all frontal images, the face recognition effect corresponding to the face image captured by the camera device with the frontal shooting angle is good, otherwise it will be worse. .

Summary of the invention

In view of the above, it is necessary to propose a face recognition method based on multiple cameras, a face recognition device based on multiple cameras, a terminal and a storage medium. When performing face recognition, each camera obtains the under-test The face image is only compared with the face image in the corresponding target sample database to improve the accuracy of face recognition and increase the rate of face recognition.

The first aspect of the embodiments of the present application provides a face recognition method based on multi-channel camera. The face recognition method based on multi-channel camera includes:

Establishing a sample database for each camera device in the preset multi-channel camera device, and the shooting angle of the face image in the sample database is the same as the shooting angle of the corresponding camera device;

Acquire multiple face images to be measured of faces to be measured that are simultaneously photographed by each camera device;

Identifying the shooting angle of the face image to be tested according to the pre-trained face angle recognition model;

Calculating the angle difference between the shooting angle of the face image to be measured and the shooting angle of each camera device;

The smallest angle difference is selected from the angle difference as the target angle difference;

Determining the sample database of the camera device corresponding to the target angle difference as the target sample database;

Identify the face image to be measured taken by the camera device corresponding to the target angle difference according to the face image in the target sample database.

The second aspect of the embodiments of the present application also provides a face recognition device based on multi-channel camera, the face recognition device based on multi-channel camera includes:

A sample database establishment module for establishing a sample database corresponding to each camera device in the preset multi-channel camera device, and the photographing angle of the face image in the sample database is the same as the photographing angle of the corresponding camera device;

The face image acquisition module is used to acquire multiple faces to be measured in the face images to be measured taken by each camera at the same time;

A shooting angle recognition module, configured to recognize the shooting angle of the face image to be tested according to a pre-trained face angle recognition model;

An angle difference calculation module, configured to calculate the angle difference between the shooting angle of the face image to be measured and the shooting angle of each camera device;

The target difference determination module is configured to filter out the smallest angle difference from the angle difference as the target angle difference;

A sample database determining module, configured to determine the sample database of the camera device corresponding to the target angle difference as the target sample database;

The face recognition module is configured to recognize, according to the face image in the target sample database, the face image to be measured taken by the camera device corresponding to the target angle difference.

The third aspect of the embodiments of the present application further provides a terminal, the terminal includes a processor, and the processor is configured to implement the multi-channel camera-based face of any one of the above when executing computer-readable instructions stored in the memory. recognition methods.

The fourth aspect of the embodiments of the present application also provides a non-volatile readable storage medium having computer readable instructions stored on the non-volatile readable storage medium, and the computer readable instructions are implemented when executed by a processor The face recognition method based on multi-channel camera described in any one of the above.

The embodiment of the application provides a face recognition method based on multiple cameras, a face recognition device based on multiple cameras, a terminal, and a non-volatile readable storage medium, which are preset for each camera in the multiple camera The device establishes a sample database, and the photographing angle of the face image in the sample database is the same as the photographing angle of the corresponding camera device; acquiring a plurality of faces to be measured of the faces to be measured that are simultaneously photographed by each camera Image; recognize the shooting angle of the face image to be tested according to a pre-trained face angle recognition model; calculate the angle difference between the shooting angle of the face image to be tested and the shooting angle of each camera The smallest angle difference is selected from the angle difference as the target angle difference; the sample database of the camera device corresponding to the target angle difference is determined as the target sample database; according to the target sample database The face image recognizes the face image to be measured taken by the camera device corresponding to the target angle difference. Using the embodiment of the present application, a sample database is established for each camera setting of the multi-channel camera device, and the shooting angle of the face image stored in the sample database is the same as the shooting angle of the corresponding camera device. When performing face recognition, for the captured face image to be tested, select the face image in the sample database that is closest to its shooting angle for comparison, so that the face image to be tested is the closest to its shooting angle and environment Perform face comparison to improve the accuracy of face recognition and increase the rate of face recognition.

Description of the drawings

FIG. 1 is a flowchart of a face recognition method based on multi-channel camera provided by an embodiment of the present application.

FIG. 2 is a schematic structural diagram of a terminal according to an embodiment of the present application.

Fig. 3 is an exemplary functional block diagram of the terminal shown in Fig. 2.

The following specific implementations will further illustrate the embodiments of the present application in conjunction with the above-mentioned drawings.

Detailed ways

FIG. 1 is a flowchart of a face recognition method based on multiple cameras in the first embodiment of the present application. The face recognition method based on multiple cameras can be applied to a terminal 1, and the terminal 1 may be, for example, a smart phone, Smart devices such as notebook computers, desktop/tablet computers, smart watches, and personal digital assistants (PDAs). As shown in Fig. 1, the face recognition method based on multi-channel camera may include the following steps:

S101: Establish a sample database for each of the preset multi-channel camera devices, and the photographing angle of the face image in the sample database is the same as the photographing angle of the corresponding camera device.

In at least one embodiment of the present application, the preset multi-channel camera device is a camera device preset at multiple shooting positions. Due to the different shooting positions, the corresponding camera devices have different angles, and the angles of the captured face images are also different. The camera device may be one or more of a camera, a digital camera, and a digital camera. The number of the camera devices is not limited here, and the number of the camera devices can be set according to the actual shooting angle. Taking the number of the camera devices as three as an example, the preset multi-channel camera device includes a first camera device, a second camera device, and a third camera device arranged in a common center.

It should be noted that the shooting angle is relative. For example, it is possible to set the position of the first camera device located in the center as 0°, and the angle of the face image captured by the corresponding first camera device is 0°; The position of the second camera on the right becomes a positive angle, and the angle of the face image taken by the corresponding second camera is positive; all the third cameras located on the left of the first camera The position of the road camera device becomes a negative angle, and the corresponding angle of the face image taken by the third road camera device is negative.

Exemplarily, assuming that the center is used as the center point, three different positions are set, namely the first position, the second position, and the third position. Among them, the first position is 0°, and the first position is installed with the first camera; the second position is +90°, and the second position is installed with the second camera; the third position is -90° , And a third camera is installed at the third position. Then, the angle of the face taken by the first camera is 0°, and the face taken at this angle is a front face; the angle of the face taken by the second camera is +90°, and the person taken at this angle The face is the right face; the face angle taken by the third camera device is -90°, and the face taken at this angle is the left face. Then, an independent sample database of left face/front face/right face is set up, and the independent sample database of left face/front face/right face is used to store the left face/front face/right face sample photos of different users.

The establishing a sample database for each camera device in the preset multi-channel camera device includes: taking the center as the center point, setting the shooting angle of the first camera device to 0°, and setting the second camera device The shooting angle of the camera device is +90°, and the shooting angle of the third camera device is set to -90°; the first camera device, the second camera device, and the third camera device are controlled Simultaneously photograph a preset number of target testers for each of the target testers to obtain face images of different shooting angles; take the preset number of faces of the target testers captured by the first camera The image is used as the first sample database of the first camera, and the preset number of face images of the target tester captured by the second camera are used as the second The sample database uses the preset number of face images of the target tester captured by the third camera device as the third sample database of the third camera device. Wherein, the preset number is preset, and the number of target testers who need to collect face images.

It is understandable that, in addition to storing a preset number of face images of the target tester in the first sample database, the second sample database, and the third sample database, the sample database also includes human faces. Image identification, the face image identification of the same target tester is unique, and the face image identification is used to identify the identity information of the corresponding tester. Therefore, in one embodiment, it is also possible to perform face recognition by judging whether the face image identifiers are consistent. When the judgment result is that the face image identifiers are consistent, the face recognition is successful; when the judgment result is all When the face image identification is inconsistent, the face recognition fails. Performing face recognition through the face image identification can improve the efficiency of face recognition.

Specifically, multiple target testers can be contacted, so that each target tester is facing the first camera device located in the center, and the target tester is photographed simultaneously through the first, second and third camera devices, namely Face images of different shooting angles can be obtained. The multiple face images of multiple target testers captured by the first camera device are used as the first sample database corresponding to the first camera device; the multiple face images of multiple target testers captured by the second camera device One face image is used as the second sample database corresponding to the second camera; the multiple face images of multiple target testers captured by the third camera are taken as the third camera corresponding to the third camera. Sample database. In this way, it is possible to establish a sample database corresponding to the camera devices at a plurality of preset positions, and the shooting angle of the face image in each sample database corresponds to the shooting angle of the camera device.

S102: Acquire multiple face images to be measured of faces to be measured that are simultaneously photographed by each camera.

In at least one embodiment of the present application, after the acquiring multiple face images of the face to be measured that are simultaneously captured by each camera device, the method further includes: photographing each camera device The face image to be tested is preprocessed, and the preprocessing includes one or more of the following: image denoising processing; light normalization processing; posture calibration processing; gray-scale normalization processing. It is understandable that, after acquiring multiple face images of the face to be measured that are simultaneously captured by each camera device, preprocessing the face image to be measured can further improve the face of the face to be measured. The sharpness of the image facilitates the extraction of more discriminative facial features.

S103: Recognize the shooting angle of the face image to be tested according to the pre-trained face angle recognition model.

In at least one embodiment of the present application, for the captured face image to be tested, the shooting angle of the face image to be tested is recognized according to a pre-trained face angle recognition model. The recognizing the shooting angle of the face image to be tested includes: acquiring the face image to be tested taken by the first camera; inputting the face image to be tested into a pre-trained face angle recognition model; Confirm the shooting angle of the face to be detected through the pre-trained face angle recognition model. The pre-trained face angle recognition model is a model pre-trained by a terminal user for recognizing the shooting angle of a face image.

Specifically, the training process of the face angle recognition model includes: acquiring a face image data set, the face image data set including a plurality of people's face images of different shooting angles; Divide to obtain a training set and a test set; input the face images in the training set into a preset neural network model for training to obtain a face angle recognition model; input the face images in the test set to the The face angle recognition model is tested to obtain a test pass rate; when the test pass rate is greater than or equal to the preset pass rate threshold, the training of the face angle recognition model ends; if the test pass rate is less than the When the pass rate threshold is preset, the number of face images in the training set is increased, and the face angle recognition model is retrained. Wherein, the preset pass rate threshold is preset by the terminal user according to an empirical value.

S104: Calculate the angle difference between the shooting angle of the face image to be measured and the shooting angle of each camera.

In at least one embodiment of the present application, the shooting angle of each camera device is known, and the difference between the shooting angle of the face image to be measured and the shooting angle of each camera device is obtained by calculation. Angle difference.

S105: Screen the smallest angle difference from the angle difference as the target angle difference.

In at least one embodiment of the present application, the smallest angle difference is selected from the angle difference as the target angle difference. It can be understood that the target angle difference indicates the shooting angle of the face image to be measured It is the closest to the shooting angle of the corresponding camera device, so the face recognition effect will be the best.

S106: Determine the sample database of the camera device corresponding to the target angle difference as the target sample database.

In at least one embodiment of the present application, the sample database of the camera device corresponding to the target angle difference is determined as the target sample database. It can be understood that the face image in the target sample database is the same as the target sample database. The shooting angle of the face image is the closest, so the face recognition effect will be the best.

S107: Identify, according to the face image in the target sample database, the face image to be measured taken by the camera device corresponding to the target angle difference.

In at least one embodiment of the present application, the recognizing, according to the face image in the target sample database, the face image to be measured taken by the camera corresponding to the target angle difference includes: extracting the face to be measured The facial features of the image and the facial features of the facial image in the target sample database; calculating the similarity between the facial features of the facial image to be tested and the facial features of the facial image in the target sample database; Determine whether the similarity is greater than or equal to the preset similarity threshold; if the judgment result is that the similarity is greater than or equal to the preset similarity threshold, output the face image corresponding to the similarity in the sample database as the result Describe the face recognition result of the face image to be tested. The shooting angle of the corresponding face image in the target sample database is consistent with the shooting angle of the face image to be tested. The preset similarity threshold can be flexibly set according to specific application scenarios, for example, a simulation can be performed with facial image features corresponding to multiple photos, and finally a threshold is determined as the preset similarity threshold.

For multi-channel camera devices, when each camera device performs face recognition, there may be multiple candidate objects in its corresponding face recognition result. Therefore, it is necessary to comprehensively analyze the candidate objects corresponding to the multi-channel camera device to obtain the face Recognition results.

In one embodiment, when it is determined that the similarity is less than the preset similarity threshold, the method further includes: identifying, according to the first sample database, the to-be-photographed by the first camera The first candidate face image of the face image to be tested; the second candidate face image of the face image to be tested taken by the second camera device is identified according to the second sample database; according to the third The sample database identifies the third candidate face image of the face image to be tested taken by the third-channel camera device; determines the first candidate face image, the second candidate face image, and the first candidate face image Whether the three candidate face images are the same person's face image; if the judgment result is that the first candidate face image, the second candidate face image, and the third candidate face image are the same person's face Image, the first candidate face image is output as the final face recognition result.

It is understandable that the method further includes: respectively acquiring the face image identifiers corresponding to the first candidate face image, the second candidate face image, and the third candidate face image; and judging the first candidate face image; Whether the face image identification of a candidate face image, the face image identification of the second candidate face image, and the face image identification of the third candidate face image are the same; if the judgment result is the face image If the identifiers are the same, the first candidate face image corresponding to the facial image identifier is output as the final facial recognition result.

In other implementation manners, the number of the first candidate face image, the second candidate face image, and the third candidate face image may be one or more. When the number of the first candidate face image, the second candidate face image, and the third candidate face image is one and the face image of the same person, output the first candidate face The image is used as the final face recognition image. When the number of the first candidate's face image, the second candidate's face image, and the third candidate's face image is one and not the same person's face image, it means that each camera is taken If the received face image and the face image in the corresponding target sample database are not the same as the result of face recognition, the final face recognition fails; when the first candidate face image, the second candidate face image and When the number of the third candidate face images has the same face images but the number is not unique, acquiring the similarity between the face features corresponding to each same candidate face image and the face features in the target sample database, The three similarity values are summed to obtain the similarity and value of each identical candidate face image; the candidate face image with the highest similarity and value is output as the final face recognition result.

The embodiment of the present application provides a face recognition method based on multi-channel camera. A sample database is established for each camera device in the preset multi-channel camera device. The shooting angle of the face image in the sample database and the corresponding The shooting angles of the camera devices are the same; acquiring multiple face images to be tested of the face to be tested taken by each camera at the same time; identifying the face image of the face to be tested according to a pre-trained face angle recognition model Shooting angle; calculating the angle difference between the shooting angle of the face image to be measured and the shooting angle of each camera; selecting the smallest angle difference from the angle difference as the target angle difference Determine the sample database of the camera device corresponding to the target angle difference as the target sample database; identify the person to be tested taken by the camera device corresponding to the target angle difference according to the face image in the target sample database Face image. Using the embodiment of the present application, a sample database is established for each camera setting of the multi-channel camera device, and the shooting angle of the face image stored in the sample database is the same as the shooting angle of the corresponding camera device. When performing face recognition, for the captured face image to be tested, select the face image in the sample database that is closest to its shooting angle for comparison, so that the face image to be detected is the closest to its shooting angle and environment Perform face comparison to improve the accuracy of face recognition and increase the rate of face recognition.

The foregoing is a detailed description of the method provided by the embodiment of the present application. According to different requirements, the execution order of the blocks in the flowchart shown can be changed, and some blocks can be omitted. The following describes the terminal 1 provided in the embodiment of the present application.

An embodiment of the present application also provides a terminal 1, including a memory 10, a processor 30, and a computer program stored in the memory 10 and capable of running on the processor 30. The processor 30 implements any of the foregoing when the program is executed. The steps of the face recognition method based on multi-channel camera described in the implementation mode.

FIG. 2 is a schematic structural diagram of a terminal 1 according to an embodiment of the present application. As shown in FIG. 2, the terminal 1 includes a memory 10 in which a face recognition device 100 based on multi-channel photography is stored. The terminal 1 may be a terminal 1 with application display function, such as a mobile phone, a tablet computer, or a personal digital assistant. The face recognition device 100 based on multi-channel camera can establish a sample database for each camera device in the preset multi-channel camera device, and the shooting angle of the face image in the sample database corresponds to the corresponding camera device. The shooting angles are the same; acquiring multiple face images to be measured of the face to be measured that are shot simultaneously by each camera device; identifying the shooting angle of the face image to be measured according to a pre-trained face angle recognition model; calculate The angle difference between the shooting angle of the face image to be measured and the shooting angle of each camera device; the smallest angle difference is selected from the angle difference as the target angle difference; The sample database of the camera device of the target angle difference is determined as the target sample database; the face image to be measured taken by the camera device corresponding to the target angle difference is identified according to the face image in the target sample database. Using the embodiment of the present application, a sample database is established for each camera setting of the multi-channel camera device, and the shooting angle of the face image stored in the sample database is the same as the shooting angle of the corresponding camera device. When performing face recognition, for the captured face image to be tested, select the face image in the sample database that is closest to its shooting angle for comparison, so that the face image to be detected is the closest to its shooting angle and environment Perform face comparison to improve the accuracy of face recognition and increase the rate of face recognition.

In this embodiment, the terminal 1 may further include a display screen 20 and a processor 30. The memory 10 and the display screen 20 may be electrically connected to the processor 30 respectively.

The memory 10 may be different types of storage devices for storing various types of data. For example, it can be the memory or internal memory of the terminal 1, or a memory card that can be externally connected to the terminal 1, such as flash memory, SM card (Smart Media Card), SD card (Secure Digital Card, secure digital card) Wait. In addition, the memory 10 may include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a Secure Digital (SD) card, a flash memory card (Flash Card), At least one magnetic disk storage device, flash memory device, or other non-volatile solid state storage device. The memory 10 is used to store various types of data. For example, the memory 10 can store various types of applications installed in the terminal 1, and the memory 10 can also store applications based on the multi-channel camera. Information such as data obtained by face recognition methods.

The display screen 20 is installed in the terminal 1 for displaying information.

The processor 30 is configured to execute the face recognition method based on multi-channel cameras and various software installed in the terminal 1, such as an operating system and application display software. The processor 30 includes, but is not limited to, a processor (Central Processing Unit, CPU), a Micro Controller Unit (Micro Controller Unit, MCU) and other devices for interpreting computers and processing data in computer software.

The multi-channel camera-based face recognition device 100 may include one or more modules, and the one or more modules are stored in the memory 10 of the terminal 1 and configured to be composed of one or more processors ( This embodiment is executed by a processor 30) to complete the embodiment of the present application. For example, referring to FIG. 3, the face recognition device 100 based on multi-channel cameras may include a sample database establishment module 101, a face image acquisition module 103, a shooting angle recognition module 105, an angle difference calculation module 107, and a target difference The value determination module 109, the sample database determination module 111, and the face recognition module 113. The module referred to in the embodiment of the present application may be a program segment that completes a specific function, and is more suitable for describing the execution process of software in the processor 30 than a program.

It can be understood that, corresponding to the above-mentioned implementations of the face recognition method based on multiple cameras, the terminal 1 may include some or all of the functional modules shown in FIG. 3, and the functions of each module will be described in detail below . It should be noted that the same noun related terms and specific explanations in the above embodiments of the face recognition method based on multi-channel camera can also be applied to the following functional introduction of each module. To save space and avoid repetition, I won’t repeat them here.

The sample database establishment module 101 may be used to establish a sample database corresponding to each camera device in the preset multi-channel camera device, and the photographing angle of the face image in the sample database is the same as the photographing angle of the corresponding camera device.

The face image acquisition module 103 may be used to acquire multiple faces to be measured in the face images to be measured that are simultaneously captured by each camera.

The shooting angle recognition module 105 may be used to recognize the shooting angle of the face image to be tested according to a pre-trained face angle recognition model.

The angle difference calculation module 107 may be used to calculate the angle difference between the shooting angle of the face image to be measured and the shooting angle of each camera device.

The target difference value determining module 109 may be used to filter the smallest angle difference value from the angle difference values as the target angle difference value.

The sample database determining module 111 may be used to determine the sample database of the camera device corresponding to the target angle difference as the target sample database.

The face recognition module 113 may be configured to recognize, according to the face images in the target sample database, the face images to be measured taken by the camera device corresponding to the target angle difference.

The embodiments of the present application also provide a non-volatile readable storage medium, on which computer-readable instructions are stored, and when the computer-readable instructions are executed by the processor 30, the multi-channel camera-based system in any of the above embodiments is implemented The steps of the face recognition method.

If the multi-channel camera-based face recognition device 100/terminal 1/computer equipment integrated module/unit is realized in the form of a software functional unit and sold or used as an independent product, it can be stored in a non-volatile memory. Read the storage medium. Based on this understanding, this application implements all or part of the processes in the above-mentioned implementation methods, and can also be completed by instructing relevant hardware by a computer program, which can be stored in a non-volatile readable storage medium When the computer program is executed by the processor 30, the steps of the foregoing method embodiments can be implemented. The computer program includes computer-readable instructions, and the computer-readable instructions may be in the form of source code, object code, executable files, or some intermediate forms. When the computer-readable instructions are executed by the processor 30, the following steps are implemented: a sample database is established for each of the preset multi-channel camera devices, and the photographing angle of the face image in the sample database is corresponding to the The shooting angles of the camera devices are the same; obtain multiple face images to be tested of the face to be tested that are simultaneously shot by each camera device; identify the shooting of the face images to be tested according to the pre-trained face angle recognition model Angle; calculating the angle difference between the shooting angle of the face image to be measured and the shooting angle of each camera device; selecting the smallest angle difference from the angle difference as the target angle difference; The sample database of the camera device corresponding to the target angle difference is determined as the target sample database; the face to be measured captured by the camera device corresponding to the target angle difference is identified according to the face image in the target sample database image. The non-volatile readable storage medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM, Read -Only Memory) etc.

The so-called processor 30 may be a central processing unit (Central Processing Unit, CPU), other general processors, digital signal processors (Digital Signal Processor, DSP), application specific integrated circuits (Application Specific Integrated Circuit, ASIC), Ready-made programmable gate array (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc. The processor 30 is the control center of the multi-channel camera-based face recognition device 100/terminal 1, and uses various The interfaces and lines connect various parts of the entire face recognition device 100/terminal 1 based on multi-channel camera.

The memory 10 is used to store the computer program and/or module, and the processor 30 executes or executes the computer program and/or module stored in the memory and calls the data stored in the memory 10 to implement Various functions of the face recognition device 100/terminal 1 based on multi-channel camera. The memory 10 may mainly include a storage program area and a storage data area. The storage program area may store an operating system, an application program required by at least one function (such as a sound playback function, an image playback function, etc.), etc.; the storage data area may Store data (such as audio data, etc.) created according to the use of the terminal.

In the several specific implementation manners provided in this application, it should be understood that the disclosed terminal and method may be implemented in other ways. For example, the system implementation described above is only illustrative. For example, the division of the modules is only a logical function division, and there may be other division methods in actual implementation.

For those skilled in the art, it is obvious that the embodiments of the present application are not limited to the details of the above exemplary embodiments, and can be implemented in other specific forms without departing from the spirit or basic characteristics of the embodiments of the present application. example. Therefore, no matter from which point of view, the embodiments should be regarded as exemplary and non-limiting. The scope of the embodiments of the present application is defined by the appended claims rather than the above description, and therefore it is intended to fall on All changes within the meaning and scope of equivalent elements of the claims are included in the embodiments of the present application. Any reference signs in the claims should not be regarded as limiting the claims involved. Multiple units, modules or devices stated in the claims can also be implemented by the same unit, module or device through software or hardware.

The above implementation manners are only used to illustrate the technical solutions of the embodiments of the present application, not to limit them. Although the embodiments of the present application are described in detail with reference to the above preferred implementation manners, those of ordinary skill in the art should understand that the technical solutions of the embodiments of the application Modifications or equivalent replacements of the solutions should not depart from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

A face recognition method based on multi-channel camera, the face recognition method based on multi-channel camera includes:

Establishing a sample database for each camera device in the preset multi-channel camera device, and the shooting angle of the face image in the sample database is the same as the shooting angle of the corresponding camera device;

Acquire multiple face images to be measured of faces to be measured that are simultaneously photographed by each camera device;

Identifying the shooting angle of the face image to be tested according to the pre-trained face angle recognition model;

Calculating the angle difference between the shooting angle of the face image to be measured and the shooting angle of each camera device;

The smallest angle difference is selected from the angle difference as the target angle difference;

Determining the sample database of the camera device corresponding to the target angle difference as the target sample database;

Identify the face image to be measured taken by the camera device corresponding to the target angle difference according to the face image in the target sample database.
The face recognition method based on multi-channel cameras according to claim 1, wherein the preset multi-channel camera device includes a first camera device, a second camera device, and a third camera device arranged in a common center. Establishing a sample database for each of the preset multi-channel camera devices includes:

Taking the center as the center point, set the shooting angle of the first camera device to 0°, set the shooting angle of the second camera device to +90°, and set the shooting angle of the third camera device to -90 °;

Control the first camera device, the second camera device, and the third camera device to simultaneously capture a preset number of target testers for each of the target testers to obtain faces with different shooting angles image;

The face images of the preset number of target testers captured by the first camera device are used as the first sample database of the first camera device, and the second camera device A preset number of face images of the target tester are used as the second sample database of the second camera, and the preset number of face images of the target tester captured by the third camera are used as all The third sample database of the third camera device.
The face recognition method based on multi-channel camera according to claim 1, after said acquiring a plurality of face images to be tested of the face to be tested taken by each camera at the same time, the method further comprises:

Preprocessing is performed on the face images to be measured taken by each camera device, and the preprocessing includes one or more of the following: image denoising processing; light normalization processing; posture calibration processing; gray scale restoration One treatment.
The face recognition method based on multi-channel camera according to claim 2, wherein the recognizing the shooting angle of the face image to be tested according to the pre-trained face angle recognition model comprises:

Obtain the face image to be tested taken by the first camera;

Inputting the face image to be tested into a pre-trained face angle recognition model;

Confirm the shooting angle of the face image to be tested by the pre-trained face angle recognition model.
According to the method for face recognition based on multi-channel camera of claim 4, the training process of the face angle recognition model includes:

Acquiring a face image data set, where the face image data set includes face images of multiple people at different shooting angles;

Dividing the face image data set to obtain a training set and a test set;

Inputting the face image in the training set into a preset neural network model for training to obtain a face angle recognition model;

Input the face images in the test set into the face angle recognition model for testing, and obtain a test pass rate;

When the test pass rate is greater than or equal to the preset pass rate threshold, the training of the face angle recognition model ends; if the test pass rate is less than the preset pass rate threshold, increase the number of people in the training set For the number of face images, retrain the face angle recognition model.
The face recognition method based on multi-channel photography according to claim 1, wherein the recognition of the face image to be measured taken by the camera device corresponding to the target angle difference according to the face image in the target sample database comprises:

Extracting the facial features of the face image to be tested and the facial features of the face image in the target sample database;

Calculating the similarity between the face feature of the face image to be tested and the face feature of the face image in the target sample database;

Judging whether the similarity is greater than or equal to a preset similarity threshold;

If the judgment result is that the similarity is greater than or equal to the preset similarity threshold, output the face image corresponding to the similarity in the sample database as the face recognition result of the face image to be tested.
The face recognition method based on multi-channel camera according to claim 6, when it is determined that the similarity is less than the preset similarity threshold, the method further comprises:

Identifying, according to the first sample database, the first candidate face image of the face image to be tested taken by the first camera device;

Identifying, according to the second sample database, the second candidate face image of the face image to be tested captured by the second camera device;

Identifying, according to the third sample database, the third candidate face image of the face image to be tested taken by the third camera device;

Judging whether the first candidate face image, the second candidate face image, and the third candidate face image are the same person's face image;

If the judgment result is that the first candidate face image, the second candidate face image, and the third candidate face image are the same person's face image, output the first candidate face image as the final Face recognition result.
A face recognition device based on multi-channel camera, the face recognition device based on multi-channel camera includes:

A sample database establishment module for establishing a sample database corresponding to each camera device in the preset multi-channel camera device, and the photographing angle of the face image in the sample database is the same as the photographing angle of the corresponding camera device;

The face image acquisition module is used to acquire multiple faces to be measured in the face images to be measured taken by each camera at the same time;

A shooting angle recognition module, configured to recognize the shooting angle of the face image to be tested according to a pre-trained face angle recognition model;

An angle difference calculation module, configured to calculate the angle difference between the shooting angle of the face image to be measured and the shooting angle of each camera device;

The target difference determination module is configured to filter out the smallest angle difference from the angle difference as the target angle difference;

A sample database determining module, configured to determine the sample database of the camera device corresponding to the target angle difference as the target sample database;

The face recognition module is configured to recognize, according to the face image in the target sample database, the face image to be measured taken by the camera device corresponding to the target angle difference.
A terminal includes a processor configured to execute computer-readable instructions stored in a memory to implement the following steps:

Establishing a sample database for each camera device in the preset multi-channel camera device, and the shooting angle of the face image in the sample database is the same as the shooting angle of the corresponding camera device;

Acquire multiple face images to be measured of faces to be measured that are simultaneously photographed by each camera device;

Identifying the shooting angle of the face image to be tested according to the pre-trained face angle recognition model;

Calculating the angle difference between the shooting angle of the face image to be measured and the shooting angle of each camera device;

The smallest angle difference is selected from the angle difference as the target angle difference;

Determining the sample database of the camera device corresponding to the target angle difference as the target sample database;

Identify the face image to be measured taken by the camera device corresponding to the target angle difference according to the face image in the target sample database.
The terminal according to claim 9, wherein the preset multi-channel camera device includes a first camera device, a second camera device, and a third camera device that are co-centered, and the processor executes the computer Reading instructions to realize the establishment of a sample database for each of the preset multi-channel camera devices includes the following steps:

Taking the center as the center point, set the shooting angle of the first camera device to 0°, set the shooting angle of the second camera device to +90°, and set the shooting angle of the third camera device to -90 °;

Control the first camera device, the second camera device, and the third camera device to simultaneously capture a preset number of target testers for each of the target testers to obtain faces with different shooting angles image;

The face images of the preset number of target testers captured by the first camera device are used as the first sample database of the first camera device, and the second camera device A preset number of face images of the target tester are used as the second sample database of the second camera, and the preset number of face images of the target tester captured by the third camera are used as all The third sample database of the third camera device.
The terminal according to claim 10, when the processor executes the computer-readable instruction to realize the recognition of the shooting angle of the face image to be measured according to a pre-trained face angle recognition model, the following steps are included: step:

Obtain the face image to be tested taken by the first camera;

Inputting the face image to be tested into a pre-trained face angle recognition model;

Confirm the shooting angle of the face image to be tested by the pre-trained face angle recognition model.
The terminal according to claim 11, when the processor executes the computer-readable instructions to implement the training process of the face angle recognition model, the processor comprises the following steps:

Acquiring a face image data set, where the face image data set includes face images of multiple people at different shooting angles;

Dividing the face image data set to obtain a training set and a test set;

Inputting the face image in the training set into a preset neural network model for training to obtain a face angle recognition model;

Input the face images in the test set into the face angle recognition model for testing, and obtain a test pass rate;

When the test pass rate is greater than or equal to the preset pass rate threshold, the training of the face angle recognition model ends; if the test pass rate is less than the preset pass rate threshold, increase the number of people in the training set For the number of face images, retrain the face angle recognition model.
The terminal according to claim 9, wherein the processor is executing the computer-readable instructions to realize the recognition of the image taken by the camera device corresponding to the target angle difference according to the face image in the target sample database. When measuring a face image, it includes the following steps:

Extracting the facial features of the face image to be tested and the facial features of the face image in the target sample database;

Calculating the similarity between the face feature of the face image to be tested and the face feature of the face image in the target sample database;

Judging whether the similarity is greater than or equal to a preset similarity threshold;

If the judgment result is that the similarity is greater than or equal to the preset similarity threshold, output the face image corresponding to the similarity in the sample database as the face recognition result of the face image to be tested.
The terminal according to claim 13, when it is determined that the similarity is less than the preset similarity threshold, the processor executes computer-readable instructions to further implement the following steps:

Identifying, according to the first sample database, the first candidate face image of the face image to be tested taken by the first camera device;

Identifying, according to the second sample database, the second candidate face image of the face image to be tested captured by the second camera device;

Identifying, according to the third sample database, the third candidate face image of the face image to be tested taken by the third camera device;

Judging whether the first candidate face image, the second candidate face image, and the third candidate face image are the same person's face image;

If the judgment result is that the first candidate face image, the second candidate face image, and the third candidate face image are the same person's face image, output the first candidate face image as the final Face recognition result.
A non-volatile readable storage medium having computer readable instructions stored on the non-volatile readable storage medium, and when the computer readable instructions are executed by a processor, the following steps are implemented:

Establishing a sample database for each camera device in the preset multi-channel camera device, and the shooting angle of the face image in the sample database is the same as the shooting angle of the corresponding camera device;

Acquire multiple face images to be measured of faces to be measured that are simultaneously photographed by each camera device;

Identifying the shooting angle of the face image to be tested according to the pre-trained face angle recognition model;

Calculating the angle difference between the shooting angle of the face image to be measured and the shooting angle of each camera device;

The smallest angle difference is selected from the angle difference as the target angle difference;

Determining the sample database of the camera device corresponding to the target angle difference as the target sample database;

Identify the face image to be measured taken by the camera device corresponding to the target angle difference according to the face image in the target sample database.
The non-readable storage medium of claim 15, wherein the preset multi-channel camera device includes a first camera device, a second camera device, and a third camera device that are arranged in a common center, and the computer can When the read instruction is executed by the processor to realize the establishment of a sample database for each of the preset multi-channel camera devices, the following steps are included:

With the center as the center point, set the shooting angle of the first camera to 0°, set the shooting angle of the second camera to +90°, and set the shooting angle of the third camera to -90 °;

Control the first camera device, the second camera device, and the third camera device to simultaneously capture a preset number of target testers for each of the target testers to obtain faces with different shooting angles image;

The face images of the preset number of target testers captured by the first camera device are used as the first sample database of the first camera device, and the second camera device A preset number of face images of the target tester are used as the second sample database of the second camera, and the preset number of face images of the target tester captured by the third camera are used as all The third sample database of the third camera device.
The non-volatile readable storage medium according to claim 16, when the computer-readable instructions are executed by a processor, the photographing of the face image to be measured is recognized according to the pre-trained face angle recognition model Angle, including the following steps:

Obtain the face image to be tested taken by the first camera;

Inputting the face image to be tested into a pre-trained face angle recognition model;

Confirm the shooting angle of the face image to be tested by the pre-trained face angle recognition model.
According to the non-volatile readable storage medium of claim 17, when the computer readable instructions are executed by a processor, the training process of the face angle recognition model is realized, comprising the following steps:

Acquiring a face image data set, where the face image data set includes face images of multiple people at different shooting angles;

Dividing the face image data set to obtain a training set and a test set;

Inputting the face image in the training set into a preset neural network model for training to obtain a face angle recognition model;

Input the face images in the test set into the face angle recognition model for testing, and obtain a test pass rate;

When the test pass rate is greater than or equal to the preset pass rate threshold, the training of the face angle recognition model ends; if the test pass rate is less than the preset pass rate threshold, increase the number of people in the training set For the number of face images, retrain the face angle recognition model.
The non-volatile readable storage medium according to claim 15, when the computer-readable instructions are executed by a processor to realize the recognition corresponding to the target angle difference according to the face image in the target sample database The face image to be tested captured by the camera device includes the following steps:

Extracting the facial features of the face image to be tested and the facial features of the face image in the target sample database;

Calculating the similarity between the face feature of the face image to be tested and the face feature of the face image in the target sample database;

Judging whether the similarity is greater than or equal to a preset similarity threshold;

If the judgment result is that the similarity is greater than or equal to the preset similarity threshold, output the face image corresponding to the similarity in the sample database as the face recognition result of the face image to be tested.
According to the non-volatile readable storage medium of claim 19, when it is determined that the similarity is less than the preset similarity threshold, the computer-readable instructions are executed by the processor to further implement the following steps:

Identifying, according to the first sample database, the first candidate face image of the face image to be tested taken by the first camera device;

Identifying, according to the second sample database, the second candidate face image of the face image to be tested captured by the second camera device;

Identifying, according to the third sample database, the third candidate face image of the face image to be tested taken by the third camera device;

Judging whether the first candidate face image, the second candidate face image, and the third candidate face image are the same person's face image;

If the judgment result is that the first candidate face image, the second candidate face image, and the third candidate face image are the same person's face image, output the first candidate face image as the final Face recognition result.