CN114511914A - Face recognition method and device and terminal equipment - Google Patents

Abstract

The invention discloses a face recognition method, a face recognition device and terminal equipment, wherein the method comprises the following steps: acquiring face image information of a person to be identified; carrying out image feature extraction on the face image information to obtain face image feature data of the figure to be identified; inputting the facial image feature data into an identification degree model to calculate the facial identification degree, and obtaining an identification degree coefficient; if the identification degree coefficient is larger than a preset threshold value, acquiring an ultrasonic signal reflected by the face of the person to be identified; extracting the face ultrasonic characteristic data of the figure to be recognized from the acquired ultrasonic signals; and comparing the face image characteristic data and the face ultrasonic characteristic data with each characteristic data template in a preset characteristic database respectively to obtain the identity identification information of the figure to be identified. The invention avoids invalid operation, reduces operation pressure and improves the accuracy of face recognition.

Description

Face recognition method, device and terminal device

technical field

The present invention relates to the technical field of face recognition, in particular to a face recognition method, a recognition device and terminal number equipment.

Background technique

With the maturity of face recognition technology, more and more scenarios such as access control or checkpoints use face recognition technology to identify the identity information of people entering and leaving. Face recognition technology generally collects face information of a person in the form of collecting face picture information. However, in some specific scenarios, such as inside a processing factory or in a hospital, people often wear masks, causing part of the face to be blocked; or when there are a large number of people, the face is blocked.

At this time, if complete face information cannot be collected for recognition, the face recognition device will report an error or the recognition accuracy will be reduced, and computing power will be occupied and wasted. And under the influence of the outdoor environment, when the face is irradiated by strong light such as sunlight or lights, the collected face information of the person has the problem of overexposure, which will also cause the recognition accuracy of the face recognition device to decrease.

SUMMARY OF THE INVENTION

The purpose of the embodiments of the present invention is to provide a face recognition method, a recognition device and a terminal device, which can improve the accuracy of face recognition.

In order to solve the above technical problems, in a first aspect, an embodiment of the present invention provides a face recognition method, including:

Collect the face image information of the person to be recognized;

performing image feature extraction on the face image information to obtain face image feature data of the person to be identified;

Inputting the feature data of the face image into the recognition model to calculate the recognition degree of the face to obtain the recognition coefficient;

If the recognition coefficient is greater than a preset threshold, collect the ultrasonic signal reflected by the face of the person to be recognized;

Extract the face ultrasonic feature data of the person to be recognized from the collected ultrasonic signal;

Comparing the face image feature data and the face ultrasonic feature data with each feature data template in a preset feature database, respectively, to obtain the identification information of the person to be identified; the feature data template includes a specified user identity information and the image feature data and ultrasonic feature data corresponding to the face of the designated user.

As one of the preferred solutions, the feature data of the face image is input into the recognition degree model to calculate the recognition degree of the face, and the recognition degree coefficient is obtained, which specifically includes:

The face image feature data is divided into face regions by a classifier, and a collection of corresponding feature data in several preset face parts regions is obtained; the preset face parts include eyes, forehead, nose, mouth, cheeks and chin;

Inputting the corresponding feature data sets in the several preset face region regions into the neural network for weighting, and then calculating the recognition coefficient according to the weighted feature data sets.

As one of the preferred solutions, the neural network includes weight values corresponding to each preset face region;

The described inputting the corresponding feature data sets in the several preset facial region regions into the neural network for weighting, and then calculating the recognition coefficient according to the weighted feature data sets, specifically including:

Through the neural network, the corresponding feature data sets in each of the preset face position regions are convolved with the weight values corresponding to the preset face position regions to obtain each of the preset face positions The corresponding weighted feature data in the area;

After summing all the weighted feature data, normalization is performed to generate a recognition degree coefficient.

As one of the preferred solutions, before inputting the corresponding feature data sets in the several preset face regions into the neural network for weighting, and then calculating the recognition coefficient according to the weighted feature data sets, it also includes:

Obtain the collected historical face image information according to the received identification instruction; the identification instruction includes the target face position;

Extracting the feature data of the target face position in the historical face image information to obtain sample data;

The sample data is input into the neural network for iterative training, so as to increase the weight of the target face part in the neural network to obtain an updated neural network.

As one of the preferred solutions, after inputting the facial image feature data into the recognition degree model for face recognition degree calculation, and obtaining the recognition degree coefficient, the method further includes:

If the recognition degree coefficient is smaller than the preset threshold, the face image information of the person to be recognized is collected again.

As one of the preferred solutions, the face image information is visible light image information or infrared image information.

As one of the preferred solutions, the face image feature data and the face ultrasonic feature data are compared with each feature data template in a preset feature database, respectively, to obtain the identification information of the person to be identified. , including:

The face image feature data and the face ultrasonic feature data are respectively clustered with each feature data template in the preset feature database by a clustering method to determine the face image feature data and the face ultrasonic feature The designated user corresponding to the cluster center of the data is the target user;

The identity information of the target user is output.

As one of the preferred solutions, after clustering the face image feature data and the face ultrasonic feature data with each feature data template in the preset feature database by a clustering method, the method further includes:

If there is no designated user corresponding to the cluster center of the face image feature data and the face ultrasonic feature data, a result of recognition failure is output.

In a second aspect, an embodiment of the present invention further provides a face recognition device, including:

A face image information collection module, used to collect face image information of the person to be identified;

a first feature extraction module, configured to perform image feature extraction on the face image information to obtain the face image feature data of the person to be identified;

a coefficient calculation module, for inputting the facial image feature data into a recognition degree model to perform a face recognition degree calculation to obtain a recognition degree coefficient;

an ultrasonic signal collection module, configured to collect the ultrasonic signal reflected by the face of the person to be recognized if the recognition coefficient is greater than a preset threshold;

The second feature extraction module is used for extracting the face ultrasonic feature data of the person to be recognized from the collected ultrasonic signal;

an identification module, configured to compare the face image feature data and the face ultrasonic feature data with each feature data template in a preset feature database to obtain the identification information of the person to be identified; the feature The data template includes the identity information of the specified user and the image feature data and ultrasonic feature data corresponding to the face of the specified user.

Yet another embodiment of the present invention provides a terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, where the processor implements the first aspect when executing the computer program method.

Yet another embodiment of the present invention provides a computer-readable storage medium, where the computer-readable storage medium includes a stored computer program, wherein when the computer program runs, the device where the computer-readable storage medium is located is controlled to execute the above-mentioned first aspect Methods.

Compared with the prior art, the beneficial effects of the embodiments of the present invention lie in at least one of the following:

The face recognition method of the present invention obtains the face image feature data of the to-be-recognized person by collecting the face image information of the person to be recognized; extracting the image features from the face image information; The data is input into the recognition degree model to calculate the recognition degree of the face, and the recognition degree coefficient is obtained; if the recognition degree coefficient is greater than the preset threshold, the ultrasonic signal reflected by the face of the person to be recognized is collected; The image feature data is input into the recognition model to calculate the recognizability of the face, to know whether the degree of face occlusion of the person to be recognized currently affects the face recognition. If the recognition coefficient is greater than the preset threshold, it means that the degree of face occlusion does not affect the face. Only then can continue to perform face recognition, avoid invalid calculation, and reduce calculation pressure; extract the face ultrasonic feature data of the person to be recognized from the collected ultrasonic signal; combine the face image feature data with the The facial ultrasonic feature data is compared with each feature data template in the preset feature database to obtain the identification information of the person to be identified; the feature data template includes the identity information of the designated user and the designated user person. The image feature data and ultrasonic feature data corresponding to the face. By combining the face image information with the collected face ultrasonic signals, the collected face image information is supplemented with the missing face information due to problems such as overexposure or occlusion, so that the face recognition process can be carried out smoothly and accurately, thereby improving the accuracy of face recognition. Correspondingly, the present invention also provides a face recognition device and equipment.

Description of drawings

In order to illustrate the technical solutions of the present invention more clearly, the following will briefly introduce the accompanying drawings used in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention, which are common in the art. As far as technical personnel are concerned, other drawings can also be obtained based on these drawings without any creative effort.

1 is a schematic flowchart of an embodiment of a face recognition method of the present invention;

2 is a schematic structural diagram of an embodiment of a face recognition device of the present invention;

FIG. 3 is a schematic structural diagram of an embodiment of a terminal device of the present invention.

Detailed ways

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

In the description of the present invention, the terms "first", "second", "third", etc. are only used for description purposes, and should not be understood as indicating or implying relative importance or the number of indicated technical features. Thus, a feature defined as "first", "second", "third", etc., may expressly or implicitly include one or more of that feature. In the description of the present invention, unless otherwise specified, "plurality" means two or more.

In the description of the present invention, it should be noted that the terms "installed", "connected" and "connected" should be understood in a broad sense, unless otherwise expressly specified and limited, for example, it may be a fixed connection or a detachable connection Connection, or integral connection; can be mechanical connection, can also be electrical connection; can be directly connected, can also be indirectly connected through an intermediate medium, can be internal communication between two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood in specific situations.

In the description of the present invention, it should be noted that unless otherwise defined, all technical and scientific terms used in the present invention have the same meaning as commonly understood by those skilled in the art. The terms used in the description of the present invention are only for the purpose of describing specific embodiments, and are not intended to limit the present invention. Those of ordinary skill in the art can understand the specific meanings of the above terms in the present invention in specific situations.

Example 1

An embodiment of the present invention provides a face recognition method. Please refer to FIG. 1. FIG. 1 is a schematic flowchart of an embodiment of the face recognition method of the present invention. This embodiment can be applied to application scenarios such as access control or checkpoint where personnel identity information needs to be checked, and the method can be executed by a face recognition device, which can be a processor, a smart terminal, a tablet, or a PC. In this embodiment, the face recognition method may include steps S110-S160, and each step is as follows:

S110: collect face image information of the person to be recognized;

Since the face recognition method needs to recognize the face information of the person to be recognized, it is necessary to collect the face information of the person to be recognized. Specifically, the collection of face information is realized by collecting the face image of the person to be recognized, and the collected face image is the face image information. Optionally, the camera of the face recognition device or an external camera connected to the face recognition device can capture the face of the person to be recognized, so as to complete the collection of the face image information of the person to be recognized.

As one of the preferred solutions, the face image information is visible light image information or infrared image information. Since the external environment sometimes cannot provide sufficient light to meet the conditions for the camera to capture clear face images, an infrared camera can be used to capture clear face images under insufficient lighting or dark night conditions. Therefore, the collected face image information may include visible light image information or infrared image information.

S120: Perform image feature extraction on the face image information to obtain face image feature data of the person to be identified;

The information carried by each pixel in the collected face image information is the face feature information of the person to be identified. By extracting image features from the face image information, a face containing the face feature information of the person to be identified can be obtained. Image feature data. Specifically, the image feature extraction for the above-mentioned face image information can be realized by an image feature extraction method such as a statistical feature identification method, a geometric feature identification method, or a connection mechanism-based identification method.

S130: Input the facial image feature data into a recognition degree model to perform a face recognition degree calculation to obtain a recognition degree coefficient;

Because in some specific scenarios, such as inside a processing factory or a hospital, people often wear masks, causing part of the face to be blocked; or when there are a large number of people, the face is blocked. If the complete face information cannot be collected for recognition, the face recognition device will make an error or the face recognition will be inaccurate, and the face recognition device has already performed a complete recognition operation process when the wrong face recognition result is obtained. , resulting in the occupation and waste of computing power. In order to avoid invalid operations, the face recognition degree calculation can be performed by inputting the face image feature data into the pre-trained recognition model in advance, so as to know the face occlusion of the person to be recognized reflected by the currently collected face image feature data. Whether the degree affects face recognition; the recognition degree output by the recognition degree model reflects the degree of face occlusion of the person to be recognized. If the recognition degree coefficient is greater than the preset threshold, it means that the degree of face occlusion does not affect the face recognition, and then the face recognition can be continued to avoid invalid operations and reduce the computational pressure. Optionally, the recognition model may be a model composed of several neural network layers.

As one of the preferred solutions, the feature data of the face image is input into the recognition degree model for face recognition degree calculation, and the recognition degree coefficient is obtained, which specifically includes steps S210 and S220:

S210, dividing the face image feature data by a classifier to obtain a collection of corresponding feature data in several preset face position regions; the preset face positions include eyes, forehead, nose, mouth, cheeks and chin;

In detail, the feature data of the face image can be divided according to the face region by using a pre-trained classifier, so as to obtain a set of feature data corresponding to each preset face region. The preset face parts can be divided by the user, and the face can be divided into several regular face parts such as eyes, forehead, nose, mouth, cheeks and chin.

S220. Input the corresponding feature data sets in the preset facial region regions into the neural network for weighting, and then calculate and obtain the recognition coefficient according to the weighted feature data sets.

In detail, since the face recognition process generally compares and judges several key face parts with recognizable degrees in the face and the nearby areas, whether the collected face image feature data includes these several recognizable faces? The key face position of the degree and whether the number of features included are enough are the prerequisites for smooth and accurate face recognition. The above-mentioned key human face regions with recognizability may be such regions as facial features, cheeks, and chin.

In order to judge whether the collected face image feature data of the person to be recognized has the feature data that meets the preconditions for face recognition, it is possible to generate a weighted value based on a number of samples consisting of historical face image feature data with successful face recognition. neural network. Among them, the neural network generates the weight value corresponding to each face part region according to the importance of each face part in the face recognition process; the larger the weight ratio, the more the face part region in the face recognition process. The more effective it is.

By inputting the corresponding feature data sets in several preset face region regions into the neural network for weighting, it is known whether the degree of face occlusion of the person to be recognized currently affects face recognition. Specifically, when the face of the person to be identified is occluded, the collected face image feature data may lack relevant feature data of the occluded face position. Since the weight of the feature data of each face region in the trained neural network is configured according to the rule that the greater the effect on face recognition, the greater the weight value, when the missing occluded features in the collected face image feature data The relevant feature data of the face position has little effect in the face recognition process, and the weight value given is also small, so that the face with the recognition coefficient that is not occluded is finally calculated according to the weighted feature data collection. The influence of the relevant feature data of the face image; when the relevant feature data of the occluded face position missing in the collected face image feature data has a great effect in the face recognition process, the assigned weight value is also large, Therefore, the influence of the recognition coefficient obtained by finally calculating according to the weighted feature data collection by the relevant feature data of the occluded face parts becomes smaller. Therefore, according to the size of the recognition coefficient obtained by calculation, it can be known whether the degree of face occlusion of the person to be recognized currently affects the face recognition.

As one of the preferred solutions, the neural network includes weight values corresponding to each preset face region; in detail, the neural network includes weight values corresponding to each preset face region, and the preset face region The greater the role of the region in the face recognition process, the greater the corresponding weight value.

The described inputting the corresponding feature data sets in the several preset facial region regions into the neural network for weighting, and then calculating the recognition coefficient according to the weighted feature data sets, specifically including:

Through the neural network, the corresponding feature data sets in each of the preset face position regions are convolved with the weight values corresponding to the preset face position regions to obtain each of the preset face positions The corresponding weighted feature data in the area;

Specifically, the weight values corresponding to each preset face region included in the neural network may be set as convolution kernels in several neural network layers. After the corresponding feature data sets in several preset face part regions are input into the neural network for weighting, the corresponding feature data sets in each of the preset face part regions and the corresponding weights of the preset face part regions The values are convolved to obtain the corresponding weighted feature data in each of the preset face regions.

After summing all the weighted feature data, normalization is performed to generate a recognition degree coefficient.

In order to facilitate comparison and judgment, all the weighted feature data can be summed and then normalized to generate an identification coefficient.

Because in some specific scenarios, such as inside a processing factory or a hospital, people often wear masks, and the person to be identified cannot easily take off the mask for face recognition. At this time, it is necessary to adjust the face recognition part according to the usage scenario, and the recognition focus of the face recognition device needs to be adjusted to focus on recognizing the upper part of the face area of the person to be recognized. Specifically, the method of adjusting the recognition focus of the face recognition device to focus on recognizing the upper half of the face region of the person to be recognized may be to adjust the weight value corresponding to the upper half of the face region in the neural network in the recognition model.

As one of the preferred solutions, before inputting the corresponding feature data sets in the several preset face regions into the neural network for weighting, and then calculating the recognition coefficient according to the weighted feature data sets, it also includes:

Obtain the collected historical face image information according to the received identification instruction; the identification instruction includes the target face position;

In detail, the recognition instruction is generated according to the face position of which the weight value is adjusted, so as to obtain the collected historical face image information according to the received recognition instruction. Specifically, the identification instruction includes the face position of the target person. To illustrate in detail, if the recognition focus of the face recognition device needs to be adjusted to focus on recognizing the upper part of the face area of the person to be recognized, the upper part of the face area can be set as the target face position.

Extracting the feature data of the target face position in the historical face image information to obtain sample data;

The sample data is input into the neural network for iterative training, so as to increase the weight of the target face part in the neural network to obtain an updated neural network.

By inputting the sample data related to the target face position into the neural network for iterative training, to increase the weight of the target face position in the neural network, and to obtain an updated neural network, the human The recognition focus of the face recognition device is adjusted to focus on recognizing the target face region of the person to be recognized, so as to be more flexibly adapted to various face recognition usage scenarios.

S140: If the recognition coefficient is greater than a preset threshold, collect the ultrasonic signal reflected by the face of the person to be recognized;

In detail, the preset threshold is calculated by inputting the most recognizable face image samples into the trained recognition model. If the recognition coefficient is greater than the preset threshold, it means that the face information contained in the face image feature data is effective enough for face recognition, and face recognition can be continued, avoiding invalid operations and reducing computational pressure.

Since the collected face image information may also have problems such as overexposure or occlusion, resulting in the lack of face information, when it is judged that the recognition coefficient is greater than the preset threshold, the ultrasonic signal reflected by the face of the person to be identified can be collected. The collected face image information is supplemented with missing face information due to problems such as overexposure or occlusion, so that the face recognition process can be carried out smoothly and accurately, and the accuracy of face recognition is improved. Specifically, the ultrasonic signal transmitter connected to the face recognition device can be controlled to transmit the ultrasonic signal to the face of the person to be recognized, and then the reflection from the face of the person to be recognized collected by the ultrasonic signal sensor connected to the face recognition device can be received. ultrasonic signal.

As one of the preferred solutions, after inputting the facial image feature data into the recognition degree model for face recognition degree calculation, and obtaining the recognition degree coefficient, the method further includes:

If the recognition degree coefficient is smaller than the preset threshold, the face image information of the person to be recognized is collected again. Specifically, if the recognition degree coefficient is smaller than the preset threshold, it means that the face information contained in the face image feature data is too small to perform effective face recognition. Therefore, it is necessary to re-collect the face image information of the person to be recognized until the recognition degree coefficient is greater than the preset threshold.

S150: Extract the facial ultrasonic feature data of the person to be recognized from the collected ultrasonic signal;

The information carried by each signal in the collected ultrasonic signal is the facial feature information of the person to be recognized. By performing signal feature extraction on the ultrasonic signal, the facial ultrasonic feature data including the facial feature information of the person to be recognized can be obtained.

S160: Compare the face image feature data and the face ultrasonic feature data with each feature data template in a preset feature database, to obtain the identification information of the person to be identified; the feature data template includes The identity information of the specified user and the image feature data and ultrasonic feature data corresponding to the face of the specified user.

In order to identify a specific user or a recorded member, the user to be identified can be preset as a designated user, and the identity information of the designated user and the image feature data and ultrasonic feature data corresponding to the designated user's face can be collected. The collected information corresponding to each designated user is stored in the database for each feature data template. During data comparison, each feature data template is called from the database, and the collected face image feature data and the face ultrasonic feature data are compared respectively, so as to realize the identification information of the person to be identified.

As one of the preferred solutions, the face image feature data and the face ultrasonic feature data are compared with each feature data template in a preset feature database, respectively, to obtain the identification information of the person to be identified. , including:

The face image feature data and the face ultrasonic feature data are respectively clustered with each feature data template in the preset feature database by a clustering method to determine the face image feature data and the face ultrasonic feature The designated user corresponding to the cluster center of the data is the target user;

The identity information of the target user is output.

As one of the preferred solutions, after clustering the face image feature data and the face ultrasonic feature data with each feature data template in the preset feature database by a clustering method, the method further includes:

If there is no designated user corresponding to the cluster center of the face image feature data and the face ultrasonic feature data, a result of recognition failure is output.

To sum up, in a face recognition method provided by this embodiment, the face image feature data of the to-be-recognized person is obtained by collecting the face image information of the person to be recognized, and performing image feature extraction on the face image information. ; Input the feature data of the facial image into the recognition model to calculate the recognizable degree of the human face, and obtain the recognition degree coefficient; Ultrasonic signal; by inputting the facial image feature data into the recognition model in advance to calculate the recognition degree of the face, to know whether the degree of face occlusion of the person to be recognized currently affects the recognition of the face, if the recognition degree coefficient is greater than the preset threshold, it means that The degree of face occlusion does not affect the face recognition, so that the face recognition can be continued to avoid invalid calculation and reduce the calculation pressure; extract the face ultrasonic feature data of the person to be recognized from the collected ultrasonic signal; The face image feature data and the face ultrasonic feature data are compared with each feature data template in the preset feature database respectively to obtain the identification information of the person to be identified; the feature data template includes the specified user's data. Identity information and image feature data and ultrasonic feature data corresponding to the face of the designated user. By combining the face image information with the collected face ultrasonic signals, the collected face image information is supplemented with the missing face information due to problems such as overexposure or occlusion, so that the face recognition process can be carried out smoothly and accurately, thereby improving the accuracy of face recognition. Correspondingly, the present invention also provides a face recognition device and equipment.

Embodiment 2

On the basis of Embodiment 1, as shown in FIG. 2 , an embodiment of the present invention further provides a face recognition device 2, including:

A face image information collection module 201, configured to collect face image information of the person to be identified;

The first feature extraction module 202 is configured to perform image feature extraction on the face image information to obtain the face image feature data of the person to be identified;

The coefficient calculation module 203 is used for inputting the feature data of the face image into the recognition degree model to calculate the recognition degree of the face, and obtain the recognition degree coefficient;

The ultrasonic signal collection module 204 is configured to collect the ultrasonic signal reflected by the face of the person to be recognized if the recognition coefficient is greater than a preset threshold;

The second feature extraction module 205 is configured to extract the face ultrasonic feature data of the person to be recognized from the collected ultrasonic signal;

The identification module 206 is configured to compare the face image feature data and the face ultrasonic feature data with each feature data template in a preset feature database, to obtain the identification information of the person to be identified; the The feature data template includes the identity information of the specified user and the image feature data and ultrasonic feature data corresponding to the face of the specified user.

As one of the preferred solutions, the coefficient calculation module 203 further includes:

A classification unit, configured to divide the face image feature data by a classifier, and obtain a collection of corresponding feature data in a number of preset face position regions; for the preset person, the face positions include eyes, forehead, nose, mouth, cheeks and chin;

The coefficient calculation unit is used for inputting the corresponding feature data sets in the preset facial region regions into the neural network for weighting, and then calculating the recognition coefficient according to the weighted feature data sets.

As one of the preferred solutions, the neural network includes weight values corresponding to each preset face region;

Coefficient calculation unit, specifically for:

Through the neural network, the corresponding feature data sets in each of the preset face position regions are convolved with the weight values corresponding to the preset face position regions to obtain each of the preset face positions The corresponding weighted feature data in the area;

After summing all the weighted feature data, normalization is performed to generate a recognition degree coefficient.

As one of the preferred solutions, before inputting the corresponding feature data sets in the several preset face regions into the neural network for weighting, and then calculating the recognition coefficient according to the weighted feature data sets, it also includes:

Obtain the collected historical face image information according to the received identification instruction; the identification instruction includes the target face position;

Extracting the feature data of the target face position in the historical face image information to obtain sample data;

The sample data is input into the neural network for iterative training, so as to increase the weight of the target face part in the neural network to obtain an updated neural network.

As one of the preferred solutions, the device further includes:

The information re-collection module is configured to re-collect the face image information of the person to be recognized if the recognition degree coefficient is smaller than the preset threshold.

As one of the preferred solutions, the face image information is visible light image information or infrared image information.

As one of the preferred solutions, the identification module 206 specifically includes:

The clustering unit is used for clustering the face image feature data and the face ultrasonic feature data with each feature data template in the preset feature database by a clustering method, and determining the face image feature data and the face image feature data. The designated user corresponding to the cluster center of the ultrasonic facial feature data is the target user;

The identity information output unit is used for outputting the identity information of the target user.

As one of the preferred solutions, the identification module 206 further includes:

A recognition failure result output unit, configured to output a recognition failure result if there is no designated user corresponding to the cluster center of the face image feature data and the face ultrasonic feature data.

To sum up, a face recognition device provided in this embodiment obtains the face image feature data of the person to be recognized by collecting the face image information of the person to be recognized, and extracting the image features from the face image information. ; Input the feature data of the facial image into the recognition model to calculate the recognizable degree of the human face, and obtain the recognition degree coefficient; Ultrasonic signal; by inputting the facial image feature data into the recognition model in advance to calculate the recognition degree of the face, to know whether the degree of face occlusion of the person to be recognized currently affects the recognition of the face, if the recognition degree coefficient is greater than the preset threshold, it means that The degree of face occlusion does not affect the face recognition, so that the face recognition can be continued to avoid invalid calculation and reduce the calculation pressure; extract the face ultrasonic feature data of the person to be recognized from the collected ultrasonic signal; The face image feature data and the face ultrasonic feature data are compared with each feature data template in the preset feature database respectively to obtain the identification information of the person to be identified; the feature data template includes the specified user's Identity information and image feature data and ultrasonic feature data corresponding to the face of the designated user. By combining the face image information with the collected face ultrasonic signals, the collected face image information is supplemented with the missing face information due to problems such as overexposure or occlusion, so that the face recognition process can be carried out smoothly and accurately, thereby improving the accuracy of face recognition. Correspondingly, the present invention also provides a face recognition device and equipment.

Embodiment 3

Referring to FIG. 3 , an embodiment of the present invention further provides a terminal device, that is, a computer terminal device, including one or more processors and a memory. A memory is coupled to the processor for storing one or more programs; when the one or more programs are executed by the one or more processors, the one or more processors implement any of the above The face recognition method according to an embodiment.

The processor is used to control the overall operation of the computer terminal device, so as to complete all or part of the steps of the above-mentioned face recognition-based terminal device control apparatus 100 . The memory is used to store various types of data to support operation at the computer terminal device, such data may include, for example, instructions for any application or method for operation on the computer terminal device, as well as application-related data. The memory can be implemented by any type of volatile or nonvolatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (Electrically Erasable) Programmable Read-Only Memory (EEPROM for short), Erasable Programmable Read-Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (Read- Only Memory, referred to as ROM), magnetic memory, flash memory, magnetic disk or optical disk.

In an exemplary embodiment, the computer terminal device may be implemented by one or more application specific integrated circuits (Application Specific Integrated Circuit, AS1C for short), Digital Signal Processor (Digital Signal Processor, DSP for short), Digital Signal Processing Device (Digital Signal Processing) Processing Device, referred to as DSPD), Programmable Logic Device (Programmable Logic Device, referred to as PLD), Field Programmable Gate Array (Field Programmable Gate Array, referred to as FPGA), controller, microcontroller, microprocessor or other electronic components. , which is used to execute the above-mentioned terminal device control method based on face recognition, and achieve the same technical effect as the above-mentioned method.

In another exemplary embodiment, a computer-readable storage medium including program instructions is also provided, and when the program instructions are executed by a processor, the terminal device control based on face recognition as described in any one of the foregoing embodiments is implemented steps of the method. For example, the computer-readable storage medium can be the above-mentioned memory including program instructions, and the above-mentioned program instructions can be executed by the processor of the computer terminal device to complete the above-mentioned face recognition-based terminal device control method, and achieve the same technology as the above method. Effect.

The above are the preferred embodiments of the present invention. It should be pointed out that for those skilled in the art, without departing from the principles of the present invention, several improvements and modifications can be made, and these improvements and modifications may also be regarded as It is the protection scope of the present invention.

Claims (10)

1. a face recognition method, is characterized in that, described method comprises: Collect the face image information of the person to be recognized; performing image feature extraction on the face image information to obtain face image feature data of the person to be identified; Inputting the feature data of the face image into the recognition model to calculate the recognition degree of the face to obtain the recognition coefficient; If the recognition coefficient is greater than a preset threshold, collect the ultrasonic signal reflected by the face of the person to be recognized; Extract the face ultrasonic feature data of the person to be recognized from the collected ultrasonic signal; Comparing the face image feature data and the face ultrasonic feature data with each feature data template in a preset feature database, respectively, to obtain the identification information of the person to be identified; the feature data template includes a specified user identity information and the image feature data and ultrasonic feature data corresponding to the face of the designated user. 2. face recognition method according to claim 1, is characterized in that, described with described face image feature data input recognition degree model to carry out face recognition degree calculation, obtain recognition degree coefficient, specifically comprise: The face image feature data is divided into face regions by a classifier, and a collection of corresponding feature data in several preset face parts regions is obtained; the preset face parts include eyes, forehead, nose, mouth, cheeks and chin; Inputting the corresponding feature data sets in the several preset face region regions into the neural network for weighting, and then calculating the recognition coefficient according to the weighted feature data sets. 3. face recognition method according to claim 2, is characterized in that, described neural network comprises the corresponding weight value of each preset face position area; The described inputting the corresponding feature data sets in the several preset facial region regions into the neural network for weighting, and then calculating the recognition coefficient according to the weighted feature data sets, specifically including: Through the neural network, the corresponding feature data sets in each of the preset face position regions are convolved with the weight values corresponding to the preset face position regions to obtain each of the preset face positions The corresponding weighted feature data in the area; After summing all the weighted feature data, normalization is performed to generate a recognition degree coefficient. 4. face recognition method according to claim 3, is characterized in that, input neural network weighting by the corresponding characteristic data collection in described several preset face position area, then according to the characteristic data after the weighting Before the identification coefficient is obtained by the collection calculation, it also includes: Obtain the collected historical face image information according to the received identification instruction; the identification instruction includes the target face position; Extracting the feature data of the target face position in the historical face image information to obtain sample data; The sample data is input into the neural network for iterative training, so as to increase the weight of the target face part in the neural network to obtain an updated neural network. 5. face recognition method according to claim 1, is characterized in that, after described face image feature data is input recognition degree model to carry out face recognizable degree calculation, after obtaining recognition degree coefficient, also comprises: If the recognition degree coefficient is smaller than the preset threshold, the face image information of the person to be recognized is collected again. 6 . The face recognition method according to claim 1 , wherein the face image information is visible light image information or infrared image information. 7 . 7. The face recognition method according to claim 1, wherein the described face image feature data and the face ultrasonic feature data are compared with each feature data template in the preset feature database respectively , to obtain the identification information of the person to be identified, specifically including: The face image feature data and the face ultrasonic feature data are respectively clustered with each feature data template in the preset feature database by a clustering method to determine the face image feature data and the face ultrasonic feature The designated user corresponding to the cluster center of the data is the target user; The identity information of the target user is output. 8. face recognition method according to claim 7, is characterized in that, by clustering method described face image characteristic data and described face ultrasonic wave characteristic data respectively with each characteristic data template in preset characteristic database After clustering, it also includes: If there is no designated user corresponding to the cluster center of the face image feature data and the face ultrasonic feature data, a result of recognition failure is output. 9. A face recognition device, comprising: A face image information collection module, used to collect face image information of the person to be identified; a first feature extraction module, configured to perform image feature extraction on the face image information to obtain the face image feature data of the person to be identified; a coefficient calculation module, for inputting the facial image feature data into a recognition degree model to perform a face recognition degree calculation to obtain a recognition degree coefficient; an ultrasonic signal collection module, configured to collect the ultrasonic signal reflected by the face of the person to be recognized if the recognition coefficient is greater than a preset threshold; The second feature extraction module is used for extracting the face ultrasonic feature data of the person to be recognized from the collected ultrasonic signal; an identification module, configured to compare the face image feature data and the face ultrasonic feature data with each feature data template in a preset feature database to obtain the identification information of the person to be identified; the feature The data template includes the identity information of the specified user and the image feature data and ultrasonic feature data corresponding to the face of the specified user. 10. A terminal device, characterized by comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, when the processor executes the computer program, the computer program as claimed in the claim is implemented The face recognition method described in any one of requirements 1 to 8 is required.

Images