CN108446660A - The method and apparatus of facial image for identification - Google Patents

Abstract

The embodiment of the present application discloses a method and device for recognizing a human face image. A specific implementation of the method includes: respectively acquiring image sequences of key parts from two face images to obtain two sets of image sequences of key parts; respectively inputting the two sets of image sequences of key parts into the convolutional layer of the convolutional neural network to obtain two sets of image sequences of key parts. Group eigenvectors; read a corresponding pair of eigenvectors from two groups of eigenvectors, and calculate the similarity of the pair of eigenvectors; based on the similarity of the pair of eigenvectors, determine whether the faces in the two face images are The same face. The method can improve the efficiency of identifying whether the faces in multiple face images are the same face.

Description

Method and device for recognizing human face images

technical field

The present application relates to the field of computer technology, in particular to the field of computer network technology, and in particular to a method and device for recognizing human face images.

Background technique

Face recognition is a biometric technology for identification based on human facial feature information. Use a camera or camera to collect images or video streams containing faces, and automatically detect and track faces in the images, and then perform a series of face image preprocessing, image feature extraction, matching and recognition on the detected faces Related technologies are also commonly called portrait recognition or facial recognition.

In the current face recognition method, when identifying whether the faces in the two face images are the same face, the two face images will be processed separately. The processing process includes: locating multiple (such as 88 or 68) key points on the human face during graphic feature extraction, and then extracting human face features by inputting these key points into the machine learning model. Afterwards, if the weighted sum of the similarities of the extracted face features is greater than a predetermined threshold, it is determined that the faces in the two face images are the same face.

Contents of the invention

Embodiments of the present application propose a method and device for recognizing a human face image.

In the first aspect, the embodiment of the present application provides a method for recognizing human face images, including: respectively obtaining key part image sequences from two human face images to obtain two sets of key part image sequences; The image sequence is respectively input into the convolutional layer of the convolutional neural network to obtain two sets of feature vectors; read the corresponding pair of feature vectors from the two sets of feature vectors, and calculate the similarity of the pair of feature vectors; based on the pair of feature vectors Similarity, to determine whether the faces in two face images are the same face.

In some embodiments, based on the similarity of the pair of feature vectors, determining whether the faces in the two face images are the same face includes: in response to the similarity of the pair of feature vectors being less than a preset similarity threshold, determining The faces in the two face images are not the same face.

In some embodiments, based on the similarity of the pair of feature vectors, determining whether the faces in the two face images are the same face further includes: in response to the similarity of the pair of feature vectors being greater than a preset similarity threshold and The pair of eigenvectors is not the last pair of eigenvectors in the two groups of eigenvectors, jump to read the corresponding pair of eigenvectors from the two groups of eigenvectors, and calculate the similarity of the pair of eigenvectors; based on the pair of eigenvectors to determine whether the faces in the two face images are the same face.

In some embodiments, based on the similarity of the pair of feature vectors, determining whether the faces in the two face images are the same face further includes: in response to the similarity of the pair of feature vectors being greater than a preset similarity threshold and The pair of feature vectors is the last pair of feature vectors in the two groups of feature vectors, and it is determined that the faces in the two face images are the same face.

In some embodiments, the key parts in the key part image sequence include: left eye, right eye, nose, left corner of mouth and right corner of mouth.

In some embodiments, calculating the similarity of the pair of eigenvectors includes calculating any one of the following of the pair of eigenvectors: Pearson correlation coefficient, Euclidean distance, cosine similarity, Manhattan distance and Chebyshev distance.

In the second aspect, the embodiment of the present application provides a device for recognizing human face images, including: an image sequence acquisition unit, which is used to obtain key part image sequences from two human face images respectively, and obtain two sets of key part images sequence; the feature vector acquisition unit is used to input two sets of key part image sequences into the convolutional layer of the convolutional neural network respectively to obtain two sets of feature vectors; the similarity calculation unit is used to read the corresponding A pair of eigenvectors, calculating the similarity of the pair of eigenvectors; the same face determination unit is used to determine whether the faces in the two face images are the same face based on the similarity of the pair of eigenvectors.

In some embodiments, the same face determining unit is further configured to: determine that the faces in the two face images are not the same face in response to the similarity between the pair of feature vectors being less than a preset similarity threshold.

In some embodiments, the same face determination unit is further configured to: in response to the similarity of the pair of feature vectors being greater than a preset similarity threshold and the pair of feature vectors being not the last pair of feature vectors in the two groups of feature vectors, jumping To read the corresponding pair of feature vectors from the two sets of feature vectors, calculate the similarity of the pair of feature vectors; based on the similarity of the pair of feature vectors, determine whether the faces in the two face images are the same person Face.

In some embodiments, the same face determining unit is further used for: in response to the similarity of the pair of feature vectors being greater than a preset similarity threshold and the pair of feature vectors being the last pair of feature vectors in the two groups of feature vectors, determine two The faces in the face images are the same face.

In some embodiments, the key parts in the image sequence acquisition unit The key parts in the image sequence include: left eye, right eye, nose, left corner of mouth and right corner of mouth.

In some embodiments, calculating the similarity of the pair of feature vectors in the similarity calculation unit includes calculating any of the following of the pair of feature vectors: Pearson correlation coefficient, Euclidean distance, cosine similarity, Manhattan distance and Chebyshev distance.

In a third aspect, the embodiment of the present application provides a device, including: one or more processors; a storage device for storing one or more programs; when one or more programs are executed by one or more processors, Make one or more processors realize the method for recognizing a human face image as described above.

In a fourth aspect, the embodiment of the present application provides a computer-readable medium, on which a computer program is stored, and when the program is executed by a processor, the method for recognizing a face image as described above is implemented.

In the method and device for recognizing human face images provided in the embodiments of the present application, firstly, the image sequences of key parts are respectively obtained from two face images to obtain two sets of image sequences of key parts; after that, the two sets of image sequences of key parts are separately Enter the convolutional layer of the convolutional neural network to obtain two sets of feature vectors; then, read the corresponding pair of feature vectors from the two sets of feature vectors, and calculate the similarity of the pair of feature vectors; finally, based on the pair of feature vectors to determine whether the faces in the two face images are the same face. In this process, based on whether the similarity of the read pair of feature vectors is less than the similarity threshold, it can be quickly determined whether the key part images corresponding to the read pair of feature vectors have a high similarity, so as to determine the two faces Whether the human faces in the image are the same human face, therefore, the method for identifying the human face image improves the efficiency of identifying the same human face in the human face image.

Description of drawings

Other characteristics, objects and advantages of the present application will become more apparent by reading the detailed description of non-limiting embodiments made with reference to the following drawings:

FIG. 1 shows an exemplary system architecture to which the embodiment of the present application can be applied;

Fig. 2 is a schematic flowchart of an embodiment of a method for recognizing a face image according to an embodiment of the present application;

FIG. 3 is an exemplary application scenario of an embodiment of a method for recognizing a face image according to an embodiment of the present application;

FIG. 4 is a schematic flowchart of another embodiment of a method for recognizing a face image according to an embodiment of the present application;

FIG. 5 is an exemplary structural diagram of an embodiment of a device for recognizing a face image according to an embodiment of the present application;

Fig. 6 is a schematic structural diagram of a computer system suitable for implementing a terminal device or a server according to an embodiment of the present application.

Detailed ways

The application will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain related inventions, rather than to limit the invention. It should also be noted that, for the convenience of description, only the parts related to the related invention are shown in the drawings.

It should be noted that, in the case of no conflict, the embodiments in the present application and the features in the embodiments can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and embodiments.

Please refer to FIG. 1 , which shows an exemplary system architecture 100 to which embodiments of the method for recognizing a human face image or the device for recognizing a human face image of the present application can be applied.

As shown in FIG. 1 , the system architecture 100 may include terminal devices 101 , 102 , 103 , a network 104 and servers 105 , 106 . The network 104 serves as a medium for providing communication links between the terminal devices 101 , 102 , 103 and the servers 105 , 106 . Network 104 may include various connection types, such as wires, wireless communication links, or fiber optic cables, among others.

The user 110 can use the terminal device 101 , 102 , 103 to interact with the server 105 , 106 through the network 104 to receive or send messages and the like. Various communication client applications can be installed on the terminal devices 101, 102, 103, such as shooting applications, search engine applications, shopping applications, instant messaging tools, email clients, social platform software, video playback applications, etc.

The terminal devices 101, 102, and 103 may be hardware or software. When the terminal devices 101, 102, 103 are hardware, they can be various electronic devices with display screens, including but not limited to smart phones, tablet computers, e-book readers, MP3 players (Moving Picture Experts Group Audio Layer III, Moving Picture Experts Compression Standard Audio Layer 3), MP4 (Moving Picture Experts Group Audio Layer IV, Moving Picture Experts Compression Standard Audio Layer 4) players, laptops and desktop computers, etc. When the terminal devices 101, 102, 103 are software, they can be installed in the electronic devices listed above. It can be implemented as a plurality of software or software modules (for example, to provide distributed services), or as a single software or software module. No specific limitation is made here.

The servers 105, 106 may be servers that provide various services. For example, the servers 105 , 106 may be background servers that provide support for the terminal devices 101 , 102 , 103 . The background server can analyze, store or calculate the data submitted by the terminal, and push the obtained data processing results to the terminal device.

Usually, the method for recognizing a human face image provided by the embodiment of the present application is generally executed by the servers 105 and 106 , and correspondingly, the device for recognizing a human face image is generally set in the servers 105 and 106 .

It should be understood that the numbers of terminal devices, networks and servers in Fig. 1 are only illustrative. According to the implementation needs, there can be any number of terminal devices, networks and servers.

Further referring to FIG. 2 , FIG. 2 shows a schematic flowchart of an embodiment of a method for recognizing a face image according to an embodiment of the present application.

As shown in FIG. 2, a method 200 for recognizing a face image includes:

In step 210, image sequences of key parts are respectively acquired from the two face images to obtain two sets of image sequences of key parts.

In this embodiment, the execution subject of the method for recognizing a face image (for example, the servers 105, 106 shown in FIG. 1 ) may receive the The submitted original collected images, such as static images and dynamic images of different positions and expressions.

At this time, the original captured image cannot be used directly due to the limitation of various conditions and random interference, and image preprocessing such as grayscale correction and noise filtering must be performed on it in the early stage of image processing. For face images, the preprocessing process mainly includes light compensation, grayscale transformation, histogram equalization, normalization, geometric correction, filtering and sharpening of face images.

After preprocessing the original collected images, detection algorithms can be used to detect key parts of the face image, such as eyebrows, eyes, nose, mouth, ears, etc. Afterwards, the above-mentioned executive body can use the labeling algorithm to center on the key points of key parts (such as the head of the eyebrow, the tail of the eyebrow, the inner corner of the eye, the outer corner of the eye, the center of the pupil, the tip of the nose, the corner of the left mouth, the corner of the right mouth, the vertex of the left ear, the vertex of the right ear, etc.) 1. Mark these key parts with a rectangular frame of predetermined size, and use the image in the rectangular frame as the image of the key part. According to the order in which the images of the key parts are obtained, an image sequence of the key parts can be obtained. It should be understood that, among the images of key parts marked with rectangular frames of a predetermined size, the images in the rectangular frames may partially overlap.

Alternatively or additionally, after preprocessing the original captured image, a detection algorithm may also be used to detect key parts of the face image, such as eyebrows, eyes, nose, mouth, ears, etc. Afterwards, the above-mentioned executive body may use a labeling algorithm to mark the contours of these key parts, and use the images in the contours as images of key parts. According to the order in which the images of the key parts are obtained, an image sequence of the key parts can be obtained. For example, if the order of marking the contours of key parts is eyes, nose and mouth, then the obtained key part image sequence will include eye images, nose images and mouth images in sequence.

Specifically, in some optional implementation manners, the key parts in the key part image sequence may include: the left eye, the right eye, the nose, the left corner of the mouth, and the right corner of the mouth.

In step 220, two sets of image sequences of key parts are respectively input into the convolution layer of the convolutional neural network to obtain two sets of feature vectors.

In this embodiment, the aforementioned image sequences of key parts obtained from the two face images are input into the convolutional layer of the pre-trained neural network, and two sets of feature vectors respectively corresponding to the image sequences of key parts can be obtained. For example, if a set of key part image sequences includes eye images, nose images and mouth images, then the obtained set of feature vectors corresponding to the set of key part image sequences may include: feature vectors corresponding to eye images, feature vectors corresponding to nose images The feature vector corresponding to the mouth image.

The convolution layer here has a convolution kernel with a size of N×N (N is a natural number greater than 1, such as 2 or 3, etc.). Assuming that each convolutional layer performs a convolution operation with a step size of 1, it means that the convolution kernel moves to the right one pixel at a time (when moving to the boundary, it returns to the leftmost end and moves down by one unit). The weight of the convolution kernel is learned, and the weight of the convolution kernel will not change during the convolution process. There is a weight in each unit of the convolution kernel, that is, there are N ² weights in a convolution kernel. During the movement of the convolution kernel, the pixels on the picture can be multiplied by the corresponding weights of the convolution kernel, and finally all the products are added to obtain an output. Here, multiple convolutional layers can be used to obtain deeper feature vectors of each key part, that is, to extract more discriminative features of each key part.

In step 230, a corresponding pair of feature vectors is read from the two groups of feature vectors, and the similarity of the pair of feature vectors is calculated.

In this embodiment, a pair of corresponding feature vectors may be read from the two groups of feature vectors according to the arrangement order of the feature vectors in the two groups of feature vectors, and then the similarity of the pair of feature vectors may be calculated. Here, the method for calculating the similarity may be a method for calculating the similarity in the existing technology or a technology developed in the future, which is not limited in the present application. For example, the method of computing similarity can be done by any of the following: Pearson correlation coefficient, Euclidean distance, cosine similarity, Manhattan distance, and Chebyshev distance, etc.

In step 240, based on the similarity of the pair of feature vectors, it is determined whether the faces in the two face images are the same face.

In this embodiment, according to the read similarity of the pair of feature vectors, it can be determined whether the faces in the two face images are the same face. For example, when determining whether the faces in two face images are the same face according to the similarity, if the similarity of the current pair of feature vectors is lower than the similarity threshold, it indicates that a pair of key parts corresponding to the pair of feature vectors If the similarity of the images is low, it can be determined that the faces in the two face images are not the same face. If the similarity of the current pair of feature vectors is not lower than the similarity threshold, it indicates that the similarity of a pair of key part images corresponding to the pair of feature vectors is relatively high, so based on this result, it is possible to determine the Whether the faces are the same face.

The method for identifying human face images provided by the above-mentioned embodiments of the present application can obtain the key part image sequences of two human face images, and then obtain a set of feature vectors from each group of key part image sequences, and then obtain the two sets of feature vectors. A corresponding pair of eigenvectors, and calculate the similarity of the pair of eigenphasors, and then determine whether the faces in the two face images are the same face based on the similarity. In this process, based on whether the similarity of the read pair of feature vectors is less than the similarity threshold, it can be quickly determined whether the faces in the two face images are the same face. The image method improves the efficiency of recognizing the same face in face images.

Further, please refer to FIG. 3 , which shows an exemplary application scenario of the method for recognizing a face image according to an embodiment of the present application.

As shown in FIG. 3, the method for recognizing a human face image runs in an electronic device 320, and the method includes:

Firstly, the key part image sequence 302 is obtained from the face image 301 ; the key part image sequence 304 is obtained from the face image 303 .

After that, the key part image sequence 302 and the key part image sequence 304 are respectively input into the convolutional layer 305 of the convolutional neural network to obtain a set of feature vectors 306 corresponding to the key part image sequence 302 and a set of corresponding key part image sequences 304 Eigenvector307.

Afterwards, read a corresponding pair of feature vectors 308 and 309 from the two sets of feature vectors 306 and 307 , and calculate the similarity 310 of the pair of feature vectors 308 and 309 .

Finally, based on the similarity 310 of the pair of feature vectors 308 and 309 , it is determined whether the faces in the two face images are the same face 311 .

It should be understood that the method for recognizing a human face image shown in FIG. 3 is only an exemplary application scenario of the method for recognizing a human face image, and does not represent a limitation to the present application. For example, the above-mentioned calculation of the similarity 310 of the pair of feature vectors 308 and 309 can be accomplished by using various similarity calculation methods, which will not be repeated here. It should be understood that the method for identifying human face images provided in the above application scenarios of the present application can improve the efficiency of determining whether the human faces in two human face images are the same human face.

Further, please refer to FIG. 4 , which shows a schematic flowchart of another embodiment of a method for recognizing a human face image according to an embodiment of the present application.

As shown in FIG. 4, a method 400 for recognizing a face image includes:

In step 410, image sequences of key parts are respectively acquired from the two face images to obtain two sets of image sequences of key parts.

In step 420, the two sets of key part image sequences are respectively input into the convolution layer of the convolutional neural network to obtain two sets of feature vectors.

In step 430, a corresponding pair of feature vectors is read from the two groups of feature vectors, and the similarity of the pair of feature vectors is calculated.

It should be understood that step 410, step 420, and step 430 in this embodiment correspond to step 210, step 220, and step 230 in the embodiment shown in FIG. 2 respectively, and details are not repeated here.

In step 440 , it is judged whether the similarity of the pair of feature vectors is smaller than a preset similarity threshold, if yes, go to step 450 , if not, go to step 460 .

In this embodiment, the similarity threshold is the similarity probability threshold set by the same human face in two face images determined in advance based on historical data or artificial experience. When the similarity calculated in step 430 is less than When the similarity threshold is reached, the similarity of the key part images corresponding to the eigenvectors in the two face images is low, and it can be considered that the faces in the two face images are not the same face; when the similarity calculated in step 430 When the degree is not less than the similarity threshold, the image similarity of key parts corresponding to the pair of feature vectors in the two face images is relatively high, and the faces in the two face images may be the same face.

In step 450, it is determined that the faces in the two face images are not the same face.

In this embodiment, since the similarity threshold represents the probability threshold that the faces in the two face images are the same face, when the similarity is less than the similarity threshold, it means that the faces in the two face images are the same. The probability of the face is less than the probability threshold, that is, the faces in the two face images are not the same face.

In step 460 , it is judged whether the pair of eigenvectors is the last pair of eigenvectors in the two groups of eigenvectors, if yes, execute step 470 , if not, skip to execute step 430 .

In this embodiment, when performing further feature confirmation based on the similarity not less than the similarity threshold to determine whether the faces in the two face images are the same face, you can jump to step 430 and read the two face images. The next pair of eigenvectors in the group eigenvectors, and calculate the similarity, and then determine whether to continue to judge the next pair of eigenvectors according to the comparison result of the similarity and the similarity threshold, until there are no other vectors in the two groups of eigenvectors. In the process of judging, if the similarity of a pair of feature vectors is less than the similarity threshold, it indicates that the similarity of the key part images corresponding to the pair of feature vectors is low, so it can be determined that the faces in the two face images are not the same human face.

In step 470, it is determined that the faces in the two face images are the same face.

In this embodiment, if the similarity of the last pair of feature vectors is still not less than the similarity threshold, it indicates that the similarity of each pair of feature vectors in the two groups of feature vectors is greater than the similarity threshold, that is, each group of key The part images are all images of the same key part, therefore, it can be determined that the faces in the two face images are the same face.

It should be understood that the method for recognizing a human face image shown in FIG. 4 above is only an exemplary embodiment of the method for recognizing a human face image, and does not represent a limitation to this application. For example, step 410 of this application The key part image sequences are obtained from two face images respectively, and two sets of key part image sequences are obtained, in which the key parts can be 95 key points, 85 key points or 68 key points of the reference face image, etc. identified key areas.

Further referring to FIG. 5 , as an implementation of the above method, an embodiment of the present application provides an embodiment of a device for recognizing a face image, and the embodiment of the device for recognizing a face image is the same as that shown in FIGS. 4 corresponds to the embodiment of the method for recognizing a human face image, thus, the operations and features described above for the method for recognizing a human face image in FIGS. The face image device 500 and the units included therein will not be described in detail here.

As shown in FIG. 5, the device 500 for recognizing human face images may include: an image sequence acquisition unit 510, which is used to acquire key part image sequences from two human face images respectively to obtain two sets of key part image sequences; The vector acquisition unit 520 is used to input two groups of key part image sequences into the convolutional layer of the convolutional neural network respectively to obtain two groups of feature vectors; the similarity calculation unit 530 is used to read the corresponding two groups of feature vectors. A pair of eigenvectors, calculating the similarity of the pair of eigenvectors; the same face determination unit 540 is used to determine whether the faces in the two face images are the same face based on the similarity of the pair of eigenvectors.

In some optional implementations of this embodiment, the same face determining unit 540 is further configured to: determine that the faces in the two face images are not the same in response to the similarity between the pair of feature vectors being less than a preset similarity threshold face.

In some optional implementations of this embodiment, the same face determination unit 540 is further configured to: respond to the similarity of the pair of feature vectors being greater than a preset similarity threshold and the pair of feature vectors is not the last of the two groups of feature vectors A pair of eigenvectors, jump to read the corresponding pair of eigenvectors from the two sets of eigenvectors, and calculate the similarity of the pair of eigenvectors; based on the similarity of the pair of eigenvectors, determine the Whether the faces are the same face.

In some optional implementations of this embodiment, the same face determining unit 540 is further configured to: in response to the similarity of the pair of feature vectors being greater than a preset similarity threshold and the pair of feature vectors being the last of the two groups of feature vectors A pair of eigenvectors, identifying the faces in the two face images as the same face.

In some optional implementations of this embodiment, the key parts in the key part image sequence in the image sequence acquiring unit 510 include: left eye, right eye, nose, left corner of mouth, and right corner of mouth.

In some optional implementations of this embodiment, calculating the similarity of the pair of feature vectors in the similarity calculation unit 530 includes calculating any of the following items of the pair of feature vectors: Pearson correlation coefficient, Euclidean distance, cosine similarity , Manhattan distance and Chebyshev distance.

The present application also provides an embodiment of a device, including: one or more processors; a storage device for storing one or more programs; when one or more programs are executed by one or more processors, a or a plurality of processors to implement the method for recognizing a human face image as described in any one of the above.

The present application also provides an embodiment of a computer-readable medium, on which a computer program is stored, and when the program is executed by a processor, the method for recognizing a human face image as described in any one of the above items is implemented.

Referring now to FIG. 6 , it shows a schematic structural diagram of a computer system 600 suitable for implementing a terminal device or a server according to an embodiment of the present application. The terminal device shown in FIG. 6 is only an example, and should not limit the functions and scope of use of this embodiment of the present application.

As shown in FIG. 6 , a computer system 600 includes a central processing unit (CPU) 601 that can execute programs according to programs stored in a read-only memory (ROM) 602 or loaded from a storage section 606 into a random-access memory (RAM) 603 Instead, various appropriate actions and processes are performed. In the RAM 603, various programs and data necessary for the operation of the system 600 are also stored. The CPU 601 , ROM 602 , and RAM 603 are connected to each other through a bus 604 . An input/output (I/O) interface 605 is also connected to the bus 604 .

The following components are connected to the I/O interface 605: an input section 606 including a keyboard, a mouse, etc.; an output section 607 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker; a storage section 608 including a hard disk, etc. and a communication section 609 including a network interface card such as a LAN card, a modem, or the like. The communication section 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to the I/O interface 605 as needed. A removable medium 611 such as a magnetic disk, optical disk, magneto-optical disk, semiconductor memory, etc. is mounted on the drive 610 as necessary so that a computer program read therefrom is installed into the storage section 608 as necessary.

In particular, according to an embodiment of the present disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, the embodiments of the present disclosure include a computer program product including a computer program carried on a computer-readable medium, the computer program including program codes for executing the methods shown in the flowcharts. In such an embodiment, the computer program may be downloaded and installed from a network via communication portion 609 and/or installed from removable media 611 . When the computer program is executed by the central processing unit (CPU) 601, the above-mentioned functions defined in the method of the present application are performed.

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

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

The units involved in the embodiments described in the present application may be implemented by means of software or by means of hardware. The described units can also be set in a processor, for example, it can be described as: a processor includes an image sequence acquisition unit, a feature vector acquisition unit, a similarity calculation unit and an identical face determination unit. The names of these units do not constitute a limitation of the unit itself in some cases. For example, the image sequence acquisition unit can also be described as "obtaining the image sequence of key parts from two face images respectively, and obtaining two sets of key parts A unit of an image sequence".

As another aspect, the present application also provides a non-volatile computer storage medium, which may be the non-volatile computer storage medium contained in the device described in the above embodiments; It may be a non-volatile computer storage medium that exists independently and is not assembled into the terminal. The above-mentioned non-volatile computer storage medium stores one or more programs, and when the one or more programs are executed by a device, the device: acquires key part image sequences from two face images respectively, and obtains Two sets of image sequences of key parts; input the two sets of image sequences of key parts into the convolutional layer of the convolutional neural network to obtain two sets of feature vectors; read the corresponding pair of feature vectors from the two sets of feature vectors, and calculate the pair The similarity of the feature vectors; based on the similarity of the pair of feature vectors, it is determined whether the faces in the two face images are the same face.

The above description is only a preferred embodiment of the present application and an illustration of the applied technical principle. Those skilled in the art should understand that the scope of the invention involved in this application is not limited to the technical solution formed by the specific combination of the above-mentioned technical features, and should also cover the technical solutions formed by the above-mentioned technical features or without departing from the above-mentioned inventive concept. Other technical solutions formed by any combination of equivalent features. For example, a technical solution formed by replacing the above-mentioned features with technical features with similar functions disclosed in (but not limited to) this application.

Claims (14)

1. a kind of method of facial image for identification, including：

It obtains key position image sequence respectively from two facial images, obtains two groups of key position image sequences；

The convolutional layer that key position image sequence described in two groups is inputted to convolutional neural networks respectively obtains two groups of feature vectors；

Corresponding a pair of of feature vector is read from two groups of described eigenvectors, calculates the similarity to feature vector；

Similarity based on this to feature vector determines whether the face in two facial images is same face.

2. according to the method described in claim 1, wherein, the similarity based on this to feature vector determines described in two Whether the face in facial image is that same face includes：

It is less than default similarity threshold in response to the similarity to feature vector, determines the face in two facial images Not same face.

3. according to the method described in claims 1 or 2 any one, wherein the similarity based on this to feature vector, really Whether the face in fixed two facial images is that same face further includes：

It is not spy described in two groups to feature vector to be more than default similarity threshold and this in response to the similarity to feature vector Last in sign vector jumps to feature vector and described reads corresponding a pair of of feature from two groups of described eigenvectors Vector calculates the similarity to feature vector；Similarity based on this to feature vector determines in two facial images Face whether be same face.

4. according to the method described in claims 1 to 3 any one, wherein the similarity based on this to feature vector, really Whether the face in fixed two facial images is that same face further includes：

It is feature described in two groups to feature vector to be more than default similarity threshold and this in response to the similarity to feature vector Last in vector determines that the face in two facial images is same face to feature vector.

5. according to the method described in claim 1, wherein, the key position in the key position image sequence includes：Left eye, Right eye, nose, the left corners of the mouth and the right corners of the mouth.

6. according to the method described in claim 1, wherein, the described similarity to feature vector that calculates includes calculating this to spy Levy the following any one of vector：Pearson correlation coefficient, Euclidean distance, cosine similarity, manhatton distance and Chebyshev Distance.

7. a kind of device of facial image for identification, including：

Image sequence acquiring unit obtains two groups of passes for obtaining key position image sequence respectively from two facial images Key position image sequence；

Feature vector acquiring unit, the convolution for key position image sequence described in two groups to be inputted to convolutional neural networks respectively Layer, obtains two groups of feature vectors；

It is right to calculate this for reading corresponding a pair of of feature vector from two groups of described eigenvectors for similarity calculated The similarity of feature vector；

Identical face determination unit is used for similarity to feature vector based on this, determines the people in two facial images Whether face is same face.

8. device according to claim 7, wherein the identical face determination unit is further used for：

It is less than default similarity threshold in response to the similarity to feature vector, determines the face in two facial images Not same face.

9. according to the device described in claim 7 or 8 any one, wherein the identical face determination unit is further used for：

It is not spy described in two groups to feature vector to be more than default similarity threshold and this in response to the similarity to feature vector Last in sign vector jumps to feature vector and described reads corresponding a pair of of feature from two groups of described eigenvectors Vector calculates the similarity to feature vector；Similarity based on this to feature vector determines in two facial images Face whether be same face.

10. according to the device described in claim 7 to 9 any one, wherein the identical face determination unit is further used In：

It is feature described in two groups to feature vector to be more than default similarity threshold and this in response to the similarity to feature vector Last in vector determines that the face in two facial images is same face to feature vector.

11. device according to claim 7, wherein the key position image in described image retrieval unit Key position in sequence includes：Left eye, right eye, nose, the left corners of the mouth and the right corners of the mouth.

12. device according to claim 7, wherein calculate the phase to feature vector in the similarity calculated Include the following any one for calculating this to feature vector like degree：It is Pearson correlation coefficient, Euclidean distance, cosine similarity, graceful Hatton's distance and Chebyshev's distance.

13. a kind of equipment, including：

One or more processors；

Storage device, for storing one or more programs；

When one or more of programs are executed by one or more of processors so that one or more of processors are real The now method of the facial image for identification as described in any in claim 1-6.

14. a kind of computer-readable medium, is stored thereon with computer program, such as right is realized when which is executed by processor It is required that the method for the facial image for identification described in any in 1-6.