A 3D face reconstruction method, device, and storage medium

Abstract

A three-dimensional face reconstruction method, device and storage medium, the method comprising the steps of: acquiring multi-view three-dimensional face images; performing three-dimensional reconstruction processing on the multi-view three-dimensional face images; performing shape characterization on the multi-view three-dimensional face images; performing texture rendering on the multi-view three-dimensional face images; constructing a three-dimensional face model according to the multi-view three-dimensional face images. The three-dimensional face reconstruction method, device and storage medium provided by the present application can accurately render face textures and high-frequency details while ensuring rapid acquisition of three-dimensional face contour features, and construct a more accurate 3D face model using a more reliable face region.

Description

Technical Field

[0001] The present invention belongs to the technical field of three-dimensional face models, and particularly relates to a three-dimensional face reconstruction method, device and storage medium. Background Art

[0002] The three-dimensional face model can increase the reference dimension of applications such as face editing and face recognition, and can achieve more accurate face image generation, face image editing and other tasks. The existing technology of using a deep neural network to generate a three-dimensional model from multi-view face images based on 3DMM has deficiencies in the reconstruction accuracy of both the shape and texture models. When reconstructing multi-view face images, the reconstructed model is normalized, which will not only result in insufficient application of face data but also cause a significant loss or change in the three-dimensional characteristics of the face.

[0003] Using the traditional Gan method to optimize the face texture model can restore the texture accuracy. In the face expression method of 3DMM, the shape vector and texture vector of the three-dimensional face are separated. That is, the high-precision optimization of the texture model cannot improve the accuracy of the shape model. That is, the color at the pixel level on the skin of the generated face model has been improved, but the expression accuracy of the internal three-dimensional structure point cloud data for the three-dimensional face has not been improved.

[0004] The BFM model used in traditional shape model reconstruction is a basic 3DMM model. The dimension of the shape vector parameters is limited. At the same time, in order to prevent overfitting, the dimension of the shape parameters regressed by traditional neural network algorithms will also be limited to less than 99 dimensions. At the same time, for the general three-dimensional face reconstruction algorithm that adopts the strategy of reconstructing shape and texture simultaneously, the attention of the network in the loss function cannot be fully concentrated on the fitting of the shape parameters, resulting in insufficient shape accuracy. At the same time, when fitting the shape parameters, the landmark loss function used does not distinguish between important and unimportant regions of the face, resulting in insufficient attention and fitting degree for areas with wrinkles such as facial features and eye corners.

[0005] For multi-view face data, existing algorithms also simply perform weighted averaging or high-performance selection on the models reconstructed from each face separately, and cannot make full use of information such as the angle and regional confidence of the face. The current three-dimensional face reconstruction method constructs face shape and face texture information through the same network model, and uses a convolutional neural network to regress the shape model parameters, with insufficient accuracy. The multi-view face reconstruction three-dimensional image does not make full use of multi-view face features, and the shape details of the facial features area of the simple face model are not depicted enough. Summary of the Invention

[0006] In view of the above problems, the present invention provides a three-dimensional face reconstruction method, apparatus, and storage medium that overcome the above problems or at least partially solve the above problems.

[0007] To solve the above technical problems, the present invention provides a three-dimensional face reconstruction method, the method including the steps of:

[0008] Obtain multi-view three-dimensional face images;

[0009] Perform three-dimensional reconstruction processing on the multi-view three-dimensional face images;

[0010] Characterize the shape of the multi-view three-dimensional face images;

[0011] Perform texture rendering on the multi-view three-dimensional face images;

[0012] Construct a three-dimensional face model based on the multi-view three-dimensional face images.

[0013] Preferably, the performing three-dimensional reconstruction processing on the multi-view three-dimensional face images includes the steps of:

[0014] Perform identity annotation on all the multi-view three-dimensional face images and obtain an initial face database;

[0015] Intercept and select each face image in the initial face database;

[0016] Perform 68-point feature point annotation on the intercepted and selected area and obtain a preliminarily processed face database;

[0017] Perform non-rigid transformation alignment on the preliminarily processed face database and obtain a face database in 3DMM format.

[0018] Preferably, the characterizing the shape of the multi-view three-dimensional face images includes the steps of:

[0019] Obtain a deep convolutional neural network;

[0020] Perform 3D face reconstruction on the deep convolutional neural network and obtain a reconstructed convolutional neural network;

[0021] Remove the network structure for texture construction in the reconstructed convolutional neural network;

[0022] Train the reconstructed convolutional neural network using the face database in 3DMM format;

[0023] Solve for the parameters of the face shape parameters using a loss function;

[0024] Increase the weights of the facial feature points.

[0025] Preferably, the texture rendering of the multi-view three-dimensional face image includes the steps of:

[0026] Obtain a face database in 3DMM format;

[0027] Use a generative adversarial network to train the face database in 3DMM format and obtain a three-dimensional face shape;

[0028] Use the generative adversarial network to perform texture detail rendering on the three-dimensional face shape.

[0029] Preferably, the construction of the three-dimensional face model according to the multi-view three-dimensional face image includes the steps of:

[0030] Obtain a reconstructed three-dimensional face;

[0031] Define a confidence region of the reconstructed three-dimensional face on the multi-view three-dimensional face image;

[0032] Weightedly construct the multi-view three-dimensional face image according to the confidence region.

[0033] Preferably, the marking of 68 feature points on the intercepted selected region and obtaining a preliminary processed face database includes the steps of:

[0034] Establish a three-dimensional coordinate system on the multi-view three-dimensional face image with the left ear, the top of the head, and the nose as the positive directions of the X, Y, and Z axes respectively;

[0035] Select the positions of two points at the chin of the face as the interception boundaries along the negative direction of the Z axis;

[0036] Mark the 68 marking points in the face feature point specification on the intercepted multi-view three-dimensional face image.

[0037] Preferably, the defining of the confidence region of the reconstructed three-dimensional face on the multi-view three-dimensional face image includes the steps of:

[0038] Calculate the deformation degree of the multi-view three-dimensional face image in different views compared with the frontal face image;

[0039] Define the confidence region according to the deformation degree;

[0040] Compare the usability of all the confidence regions;

[0041] Set a weight greater than a preset threshold for the confidence region corresponding to the usability greater than the preset threshold.

[0042] This application also provides a three-dimensional face reconstruction device, and the device includes the steps of:

[0043] A multi-view three-dimensional face image acquisition module, configured to acquire a multi-view three-dimensional face image;

[0044] A three-dimensional reconstruction processing module for performing three-dimensional reconstruction processing on the multi-view three-dimensional face images;

[0045] A shape characterization module for characterizing the shape of the multi-view three-dimensional face images;

[0046] A texture rendering module for performing texture rendering on the multi-view three-dimensional face images;

[0047] A three-dimensional face model construction module for constructing a three-dimensional face model based on the multi-view three-dimensional face images.

[0048] This application also provides an electronic device, which includes:

[0049] At least one processor; and,

[0050] A memory communicatively connected to the at least one processor; wherein,

[0051] The memory stores instructions executable by the at least one processor, and when the instructions are executed by the at least one processor, the at least one processor is enabled to execute any one of the foregoing three-dimensional face reconstruction methods.

[0052] This application also provides a non-transitory computer-readable storage medium, which stores computer instructions for causing the computer to execute any one of the foregoing three-dimensional face reconstruction methods.

[0053] One or more technical solutions in the embodiments of the present invention have at least the following technical effects or advantages: A three-dimensional face reconstruction method, apparatus, and storage medium provided by this application can accurately render face textures and high-frequency details while ensuring rapid acquisition of three-dimensional face contour features, and construct a more accurate 3D face model using a more reliable face region. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] To more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0055] Figure 1 is a flowchart of a three-dimensional face reconstruction method provided by an embodiment of the present invention;

[0056] Figure 2 is a structural diagram of a three-dimensional face reconstruction apparatus provided by an embodiment of the present invention;

[0057] Figure 3 It is a schematic structural diagram of an electronic device provided by an embodiment of the present invention;

[0058] Figure 4 It is a schematic structural diagram of a non-transitory computer-readable storage medium provided by an embodiment of the present invention. Specific embodiments

[0059] The following will specifically describe the present invention in combination with specific embodiments and examples, and the advantages and various effects of the present invention will be presented more clearly therefrom. Those skilled in the art should understand that these specific embodiments and examples are used to illustrate the present invention, rather than to limit the present invention.

[0060] Throughout the specification, unless otherwise specifically stated, the terms used herein should be understood as having the meanings commonly used in the art. Therefore, unless otherwise defined, all technical and scientific terms used herein have the same meaning as the general understanding of those skilled in the technical field to which the present invention belongs. In case of conflict, this specification prevails.

[0061] Unless otherwise specifically stated, various raw materials, reagents, instruments, and equipment used in the present invention can be obtained through market purchases or can be prepared by existing methods.

[0062] Such as Figure 1 , in an embodiment of the present application, the present invention provides a three-dimensional face reconstruction method, and the method includes the steps of:

[0063] S1: Obtain multi-view three-dimensional face images;

[0064] In an embodiment of the present application, a three-dimensional face acquisition device is used to collect face images at multiple angles for each identity, and multi-view three-dimensional face images of each identity can be captured.

[0065] S2: Perform three-dimensional reconstruction processing on the multi-view three-dimensional face images;

[0066] In an embodiment of the present application, the performing three-dimensional reconstruction processing on the multi-view three-dimensional face images includes the steps of:

[0067] Perform identity annotation on all the multi-view three-dimensional face images and obtain an initial face database;

[0068] Intercept and select each face image in the initial face database;

[0069] Perform 68-point feature point annotation on the intercepted and selected area and obtain a preliminary processed face database;

[0070] Perform non-rigid transformation alignment on the preliminary processed face database to obtain a face database in 3DMM format.

[0071] In the embodiment of the present application, after collecting multi-view three-dimensional face images, identity annotation is performed one by one, and at this time, an initial face database can be obtained; then, each face image in the initial face database is intercepted and selected, and the intercepted and selected area is marked with 68 feature points as the preliminary processed face database after preliminary processing; then, the preliminary processed face database after preliminary processing is aligned using non-rigid transformation operation to obtain a 3DMM format face database in 3DMM format.

[0072] In the embodiment of the present application, the marking of 68 feature points on the intercepted and selected area to obtain the preliminary processed face database includes the steps of:

[0073] Establish a three-dimensional coordinate system on the multi-view three-dimensional face image with the left ear, the top of the head, and the nose as the positive directions of the X, Y, and Z axes respectively;

[0074] Select the positions of two points on the chin of the face along the negative direction of the Z axis as the interception boundaries;

[0075] Mark the 68 marking points in the face feature point specification on the intercepted multi-view three-dimensional face image.

[0076] In the embodiment of the present application, first establish a three-dimensional coordinate system for the three-dimensional face. At this time, a three-dimensional coordinate system is established on the multi-view three-dimensional face image with the left ear, the top of the head, and the nose as the positive directions of the X, Y, and Z axes respectively. According to the marked face landmark points, select the positions of two points on the chin of the face along the negative direction of the Z axis as the interception boundaries, and mark the 68 marking points in the face feature point specification for the intercepted face.

[0077] S3: Perform shape characterization on the multi-view three-dimensional face image;

[0078] In the embodiment of the present application, the performing of shape characterization on the multi-view three-dimensional face image includes the steps of:

[0079] Obtain a deep convolutional neural network;

[0080] Perform 3D face reconstruction on the deep convolutional neural network to obtain a reconstructed convolutional neural network;

[0081] Remove the network structure for texture construction in the reconstructed convolutional neural network;

[0082] Train the reconstructed convolutional neural network using the 3DMM format face database;

[0083] Use a loss function to solve the parameters of the face shape parameters;

[0084] Increase the weights of facial feature characteristics.

[0085] In the embodiment of the present application, when performing shape characterization on the multi-view three-dimensional face image, first, a 3D face reconstruction is performed on the deep convolutional neural network to obtain a reconstructed convolutional neural network. Then, the network structure part for texture construction in the reconstructed convolutional neural network is removed, and the processed reconstructed convolutional neural network is trained in a supervised learning manner using the labeled 3DMM format face database. During training, the asymmetric Euler loss and the landmark loss are used as loss functions to solve for the face shape parameters, strengthening the weights of the facial feature characteristics that can better distinguish face identities, so as to increase the distinguishability of different faces and widen the inter-class distance.

[0086] In the embodiment of the present application, the reconstruction of the 3D face shape includes the reconstruction of the face contour and the shape reconstruction of facial features. The asymmetric Euler loss can perform regression reconstruction on both simultaneously, and the landmark loss can strengthen the penalty for feature points, so that the facial features can be well fitted.

[0087] S4: Perform texture rendering on the multi-view three-dimensional face image;

[0088] In the embodiment of the present application, the performing texture rendering on the multi-view three-dimensional face image includes the steps of:

[0089] Obtain a 3DMM format face database;

[0090] Use a generative adversarial network to train the 3DMM format face database to obtain a three-dimensional face shape;

[0091] Use the generative adversarial network to perform texture detail rendering on the three-dimensional face shape.

[0092] In the embodiment of the present application, when using a generative adversarial network to train the 3DMM format face database to obtain a three-dimensional face shape, specifically, during training, the labels of the real sample set are all set to 1 for training the discriminative network. Then, in the way of connecting the generative-discriminative network in series, first, a sample set is generated by the generative adversarial network, and the label of this sample set is set to 1. Then, the data is input into the discriminative network, and the generated error does not update the parameters of the discriminative network but only backpropagates, and the parameters of the generative adversarial network are updated. In this way, the final generative adversarial network is obtained through alternating iterative training. When using the generative adversarial network to perform texture detail rendering on the three-dimensional face shape, specifically, the constructed three-dimensional face shape image is input into the trained generative adversarial network to construct a face image with texture detail information.

[0093] S5: Construct a three-dimensional face model according to the multi-view three-dimensional face image.

[0094] In an embodiment of the present application, the constructing a three-dimensional face model based on the multi-view three-dimensional face images includes the steps of:

[0095] Obtain a reconstructed three-dimensional face;

[0096] Define a confidence region of the reconstructed three-dimensional face on the multi-view three-dimensional face images;

[0097] Perform weighted construction on the multi-view three-dimensional face images according to the confidence region.

[0098] In an embodiment of the present application, the defining a confidence region of the reconstructed three-dimensional face on the multi-view three-dimensional face images includes the steps of:

[0099] Calculate the deformation degree of the multi-view three-dimensional face images of different views compared with the frontal face image;

[0100] Define the confidence region according to the deformation degree;

[0101] Compare the availability of all the confidence regions;

[0102] Set the weight greater than a preset threshold for the confidence region corresponding to the availability greater than the preset threshold.

[0103] In an embodiment of the present application, the confidence level can be defined according to the deformation degree of the face images of different views compared with the frontal face image, and a higher weight is set for the area range with high availability. For the detection of the deformation degree, opencv is used to extract the face key points, calculate the face rotation angle, and according to the rotation angle; according to the rotation angle, the face credibility region is selected, and the size of the credibility region is determined as follows: the width is the distance between the outer canthi of the two eyes, and the length is the length of the entire 3DMM face, which is a bar-shaped region.

[0104] In an embodiment of the present application, rotate and move the bar-shaped region according to the rotation angle; use the face recognition algorithm to calculate the similarity of the face images, and determine the final weight through the similarity. The confidence weight of the non-confidence region is set to 1; the weight formula of the confidence region is as follows:

[0105]

[0106] In an embodiment of the present application, all the images of the same identity of different views constructed are weighted and fused according to the set face confidence region to construct the final three-dimensional face image. The calculation formula of the final coordinate of each three-dimensional point is as follows:

[0107]

[0108] Among them, W i represents the weight of the point point.

[0109] As Figure 2 , in the embodiments of the present application, the present application further provides a three-dimensional face reconstruction device, and the device includes the following steps:

[0110] A multi-view three-dimensional face image acquisition module 10, configured to acquire multi-view three-dimensional face images;

[0111] A three-dimensional reconstruction processing module 20, configured to perform three-dimensional reconstruction processing on the multi-view three-dimensional face images;

[0112] A shape characterization module 30, configured to perform shape characterization on the multi-view three-dimensional face images;

[0113] A texture rendering module 40, configured to perform texture rendering on the multi-view three-dimensional face images;

[0114] A three-dimensional face model construction module 50, configured to construct a three-dimensional face model according to the multi-view three-dimensional face images.

[0115] The three-dimensional face reconstruction device provided by the present application can execute the three-dimensional face reconstruction method provided by the above steps.

[0116] Next, referring to Figure 3 , which shows a schematic structural diagram of an electronic device 100 suitable for implementing the embodiments of the present disclosure. The electronic devices in the embodiments of the present disclosure may include, but are not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Tablet Computers), PMPs (Portable Multimedia Players), in-vehicle terminals (such as in-vehicle navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 3 The electronic device shown is only an example and should not impose any limitation on the functions and usage scope of the embodiments of the present disclosure.

[0117] As Figure 3 shown, the electronic device 100 may include a processing device (such as a central processing unit, a graphics processing unit, etc.) 101, which can perform various appropriate actions and processes according to the programs stored in the read-only memory (ROM) 102 or the programs loaded from the storage device 108 into the random access memory (RAM) 103. In the RAM 103, various programs and data required for the operation of the electronic device 100 are also stored. The processing device 101, the ROM 102, and the RAM 103 are connected to each other through a bus 104. The input / output (I / O) interface 105 is also connected to the bus 104.

[0118] Generally, the following devices can be connected to the I / O interface 105: an input device 106 including, for example, a touch screen, a touchpad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 107 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 108 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 109. The communication device 109 can allow the electronic device 100 to communicate with other devices wirelessly or wiredly to exchange data. Although the electronic device 100 with various devices is shown in the figure, it should be understood that it is not required to implement or have all the shown devices. More or fewer devices can be alternatively implemented or had.

[0119] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program contains program codes for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication device 109, or installed from the storage device 108, or installed from the ROM 102. When the computer program is executed by the processing device 101, the above functions defined in the method of the embodiment of the present disclosure are executed.

[0120] Refer to the following Figure 4 , which shows a schematic structural diagram of a computer-readable storage medium suitable for implementing an embodiment of the present disclosure. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, it can implement the three-dimensional face reconstruction method described in any one of the above.

[0121] It should be noted that the above-mentioned computer-readable medium in the present disclosure can be a computer-readable signal medium, a computer-readable storage medium, or any combination of the two. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples of the computer-readable storage medium can include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present disclosure, the computer-readable storage medium can be any tangible medium that contains or stores a program, and this program can be used by or in combination with an instruction execution system, apparatus, or device. In the present disclosure, the computer-readable signal medium can include a data signal propagated in a baseband or as part of a carrier wave, which carries computer-readable program code. Such a propagated data signal can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The computer-readable signal medium can also be any computer-readable medium other than the computer-readable storage medium, and this computer-readable signal medium can send, propagate, or transmit a program for use by or in combination with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted by any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0122] The above-mentioned computer-readable medium can be included in the above-mentioned electronic device; or it can exist separately without being assembled into the electronic device.

[0123] The above-mentioned computer-readable medium carries one or more programs. When the above one or more programs are executed by the electronic device, the electronic device is caused to: obtain at least two Internet protocol addresses; send a node evaluation request including the at least two Internet protocol addresses to a node evaluation device, where the node evaluation device selects an Internet protocol address from the at least two Internet protocol addresses and returns it; receive the Internet protocol address returned by the node evaluation device; where the obtained Internet protocol address indicates an edge node in a content delivery network.

[0124] Alternatively, the above computer-readable medium carries one or more programs which, when executed by the electronic device, cause the electronic device to: receive a node evaluation request including at least two Internet Protocol addresses; select an Internet Protocol address from the at least two Internet Protocol addresses; return the selected Internet Protocol address; wherein the received Internet Protocol addresses indicate edge nodes in a content delivery network.

[0125] Computer program code for carrying out operations of the present disclosure may be written in one or more programming languages or combinations thereof. The programming languages include object-oriented programming languages such as Java, Smalltalk, C++, and also include conventional procedural programming languages such as the "C" language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer, or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0126] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagram may represent a module, a segment of a program, or a part of code, which contains one or more executable instructions for implementing the specified logical function. It should also be noted that, in some alternative implementations, the functions noted in the blocks may occur in an order different from that noted in the drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, or they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented by a dedicated hardware-based system that performs the specified functions or operations, or by a combination of dedicated hardware and computer instructions.

[0127] The units involved in the embodiments of the present disclosure may be implemented in software or in hardware. Wherein, the name of the unit does not constitute a limitation to the unit itself in some cases. For example, the first acquisition unit may also be described as "the unit for acquiring at least two Internet Protocol addresses".

[0128] A 3D face reconstruction method, device, and storage medium provided by this application can accurately render face textures and high-frequency details while ensuring rapid acquisition of 3D face contour features, and construct a more accurate 3D face model using a more reliable face region.

[0129] It should be understood that the above specific embodiments of the present invention are only for illustrative explanation or interpretation of the principles of the present invention, and do not constitute a limitation on the present invention. Therefore, any modifications, equivalent replacements, improvements, etc. made without departing from the spirit and scope of the present invention shall be included within the protection scope of the present invention. In addition, the appended claims of the present invention are intended to cover all variations and modification examples falling within the scope and boundaries of the appended claims, or equivalent forms of such scope and boundaries.

Claims

1. A three-dimensional face reconstruction method, characterized in that, the method comprises the steps of: acquiring multi-view three-dimensional face images; performing three-dimensional reconstruction processing on the multi-view three-dimensional face images; characterizing the shape of the multi-view three-dimensional face images; rendering the texture of the multi-view three-dimensional face images; constructing a three-dimensional face model based on the multi-view three-dimensional face images; the performing three-dimensional reconstruction processing on the multi-view three-dimensional face images comprises the steps of: performing identity annotation on all the multi-view three-dimensional face images to obtain an initial face database; performing cropping and selection on each face image in the initial face database; performing sixty-eight feature point annotations on the cropped and selected area to obtain a preliminarily processed face database; performing non-rigid transformation alignment on the preliminarily processed face database to obtain a face database in 3DMM format; the characterizing the shape of the multi-view three-dimensional face images comprises the steps of: acquiring a deep convolutional neural network; performing 3D face reconstruction on the deep convolutional neural network to obtain a reconstructed convolutional neural network; removing the network structure for texture construction in the reconstructed convolutional neural network; training the reconstructed convolutional neural network using the face database in 3DMM format; solving the parameters of the face shape parameters using a loss function; increasing the weights of the facial feature weights; the rendering the texture of the multi-view three-dimensional face images comprises the steps of: acquiring a face database in 3DMM format; training the face database in 3DMM format using a generative adversarial network to obtain a three-dimensional face shape; performing texture detail rendering on the three-dimensional face shape using the generative adversarial network.

2. The three-dimensional face reconstruction method according to claim 1, characterized in that, the constructing a three-dimensional face model based on the multi-view three-dimensional face images comprises the steps of: acquiring a reconstructed three-dimensional face; defining a confidence region of the reconstructed three-dimensional face on the multi-view three-dimensional face images; performing weighted construction on the multi-view three-dimensional face images according to the confidence region.

3. The three-dimensional face reconstruction method according to claim 1, characterized in that, the performing sixty-eight feature point annotations on the cropped and selected area to obtain a preliminarily processed face database comprises the steps of: respectively establishing a three-dimensional coordinate on the multi-view three-dimensional face images with the left ear, the top of the head, and the nose as the positive directions of the X, Y, and Z axes; selecting the positions of two points at the chin of the face as the cropping boundaries along the negative direction of the Z axis; annotating the sixty-eight annotation points in the face feature point specification on the cropped multi-view three-dimensional face images.

4. The three-dimensional face reconstruction method according to claim 2, characterized in that, the defining a confidence region of the reconstructed three-dimensional face on the multi-view three-dimensional face images comprises the steps of: calculating the deformation degree of the multi-view three-dimensional face images at different views compared with the frontal face image; defining a confidence region according to the deformation degree; comparing the usability of all the confidence regions; setting the weight greater than a preset threshold for the confidence region corresponding to the usability greater than the preset threshold.

5. A three-dimensional face reconstruction device for implementing the method according to any one of claims 1-4, It is characterized in that The device includes the steps of: A multi-view three-dimensional face image acquisition module for acquiring multi-view three-dimensional face images; A three-dimensional reconstruction processing module for performing three-dimensional reconstruction processing on the multi-view three-dimensional face images; A shape characterization module for characterizing the shape of the multi-view three-dimensional face images; A texture rendering module for rendering the texture of the multi-view three-dimensional face images; A three-dimensional face model construction module for constructing a three-dimensional face model based on the multi-view three-dimensional face images.

6. An electronic device, It is characterized in that The electronic device includes: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the three-dimensional face reconstruction method according to any one of the preceding claims 1-4.

7. A non-transitory computer-readable storage medium storing computer instructions for causing the computer to execute the three-dimensional face reconstruction method according to any one of the preceding claims 1-4.

Images