Method and system for generating three-dimensional face animation based on physiological prior

Abstract

The application provides a three-dimensional face animation generation method and system based on physiological prior art, and belongs to the fields of computer vision and computer simulation, and comprises the following steps: S1: obtaining an initial three-dimensional face model based on CT and MRI scanning section images; correcting, mesh processing and filling the initial three-dimensional face model to obtain a three-dimensional face physiological model; S2: using an unstructured Delaunay algorithm to realize unstructured tetrahedral partitioning on the three-dimensional face physiological model to obtain a three-dimensional face physiological model mesh partitioning result; S3: performing biomechanical modeling on the three-dimensional face physiological model mesh partitioning result; S4: adding contact constraints and boundary conditions between different tissues of the three-dimensional face physiological model to construct a finite element analysis model to simulate facial deformation; and S5: solving and analyzing the finite element model to obtain a three-dimensional face animation. The method can generate high-quality three-dimensional face animations based on a small amount of samples.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of computer vision and computer simulation, and particularly relates to a three-dimensional face animation generation method and system based on physiological priori. BACKGROUND

[0002] In recent years, with the development of artificial intelligence technology and computer hardware, virtual digital people have entered the public view, and have shown great application potential in the fields of human-computer interaction, game and film industry, and virtual live broadcast. The modeling and driving of the face are the key work of digital people technology. Humans are very sensitive to changes in the face. High-quality face animation can not only quickly convey information, but also express emotions and psychological states, giving people a face-to-face communication feeling. Low-quality face animation will make people suspicious of the content and even feel uncomfortable. This puts forward requirements for the generation quality of face animation. At present, the research on face animation generation methods in the academic and industrial circles has reached the stage of high-fidelity super-realism, requiring each frame of face animation to have photo-level precision and fine detail that cannot be distinguished by the human eye.

[0003] According to the type of face animation, face animation can be divided into two categories: two-dimensional face animation and three-dimensional face animation. Compared with two-dimensional face animation, three-dimensional face animation provides comprehensive depth information and spatial dimensions, can show more realistic and lively visual effects, accurately simulates the dynamics of muscles and skin, and supports complex lighting effects. Three-dimensional face animation has two categories in terms of driving method: data-driven and physiological priori-driven methods. Although the data-driven method represented by video or motion capture can achieve highly realistic animation effects, it has high cost, strong dependence on equipment, and complex processing process. The biggest limitation at the present stage is that high-quality open-source three-dimensional face animation data is scarce, which cannot support the training of data-driven models, so over-smooth face animation will be generated.

[0004] FaceFormer: Speech-Driven 3D Facial Animation with Transformers is a paper published in CVPR 2022. The paper proposes a deep learning-based method for generating three-dimensional facial animation, and proposes a data-driven deep learning network model Faceformer, which is composed of an encoder and a decoder. Based on the FLAME three-dimensional face model, the audio features extracted from the input audio are mapped to three-dimensional facial movements, thereby realizing the generation of three-dimensional facial animation. SOTA results have been achieved on multiple datasets. However, due to the difficulty of collection and processing, the three-dimensional facial animation database used by it is relatively small, and the facial animation generated by Faceformer is too smooth, and the capture and prediction of high-frequency details in the facial area are not accurate enough. Therefore, how to generate high-quality three-dimensional facial animation based on a small amount of samples has become a problem to be solved. SUMMARY

[0005] To solve the above technical problems, the present application provides a three-dimensional facial animation generation method based on physiological priors, comprising the following steps:

[0006] Step S1: obtaining an initial three-dimensional face model based on CT and MRI scan tomographic images; simplifying the initial three-dimensional face model into a three-layer structure of exoskeleton-muscle-soft tissue, and correcting, mesh processing and filling the three-dimensional face physiological model to obtain a three-dimensional face physiological model;

[0007] Step S2: using an unstructured Delaunay algorithm to implement unstructured tetrahedral subdivision on the three-dimensional face physiological model to obtain a three-dimensional face physiological model grid subdivision result;

[0008] Step S3: performing biomechanical modeling on the three-dimensional face physiological model grid subdivision result, defining the material properties of each level, using rigid body to describe the skeleton, and using the super-elastic model Mooney-Rivlin to describe the muscle and soft tissue;

[0009] Step S4: constructing a finite element to simulate facial deformation, and adding contact constraints between different tissues of the three-dimensional face physiological model to simulate the pressure and friction between different tissues, and adding boundary conditions between the skeleton and the facial soft tissue to constrain the mutual movement between the rigid body and the soft body;

[0010] Step S5: solving and analyzing the finite element to obtain a three-dimensional facial animation.

[0011] Advantages:

[0012] 1. The application generates high-quality three-dimensional face animation in a short time with a small amount of samples, does not rely on massive data sets, and reduces the generation cost of high-quality three-dimensional face animation.

[0013] 2. The application focuses on the movement of internal organs of the face, and the animation conforms to the principle of anatomy, which has guiding significance for computer-aided language teaching and facial surgery.

[0014] 3. The animation quality is closely related to the modeling accuracy of bones, muscles and soft tissues, and the application can be gradually improved with the progress of modeling and simulation methods, and has good practicality and development prospect in the long run.

[0015] 4. The method has strong migration, and can be quickly generated when a new character or a new virtual person is made. DETAILED DESCRIPTION

[0016] Figure 1 A three-dimensional face animation generation method based on physiological prior art of the application is shown in the flowchart.

[0017] Figure 2 A modeling result diagram of a three-dimensional face physiological model is shown in the flowchart.

[0018] Figure 3 A three-dimensional face physiological model mesh partitioning result diagram is shown in the flowchart.

[0019] Figure 4 A hinge point diagram is shown in the flowchart.

[0020] Figure 5 A finite element solution flowchart is shown in the flowchart.

[0021] Figure 6 A key frame animation diagram of the synthesis result of the final vowel / ao / is shown in the flowchart.

[0022] Figure 7 A key frame animation diagram of the synthesis result of the mouth pinching action is shown in the flowchart.

[0023] Figure 8 A structure block diagram of a three-dimensional face animation generation system based on physiological prior art of the application is shown in the flowchart. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical scheme and advantages of the application clearer and more apparent, the application will be further described in detail below in combination with the drawings and examples. It should be understood that the specific examples described herein are only used to explain the application and do not limit the application. In addition, the technical features involved in each embodiment of the application described below can be combined with each other as long as they do not conflict with each other.

[0025] Example 1

[0026] As Figure 1 shown, the embodiment of the present application provides a three-dimensional face animation generation method based on physiological prior, comprising the following steps:

[0027] Step S1: obtaining an initial three-dimensional face model based on CT and MRI scanning tomographic images; simplifying the initial three-dimensional face model into a three-layer structure of exoskeleton-muscle-soft tissue, correcting, mesh processing and filling the three-layer structure to obtain a three-dimensional face physiological model;

[0028] Step S2: using an unstructured Delaunay algorithm to realize unstructured tetrahedral partitioning of the three-dimensional face physiological model to obtain a three-dimensional face physiological model mesh partitioning result;

[0029] Step S3: performing biomechanical modeling on the three-dimensional face physiological model mesh partitioning result, defining material properties of each level, using rigid body to describe the skeleton, and using a hyperelastic model Mooney-Rivlin to describe the muscle and soft tissue;

[0030] Step S4: constructing a finite element to simulate face deformation, adding contact constraints between different tissues of the three-dimensional face physiological model to simulate pressure and friction between different tissues, and adding boundary conditions between the skeleton and the face soft tissue to constrain the mutual movement between the rigid body and the soft body;

[0031] Step S5: solving and analyzing the finite element to obtain a three-dimensional face animation.

[0032] In one embodiment, the above-mentioned step S1: obtaining an initial three-dimensional face model based on CT and MRI scanning tomographic images; simplifying the initial three-dimensional face model into a three-layer structure of exoskeleton-muscle-soft tissue, correcting, mesh processing and filling the three-layer structure to obtain a three-dimensional face physiological model, specifically comprises:

[0033] At present, the conventional technology can create an accurate three-dimensional physiological model of a face from the tomographic images of CT and MRI scans by using Minics medical image processing software. However, the physiological structure of a face is relatively complex, and organs such as muscles and fat, fascia, and blood vessels are intertwined together. The result of reconstruction by Minics often appears a negative Jacobian matrix that cannot be solved when simulating animation by using a finite element analysis method. Considering the difficulty of analysis for actual face animation generation, the calculation difficulty increases sharply when using a physical method to analyze the overly complex face modeling structure. The embodiment of the present application simplifies the static three-dimensional face model into a three-layer structure from inside to outside, that is, skeleton-muscle-soft tissue. Taking the Minic reconstruction and textbook shape as a reference, the three-dimensional modeling software Zbrush is used for carving and correction. The anatomical accuracy and structural rules of the face model are considered as much as possible to be suitable for physical calculation, and then the engineering file is imported into Blender for automatic mesh processing. First, the three-dimensional model is automatically deformed and matched, and the skin, muscle, and skeleton are completely adapted. The Instent Mesh and QuadriFlow algorithms are used to re-topology optimize the mesh model. Finally, the gap between the skin and the muscle and bone model is filled by using Boolean operation to generate the soft tissue mesh. Figure 2 A modeling result schematic diagram of a three-dimensional face physiological model is shown.

[0034] In one embodiment, the step S2 of using an unstructured Delaunay algorithm to implement unstructured tetrahedral partitioning on the three-dimensional face physiological model to obtain a three-dimensional face physiological model mesh partitioning result includes the following steps.

[0035] The empty circle property and the maximum and minimum angle property of the Delaunay algorithm can obtain high-quality tetrahedrons, and the unstructured partitioning can adapt to complex face shapes. Therefore, the unstructured Delaunay algorithm is used to consider the mesh shape and the calculation amount in the finite element analysis stage.

[0036] In the use of physical methods for simulation analysis of objects, it is necessary to use volume grid. The three-dimensional face physiological model obtained in step S1 is a surface grid, which is hollow inside, so it is necessary to use grid subdivision algorithm to realize grid subdivision, and a specified space is decomposed into a set of simple geometric shapes by grid subdivision. In three-dimensional space, triangular surface grid is decomposed into a volume grid model composed of tetrahedral units. Grid subdivision is a key step for building finite element model and performing finite element analysis, and different types of grid units and subdivision algorithms will significantly affect the scale, result and accuracy of calculation. Grid subdivision is not only a basic link in the process of finite element analysis, but also is crucial to ensure the calculation efficiency and accuracy of the result. According to the type of subdivision, it can be divided into structured subdivision and unstructured subdivision, and unstructured subdivision is suitable for complex set structure. The structure of the face is relatively complex, and the three-dimensional face model is subjected to unstructured tetrahedral subdivision by using unstructured Delaunay algorithm, which can avoid the appearance of low-quality tetrahedron. Figure 3 is a schematic diagram of the three-dimensional face physiological model grid subdivision result, wherein the surface blue line represents the surface structure of the three-dimensional face model, and the section black line represents the internal tetrahedral grid distribution of the three-dimensional face model after unstructured tetrahedral subdivision by the Delaunay algorithm.

[0037] In one embodiment, the step S3 of the above-mentioned step S3: the three-dimensional face physiological model grid subdivision result is subjected to biomechanical modeling, and the material properties of each level are defined, the skeleton is described by rigid body, and the muscle and soft tissue are described by super-elastic Mooney-Rivlin model, which specifically includes:

[0038] The stress-strain properties of soft tissue and muscle are defined as nonlinearity, incompressibility and isotropy. Since Mooney-Rivlin material is a super-elastic material, it has nonlinearity and isotropic stress-strain properties, and a strain energy function is constructed based on Mooney-Rivlin to describe the stress-strain properties:

[0039] ;

[0040] Wherein, represents the strain energy function, and represent the first and second invariants of the strain tensor, and are model material parameters, which can be determined by experiment; represents the elastic volume ratio; represents the temperature related parameter; when used to represent incompressible material, the Poisson's ratio is 0.5, and at this time .

[0041] The biomechanical modeling of the three-dimensional face physiological model is to determine the constitutive equation, i.e. to define the material properties of each level of the three-dimensional face physiological model in step S2. In the embodiment of the present application, for the task of face animation synthesis, the skull will not deform under normal circumstances without suffering a heavy blow or rare situation, and thus is defined as a rigid body. The biomechanical properties of the soft tissue of the face, i.e. the muscles and other soft tissues, are very complex. The soft tissue exhibits complex biomechanical properties in its hyperelasticity, mainly including nonlinearity, anisotropy, quasi-incompressibility, and also plasticity and viscosity. The present application simplifies it to nonlinearity, incompressibility and isotropy.

[0042] Firstly, the nonlinearity. Under small deformation, the muscles and soft tissues exhibit approximately linear elastic behavior, but the slope of the strain with the stress increase is not constant. The stress-strain curve is initially gentle, and then becomes steeper and steeper. Quasi-incompressibility refers to the fact that the volume change during deformation is very small and can be ignored. The quasi-incompressibility condition is achieved by setting the Poisson's ratio of the material to 0.5. The Poisson's ratio is the ratio of the lateral strain to the longitudinal strain of the material. For a material with a Poisson's ratio of 0.5, it is theoretically completely incompressible, i.e. stretching or compression in any direction will not cause a change in volume. The anisotropic property is reflected in its complex internal structure. Due to the combination of cells, fibers and fine structures, the soft tissue exhibits different mechanical properties at different parts thereof. In particular, a fiber-rich tissue such as muscle, its mechanical behavior and response will be significantly different in local areas, which is crucial for setting local parameters. On the contrary, an isotropic material has the same mechanical properties in all directions, and a metal is a typical material that exhibits isotropy within a certain stress range. The present application simplifies the muscles and soft tissues into isotropic materials.

[0043] The muscles and soft tissues exhibit the material properties of hyperelastic materials, and have an elastic potential energy function W, which is a scalar function of the strain tensor, and the derivative of the strain component is the corresponding stress component. Upon unloading, the strain of the material can automatically recover, and the relationship between the mechanical response and the deformation does not simply follow Hooke's law but is one-to-one through the elastic energy function. After being subjected to external force and deformed, it can return to the initial state after the external force is removed.

[0044] Mooney-Rivlin material is a hyperelastic material, which describes material properties based on three principal invariants of the strain tensor. The first invariant is related to the volume change of the material, and the last two invariants are related to the shape change of the material. This model is suitable for describing the nonlinear deformation of soft tissue, and the present application uses it to describe muscles and soft tissue. Mooney-Rivlin material contains a finite number of terms of coefficients, usually including a first-order term and a second-order term. The model considers the linear approximation of the material stress-strain curve at the initial stage, and simulates the material behavior in the nonlinear region by including higher-order terms.

[0045] In one embodiment, the above step S4: constructing finite elements to simulate facial deformation, and adding contact constraints between different tissues of the three-dimensional face physiological model to simulate the pressure and friction between different tissues, and adding boundary conditions between the bones and the facial soft tissue to constrain the mutual motion between the rigid body and the soft body, specifically includes:

[0046] Step S41: setting rigid body constraints as boundary conditions to constrain the interaction between the bones and the muscles and soft tissue, simulating the constraint relationship between the soft tissue and its bones;

[0047] In the present application, facial deformation is simulated by finite element method. The structural stability of the face model is mainly provided by the skull, and appropriate constraints are applied to the skull in the model setting. The movement of the mandible is defined as rotation and freedom to simulate the opening and closing of the mouth. The muscles and part of the soft tissue are fixed to the bones by attachment. The muscles and the adjacent soft tissue contact each other during movement, and contact constraints are added in the model, which will simulate the pressure and friction between different tissues, ensure the correct transmission of force and the authenticity of biomechanical behavior. The experiment of the embodiment of the present application is carried out in the open source finite element solver FEBIO, which uses global Cartesian coordinate system to model and simulate the face model. The anatomically accurate three-dimensional face physiological model established in step S1 is meshed in step S2. The completed model can be directly imported into FEBIO for biomechanical modeling in step S3. The definition of material is realized by writing material plug-in. The skull is fixed, that is, any rotation and displacement is not allowed in the coordinate system. The mandible can produce displacement in the y-axis and z-axis directions, and displacement is not allowed in the x-axis, that is, the mandible has the freedom of rotation along the x-axis. It is realized by defining the hinge structure between the skull and the mandible which is defined as a rigid body. As shown in Fig. 2 is a schematic diagram of the hinge point, in which the blue part represents the mandible, and the hinge point position of the mandible and skull is circled and pointed by an arrow. Figure 4

[0048] ​Step S42: Construct contact constraints using the bonded contact method for the contact surfaces between muscles and soft tissues and between muscles; the contact surfaces do not allow relative sliding or separation, ensuring continuous transmission of forces and displacements in the contact area; the bonded contact relies on the enhanced Lagrangian method to precisely control the relative displacement and rotation between the contact interfaces by setting a gap function, which is defined as the distance from the slave surface to the nearest point on the master surface and ensures that the calculation is only performed once in the entire material reference frame; the gap function considers full directional displacement control, not limited to the normal direction, to ensure full fitting of the contact surface in all directions; through the gap function, the virtual work expression of the sutured contact is derived, in which the reaction force is implemented by combining the Lagrange multiplier and an additional penalty factor to impose the constraint that the gap is zero; the gap function is used to define the distance between the master surface and the slave surface, and the gap function is defined as:

[0049] ;

[0050] wherein, represents the master surface; represents the deformation function of the master surface; represents the deformation function of the corresponding slave surface; represents the projection of the slave node to the master surface;

[0051] The reaction force work equation of the bonded contact is defined as:

[0052] ;

[0053] wherein, represents the contact boundary of the master surface and the slave surface; represents the Lagrange multiplier; represents the change of the constraint function; represents the differential of the boundary.

[0054] In the present invention, the interaction between rigid and soft bodies is constrained for the boundary conditions between the skull and the facial soft tissue. The skull provides support and attachment points for the facial soft tissue, which is achieved through rigid body constraints. Specifically, the interaction between the rigid body (skull) and the soft body (muscles and other soft tissues) is achieved through rigid body constraints, ensuring that the soft tissue accurately reflects its constraint relationship with the skeleton during simulation. The rigid body constraint does not distribute any degrees of freedom at the end of the rigid body, and the motion or static position of the rigid body will directly affect the soft body connected to it. The soft body is able to closely follow the motion trajectory of the rigid body when subjected to external forces or stress and deformation due to bone movement. In addition, the contact surface between the rigid body and the soft body is usually realized by point connection elements in the numerical model. These methods allow the rigid body and the soft body to maintain the appropriate relative position when subjected to force, while allowing a certain range of sliding or separation to simulate the real physiological movement. The constraints between soft bodies include the direct contact between soft tissues and muscles and between muscles. The present invention uses a tied-elastic method to describe the contact surface between two or more objects. These contact surfaces do not allow relative sliding or separation. This type of contact can ensure the continuous transmission of force and displacement in the contact area. Tied-elastic contact relies on the enhanced Lagrangian method to accurately control the relative displacement and rotation between the contact interfaces by setting a gap function. The gap function is defined as the distance from the dependent surface to the nearest point on the master surface, and ensures that it is only calculated once in the entire material reference frame. This function considers displacement control in all directions, not just the normal direction, to ensure full fitting of the contact surface in all directions. Through the gap function, the virtual work expression of the tied-elastic contact can be derived, where the reaction force is implemented by combining the Lagrange multiplier and an additional penalty factor to enforce the constraint of zero gap.

[0055] In one embodiment, the step S5 of solving and analyzing the finite element to obtain a three-dimensional face animation specifically includes:

[0056] The finite element method (FEM) simulates the geometry and load conditions of a real physical system by mathematical approximation, and approximates the infinite real system with a limited number of unknowns by connecting simple elements to each other. By simplifying complex problems into simple problems for solving, the solution domain is divided into a number of interconnected small subdomains (finite elements), and an approximate solution is assumed for each element, and the approximate solution of the entire domain is derived. The main steps of using finite element analysis include pre-processing, assembly solving, and post-processing. Pre-processing involves establishing a mathematical model to approximate the actual problem. It mainly includes the following contents:

[0057] 1) Define the geometric region, determine the solution domain / given entity model;

[0058] 2) Select the appropriate element type, tetrahedral meshing model;

[0059] 3) specify material properties, such as elastic modulus and density;

[0060] 4) determine the geometry of the elements, such as length and area;

[0061] 5) set the connectivity between elements, ensuring the consistency of the overall structure;

[0062] 6) select the basis functions, used to approximate the expression of the solution;

[0063] 7) apply boundary conditions and external loads, simulating the actual situation.

[0064] Figure 5 The finite element solution flowchart is shown.

[0065] The data obtained by processing the foregoing steps in the Febio platform is stored in an XML file. The data of the assembly solution stage is stored in a log file, and the final result is stored in an xplt file and can be post-processed and analyzed.

[0066] The following will illustrate the synthesis of the viseme animation corresponding to the Chinese phonemes of the human face and the human face action animation using the method proposed in the application. By setting a traction force of a specific direction and size on the surface of the muscle, an experiment is performed in the FEBIO open-source finite element. The experimental environment is as shown in Table 1:

[0067] Table 1 Experimental environment configuration table

[0068]

[0069] Figure 6 is the key frame animation of the synthesis result of the final / ao / , a total of 12 frames of animation, and the total time is 3 min.

[0070] Figure 7 is the key frame animation of the synthesis result of the pucker action, a total of 15 frames of animation, and the total time is 5 min.

[0071] Example Two

[0072] As shown in Figure 8 , the embodiment of the application provides a three-dimensional human face animation generation system based on physiological priori, which comprises the following modules:

[0073] The three-dimensional human face physiological model construction module 61 is used to obtain an initial three-dimensional human face model based on tomographic images of CT and MRI scanning; the initial three-dimensional human face model is simplified into a three-layer structure of an exoskeleton-muscle-soft tissue, and the three-layer structure is corrected, meshed and filled to obtain a three-dimensional human face physiological model;

[0074] The mesh partitioning module 62 is configured to implement unstructured tetrahedral partitioning on the three-dimensional face physiological model using an unstructured Delaunay algorithm to obtain a mesh partitioning result of the three-dimensional face physiological model.

[0075] The face biomechanical modeling module 63 is configured to perform biomechanical modeling on the mesh partitioning result of the three-dimensional face physiological model, define material properties at various levels, describe the skeleton as a rigid body, and describe the muscles and soft tissues as a hyperelastic model Mooney-Rivlin.

[0076] The boundary condition and constraint setting module 64 is configured to construct a finite element to simulate face deformation, add contact constraints between different tissues of the three-dimensional face physiological model to simulate pressure and friction between different tissues, and add boundary conditions between the skeleton and the soft tissues of the face to constrain the mutual movement between the rigid body and the soft body.

[0077] The finite element analysis module 65 is configured to solve and analyze the finite element to obtain a three-dimensional face animation.

[0078] A three-dimensional face animation generation device based on physiological priors, comprising one or more electronic devices, wherein the one or more electronic devices are configured to implement a three-dimensional face animation generation method, system and device based on physiological priors.

[0079] An electronic device, comprising: one or more processors; a memory configured to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors are caused to implement a three-dimensional face animation generation method, system and device based on physiological priors.

[0080] A computer-readable storage medium having stored thereon executable instructions that, when executed by a processor, cause the processor to implement a three-dimensional face animation generation method, system and device based on physiological priors.

[0081] A non-transitory computer-readable storage medium having stored thereon a computer program that, when executed by a processor, implements a three-dimensional face animation generation method, system and device based on physiological priors.

[0082] The above description is merely one specific implementation of the present application, which enables a person skilled in the art to understand or implement the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application should not be limited to the embodiments shown herein, but should be consistent with the widest scope consistent with the principles and novel features set forth herein.

Claims

1. A method for generating 3D face animation based on physiological priors, characterized in that, include: Step S1: Obtain an initial 3D face model based on tomographic images from CT and MRI scans; The initial 3D face model is simplified into a three-layer structure of exoskeleton-muscle-soft tissue, and then corrected, meshed, and filled to obtain a 3D face physiological model. Step S2: Use the unstructured Delaunay algorithm to perform unstructured tetrahedral subdivision on the three-dimensional facial physiological model to obtain the mesh subdivision result of the three-dimensional facial physiological model; Step S3: Perform biomechanical modeling on the mesh subdivision results of the three-dimensional human face physiological model, define the material properties of each layer, use rigid bodies to describe the skeleton, and use the hyperelastic model Mooney-Rivlin to describe the muscles and soft tissues. Step S4: Construct a finite element method to simulate facial deformation, and add contact constraints between different tissues in the three-dimensional facial physiological model to simulate the pressure and friction between different tissues. Add boundary conditions between the bones and facial soft tissues to constrain the mutual motion between rigid and soft bodies. Specifically, this includes: Step S41: Set rigid body constraints as boundary conditions to constrain the interaction between bones and muscles and soft tissues, and simulate the constraint relationship between soft tissues and bones. Step S42: A binding contact method is used to construct contact constraints for the contact surfaces between muscles and soft tissues, as well as between muscles. These contact surfaces are not allowed to slide or separate relative to each other, ensuring continuous transmission of force and displacement within the contact area. The binding contact relies on an enhanced Lagrangian method, precisely controlling the relative displacement and rotation between the contact interfaces by setting a gap function. This gap function is defined as the nearest point distance from the subordinate surface to the master surface, and is calculated only once throughout the entire material reference frame. The gap function considers omnidirectional displacement control, not just in the normal direction, to ensure complete contact surface fit in all directions. Using the gap function, a virtual work expression for the suture contact is derived, where the reaction force is implemented with a zero gap constraint through a combination of Lagrangian multipliers and an additional penalty factor. The gap function is used to define the distance between the master and subordinate surfaces, and is defined as follows: ； in, Indicates the main surface; The deformation function of the main surface; Represents the deformation function of the corresponding subordinate surface; This represents the projection from the subordinate node to the main surface; The equation for the work done by the reaction force in the bonded contact is defined as: ； in, This indicates the contact boundary between the primary and secondary surfaces; Represents the Lagrange multipliers; This indicates the change in the constraint function; The differential representing the boundary; Step S5: Solve and analyze the finite element method to obtain a three-dimensional face animation.

2. The method for generating 3D face animation based on physiological priors according to claim 1, characterized in that, Step S2: Using the unstructured Delaunay algorithm to perform unstructured tetrahedral subdivision of the 3D facial physiological model, the mesh subdivision result of the 3D facial physiological model is obtained, specifically including: High-quality tetrahedrons can be obtained by utilizing the empty circle property and the maximum and minimum angle properties of the Delaunay algorithm. The unstructured meshing can adapt to complex human face shapes. Therefore, the unstructured Delaunay algorithm is used to balance the mesh shape and the computational load in the finite element analysis stage.

3. The method for generating 3D face animation based on physiological priors according to claim 1, characterized in that, Step S3: Perform biomechanical modeling on the mesh subdivision results of the 3D human face physiological model, define the material properties at each level, use rigid bodies to describe the skeleton, and use the Mooney-Rivlin hyperelastic model to describe the muscles and soft tissues, specifically including: The stress-strain characteristics of soft tissues and muscles are defined as nonlinear, incompressible, and isotropic. Since Mooney-Rivlin is a hyperelastic material with nonlinear and isotropic stress-strain characteristics, a strain energy function is constructed based on Mooney-Rivlin to describe these stress-strain characteristics. ； in, Represents the strain energy function. and The first and second invariants represent the strain tensor. and These are the material parameters of the model, which can be determined experimentally; Indicates the elastic volume ratio; This represents temperature-related parameters; when used to represent incompressible materials, Poisson's ratio is 0.

5. .

4. A 3D face animation generation system based on physiological priors, characterized in that, Includes the following modules: A three-dimensional facial physiological model module is constructed to obtain an initial three-dimensional facial model based on tomographic images from CT and MRI scans. The initial three-dimensional facial model is simplified into a three-layer structure of exoskeleton-muscle-soft tissue, and then corrected, meshed, and filled to obtain a three-dimensional facial physiological model. The mesh generation module is used to perform unstructured tetrahedral mesh generation on the three-dimensional facial physiological model using the unstructured Delaunay algorithm, thereby obtaining the mesh generation result of the three-dimensional facial physiological model. The facial biomechanical modeling module is used to perform biomechanical modeling on the mesh subdivision results of the three-dimensional facial physiological model, defining the material properties at each level, using rigid bodies to describe the skeleton, and using the hyperelastic model Mooney-Rivlin to describe the muscles and soft tissues. The boundary conditions and constraints module is used to construct finite element models to simulate facial deformation. Contact constraints are added between different tissues in the 3D facial physiological model to simulate pressure and friction between different tissues. Boundary conditions are added between the bones and facial soft tissues to constrain the mutual motion between rigid and soft bodies. Specifically, this includes: Step S41: Set rigid body constraints as boundary conditions to constrain the interaction between bones and muscles and soft tissues, and simulate the constraint relationship between soft tissues and bones. Step S42: A binding contact method is used to construct contact constraints for the contact surfaces between muscles and soft tissues, as well as between muscles. These contact surfaces are not allowed to slide or separate relative to each other, ensuring continuous transmission of force and displacement within the contact area. The binding contact relies on an enhanced Lagrangian method, precisely controlling the relative displacement and rotation between the contact interfaces by setting a gap function. This gap function is defined as the nearest point distance from the subordinate surface to the master surface, and is calculated only once throughout the entire material reference frame. The gap function considers omnidirectional displacement control, not just in the normal direction, to ensure complete contact surface fit in all directions. Using the gap function, a virtual work expression for the suture contact is derived, where the reaction force is implemented with a zero gap constraint through a combination of Lagrangian multipliers and an additional penalty factor. The gap function is used to define the distance between the master and subordinate surfaces, and is defined as follows: ； in, Indicates the main surface; The deformation function of the main surface; Represents the deformation function of the corresponding subordinate surface; This represents the projection from the subordinate node to the main surface; The equation for the work done by the reaction force in the bonded contact is defined as: ； in, This indicates the contact boundary between the primary and secondary surfaces; Represents the Lagrange multipliers; This indicates the change in the constraint function; The differential representing the boundary; The finite element analysis module is used to solve and analyze the finite element to obtain a three-dimensional face animation.

5. A three-dimensional face animation generation device based on physiological priors, characterized in that, It includes one or more electronic devices, wherein the one or more electronic devices are used to implement the method of any one of claims 1 to 3.

6. An electronic device, characterized in that, include: One or more processors; A memory for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors cause the one or more processors to implement the method of any one of claims 1 to 3.

7. A computer-readable storage medium, characterized in that, It stores executable instructions that, when executed by a processor, cause the processor to implement the method described in any one of claims 1 to 3.

8. A non-transitory computer-readable storage medium, characterized in that, It stores a computer program that, when executed by a processor, implements the steps of the method as described in any one of claims 1 to 3.

Images