Virtual character model matching method and device, medium, equipment and program product

Abstract

The embodiment of the invention provides a virtual character model matching method and device, a storage medium, equipment and a program product, and is applied to a virtual character model migration scene and a 3D reloading scene. According to the virtual character model matching method provided by the embodiment of the invention, a reference virtual character model and a target texture are obtained; determining feature points to be matched on the reference virtual character model and feature points to be matched on the target texture based on the corresponding relation marked on the reference virtual character model and the target texture; matching the to-be-matched feature points on the reference virtual character model with the to-be-matched feature points on the target texture based on a target matching algorithm to obtain matched feature points on the target texture; and obtaining a target virtual character model matched with the target texture according to the matched feature points on the target texture. According to the virtual character model matching method provided by the embodiment of the invention, the target matching algorithm is designed in a targeted manner, so that the accuracy of a model matching scheme in the traditional technology is improved.

Description

Technical Field

[0001] This application relates to the field of Internet technologies, and particularly to a method, apparatus, storage medium, device, and program product for matching virtual character models. Background Art

[0002] Model matching solutions can match a given model to a target model. However, traditional model matching solutions are difficult to register a given model to a target model with significant differences, or manual adjustment of the given model is required to obtain an initial matching result, and the operation is relatively complex. Summary of the Invention

[0003] Embodiments of this application provide a method, apparatus, storage medium, program product, and computer device for matching virtual character models, which improve the accuracy of the model matching solution and ensure that a given model can be registered to a target model with significant differences.

[0004] On the one hand, embodiments of this application provide a method for matching virtual character models, the method including: obtaining a reference virtual character model and a target texture; determining feature points to be matched on the reference virtual character model and feature points to be matched on the target texture based on the corresponding relationships marked on the reference virtual character model and the target texture; matching the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture based on a target matching algorithm to obtain the feature points after matching on the target texture; obtaining a target virtual character model that matches the target texture according to the feature points after matching on the target texture; wherein the target matching algorithm matches the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture through the marked corresponding relationships, and iteratively optimizes the matching result based on a target loss function, and the target loss function obtained by iterative optimization realizes the minimization of the matching loss between the corresponding feature points of the reference virtual character model and the target texture, and the minimization of the matching loss between the corresponding feature points of the reference virtual character model and the corresponding feature surface of the target texture.

[0005] On the other hand, an embodiment of the present application provides a model matching device, which includes a first acquisition module, a first processing module, a second processing module, and a third processing module. The first acquisition module is configured to acquire a reference virtual character model and a target texture. The first processing module is configured to determine the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture based on the corresponding relationship marked on the reference virtual character model and the target texture. The second processing module is configured to match the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture based on a target matching algorithm to obtain the feature points after matching on the target texture. The third processing module is configured to obtain a target virtual character model that matches the target texture according to the feature points after matching on the target texture; wherein, the target matching algorithm matches the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture through the marked corresponding relationship, and iteratively optimizes the matching result based on a target loss function, and the target loss function obtained by the iterative optimization realizes the minimization of the matching loss between the corresponding feature points of the reference virtual character model and the target texture, and the minimization of the matching loss from the corresponding feature points of the reference virtual character model to the corresponding feature surface of the target texture.

[0006] In some embodiments, the model matching device further includes a display module. The display module is configured to: display a model preview panel; in response to an import instruction for the reference virtual character model and the target texture, display the reference virtual character model and the target texture on the model preview panel; in response to a corresponding relationship marking event for the reference virtual character model and the target texture in the model preview panel, display the corresponding relationship marked on the reference virtual character model and the target texture, and the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture determined according to the marked corresponding relationship on the model preview panel; in response to a registration event for the reference virtual character model and the target texture in the model preview panel, display the matching result of matching the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture based on the target matching algorithm, and the result of each iterative optimization of the matching result, wherein the final result of the iterative optimization of the matching result is the feature points after matching on the target texture; obtain a target virtual character model that matches the target texture according to the feature points after matching on the target texture, and display the target virtual character model that matches the target texture on the model preview panel.

[0007] In some embodiments, the first processing module is configured to label corresponding curves on the reference virtual character model and the target texture through a wiring selection plug-in; uniformly sample the corresponding curves labeled on the reference virtual character model and the target texture to obtain the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture.

[0008] In some embodiments, the first processing module is configured to select a curve to be labeled on the reference virtual character model; on the curve to be labeled, select the starting point as the initial node, select intermediate nodes along the direction from the starting point to the ending point, and select the ending point as the terminating node; according to the initial node, intermediate nodes, and terminating node selected on the curve to be labeled, label the corresponding curve of the curve to be labeled on the target texture.

[0009] In some embodiments, the first processing module is configured to divide the corresponding curves labeled on the reference virtual character model and the target texture into n pairs of equally long curve segments, where n is a positive integer greater than 1; determine n + 1 pairs of points to be calibrated according to the endpoints of the n pairs of equally long curve segments; calibrate the n + 1 pairs of points to be calibrated on the reference virtual character model and the target texture to obtain the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture.

[0010] In some embodiments, the second processing module is configured to calculate the matching loss between the corresponding feature points of the reference virtual character model and the target texture, and the matching loss of the corresponding feature points of the reference virtual character model to the corresponding feature surface of the target texture to determine the target loss function; optimize the global displacement, joint poses, human body shape, and vertices of the reference virtual character model through an optimizer until the target loss function converges.

[0011] In some embodiments, the second processing module is configured to keep the human body shape of the reference virtual character model fixed and optimize the global displacement and joint poses of the reference virtual character model through an optimizer until the target loss function converges; optimize the global displacement, joint poses, and human body shape of the reference virtual character model through an optimizer until the target loss function converges; keep the global displacement, joint poses, and human body shape of the reference virtual character model fixed and optimize the vertices of the reference virtual character model until the target loss function converges.

[0012] In some embodiments, the second processing module is configured to keep the global displacement, joint poses, and human body shape of the reference virtual character model unchanged, and use the neural Jacobian field to optimize the shape Jacobian matrix of the triangular patches on the reference virtual character model, so as to optimize the vertices of the reference virtual character model until the target loss function converges.

[0013] In some embodiments, the model matching device further includes a second acquisition module and a fourth processing module. The second acquisition module is configured to acquire a clothing model fitted on the reference virtual character model. The fourth processing module is configured to determine the clothing model fitted on the target virtual character model according to the target virtual character model that matches the target texture and the clothing model fitted on the reference virtual character model.

[0014] On the other hand, an embodiment of the present application provides a computer-readable storage medium. The computer-readable storage medium includes a stored program. When the program is run by a processor, it executes the virtual character model matching method described in any of the above embodiments.

[0015] On the other hand, an embodiment of the present application provides a computer device. A computer program is stored in the memory, and the processor is configured to execute the virtual character model matching method described in any of the above embodiments through the computer program.

[0016] On the other hand, an embodiment of the present application provides a computer program product, including computer programs / instructions. When the computer programs / instructions are executed by a processor, they implement the virtual character model matching method described in any of the above embodiments.

[0017] In the embodiments of the present application, a reference virtual character model and a target texture are obtained; based on the corresponding relationships marked on the reference virtual character model and the target texture, the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture are determined; based on a target matching algorithm, the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture are matched to obtain the feature points after matching on the target texture; according to the feature points after matching on the target texture, a target virtual character model matching the target texture is obtained; wherein, the target matching algorithm matches the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture through the marked corresponding relationships, and iteratively optimizes the matching result based on a target loss function, and the target loss function obtained by iterative optimization realizes the minimization of the matching loss between the corresponding feature points of the reference virtual character model and the target texture, and the minimization of the matching loss between the corresponding feature points of the reference virtual character model and the corresponding feature surface of the target texture. The matching method of the virtual character model provided by the embodiments of the present application specifically designs a target matching algorithm to improve the accuracy of the model matching scheme in the traditional technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without creative efforts.

[0019] Figure 1 It is a schematic flowchart of the matching method of the virtual character model provided by the embodiments of the present application.

[0020] Figure 2 It is a schematic diagram of the reference virtual character model and the target texture provided by the embodiments of the present application.

[0021] Figure 3 It is a schematic diagram of the SMPL model provided by the embodiments of the present application.

[0022] Figure 4 It is a schematic flowchart of the process of marking the corresponding curves of the reference virtual character model and the target texture provided by the embodiments of the present application.

[0023] Figure 5 It is a schematic diagram of uniform sampling on the corresponding curves marked on the reference virtual character model and the target texture provided by the embodiments of the present application.

[0024] Figure 6 It is a schematic diagram of the target virtual character model provided by the embodiments of the present application.

[0025] Figure 7 This is a schematic structural diagram of the model matching device provided by the embodiment of the present application.

[0026] Figure 8 This is a schematic structural diagram of the computer device provided by the embodiment of the present application. Specific embodiments

[0027] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope of protection of the present application.

[0028] The embodiment of the present application provides a method, device, storage medium, device and program product for matching virtual character models. Exemplarily, the method for matching virtual character models in the embodiment of the present application can be executed by a computer device, where the computer device can be a terminal or a server, etc. The terminal can be a smart phone, a tablet computer, a notebook computer, a desktop computer, a smart TV, a smart speaker, a wearable smart device, a personal computer (Personal Computer, PC), a smart vehicle terminal, etc. The terminal can also include a client, and the client can be a video client, a shopping application client, a reading application client, a browser client or an instant messaging client, etc. The server can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (Content Delivery Network, CDN), and big data and artificial intelligence platforms.

[0029] The embodiment of the present application can be applied to various model matching scenarios, including not only the human body migration scenario and the virtual character model migration scenario, but also including, but not limited to, scenarios such as cloud technology, artificial intelligence, animation production, game development, virtual character generation and personalized customization.

[0030] First, some nouns or terms that appear in the process of describing the embodiment of the present application are explained as follows:

[0031] Human Model Transfer technology: It mainly involves the process of transferring the characteristics (such as posture, movement, body shape, etc.) of one human model to another human model. This technology plays a crucial role in multiple application fields, especially in animation production, game development, virtual character generation, and personalized customization. During the process of spatial transformation and model transfer, it is necessary to select an appropriate transfer algorithm to ensure the reliability of model transfer. In terms of muscles, bones, facial expressions, clothing, and other details, the model may lose details or produce distortions. Human Model Transfer technology can also be used in other virtual characters, such as animals, cartoon characters, and even non-living objects. This requires considering the differences in the bone structures and movement patterns of different models. For example, the bones of quadruped animals are different from those of humans, and key point matching may need to be adjusted during transfer.

[0032] 3D Virtual Try-On technology: It allows users to try on clothes or accessories in a virtual environment. By using 3D models and human pose analysis, it simulates the effects of different clothing on users, and is usually applied to scenarios such as fashion, online shopping, and virtual fitting rooms. Virtual character model transfer technology plays a crucial role in 3D virtual try-on, and can provide users with a more personalized and natural virtual dressing experience. By combining virtual character model transfer and 3D virtual try-on technology, it can ensure that the clothing not only perfectly presents on the user's virtual model, but also provides highly accurate clothing adaptation according to the user's body shape and posture. Virtual character model transfer technology transfers the body shape and posture characteristics of the user's 3D model to a standard virtual character model, which not only ensures the adaptation of clothing to different user body shapes, but also makes the clothing perfectly integrate with the user's virtual character. For example, based on the user's body shape data and motion capture information, the virtual dressing system can adjust the fit of the clothing in real time to ensure that details such as the folds and stretches of the clothes are consistent with the user's movements.

[0033] The model matching scheme can match a given model to a target model. However, traditional model matching schemes are difficult to register a given model to a target model with significant differences, or require manual adjustment of the given model to obtain an initial matching result, and the operation is relatively complex.

[0034] During the process of model matching, virtual character model migration technology is involved. In order to improve the accuracy of the model matching scheme in traditional technologies and ensure that a given model can be registered to a target model with significant differences. The matching method of the virtual character model provided by the embodiments of this application specifically designs a target matching algorithm. The target matching algorithm matches the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture through the marked corresponding relationship, and iteratively optimizes the matching result based on the target loss function. The target loss function obtained by iterative optimization minimizes the matching loss between the corresponding feature points of the reference virtual character model and the target texture, and minimizes the matching loss from the corresponding feature points of the reference virtual character model to the corresponding feature surface of the target texture, so as to improve the accuracy of the model matching scheme in traditional technologies.

[0035] The matching method of the virtual character model provided by the embodiments of this application is specifically illustrated by the following embodiments. The following will be described in detail respectively. It should be noted that the description order of the following embodiments does not limit the priority order of the embodiments.

[0036] It can be understood that in the specific implementation of this application, relevant data such as user object data and context data are involved. When the embodiments of this application are applied to specific products or technologies, user permission or consent is required, and the collection, use, and processing of relevant data need to comply with relevant laws, regulations, and standards of relevant countries and regions.

[0037] Figure 1 It is a schematic flowchart of the matching method of the virtual character model provided by the embodiments of this application. As Figure 1 shown, the matching method of the virtual character model provided by the embodiments of this application includes:

[0038] Step 01: Obtain a reference virtual character model and a target texture.

[0039] Figure 2 It is a schematic diagram of the reference virtual character model and the target texture provided by the embodiments of this application. As Figure 2 shown, the reference virtual character model can be the original virtual character model derived from the user or other targets during the virtual character model migration process. The virtual character model can be a human body, an animal, a cartoon character, or even a non-biological entity, which contains information such as the body shape, posture, and movement of the virtual character and is the starting point of the migration process. By obtaining the reference virtual character model, the body shape and posture features can be extracted. The reference virtual character model provides the basic features for migration, and these features include height, weight, joint positions, muscle structure, etc.

[0040] The target texture refers to the surface material and appearance of the target virtual character model after migration. It includes the visual effects of human body surface details such as skin, clothing, hairstyle, etc. The target texture can be a texture map synthesized from multiple images, containing information such as color, lighting, and shadows. The accuracy of the target texture directly affects the appearance of the virtual character. Especially in applications such as virtual fitting and game character customization, the correct texture is crucial for enhancing the immersion of the virtual character.

[0041] In the process of virtual character model migration, the reference virtual character model and the target texture are closely related. The reference virtual character model provides the basic structure and dynamic information for migration, while the target texture ensures the natural appearance of the migrated virtual character model. The cooperation of the two determines the effect of the migration process.

[0042] The reference virtual character model can be an SMPL (Skinned Multi-Person Linear Model). SMPL (Skinned Multi-Person Linear model) is a model used to represent the three-dimensional human body shape and pose, widely used in the fields of computer vision, graphics, and machine learning. The main feature of the SMPL model is to model the human body as a triangular mesh with parameterized shape and pose.

[0043] Figure 3 Schematic diagram of the SMPL model provided for the embodiments of this application. As Figure 3 shown, the SMPL model describes the human body shape and pose through a set of parameters, including: body shape parameters controlling the body shape and appearance characteristics of the human body (such as tall and thin, short and fat, etc.); joint pose parameters defining the joint angles and directions of the human body; global displacement parameters of the human body defining the global movement position of the human body. Given the above parameters, the SMPL model will output a 3D mesh representing the surface skin of the human body. The mesh has 20,950 vertices, and the set of its 3D coordinates is The calculation method is:

[0044] v = SMPL(θ, β, t)

[0045] In some embodiments, the method further includes: displaying a model preview panel; in response to an import instruction for the reference virtual character model and the target texture, displaying the reference virtual character model and the target texture in the model preview panel.

[0046] Specifically, the matching method of the virtual character model provided by the embodiments of this application can be implemented on editing software. The model preview panel on the editing software can display the imported reference virtual character model and the target texture.

[0047] Step 02: Based on the corresponding relationships marked on the reference virtual character model and the target texture, determine the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture.

[0048] Specifically, the feature points to be matched on the virtual character model and the feature points to be matched on the target texture can be the vertices on the model surface or key anatomical landmarks (such as joints, facial features), or key regions defined on the model by color, shape, or semantics (such as "left shoulder", "right knee"). During human body transfer, determining the corresponding points between the reference virtual character model and the target texture is the key to achieving accurate texture mapping.

[0049] In some embodiments, Step 02: Based on the corresponding relationships marked on the reference virtual character model and the target texture, determine the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture, including:

[0050] Step 021: Mark corresponding curves on the reference virtual character model and the target texture through a wiring selection plug-in.

[0051] Specifically, a wiring selection plug-in refers to a software tool for assisting in wiring work. Especially in electronic design and 3D modeling, these plug-ins can help users complete the wiring task more efficiently. The main task of the wiring selection plug-in is to help users establish an accurate corresponding relationship between the reference virtual character model and the target texture.

[0052] Figure 4 Schematic diagram of the process for marking corresponding curves on the reference virtual character model and the target texture provided by the embodiments of the present application. As Figure 4 shown, mark some key curves on the surface of the reference virtual character model, including the spine line, joint curves of the limbs, facial contours, etc. The curves marked on the surface of the reference virtual character model correspond one-to-one with specific regions on the target texture.

[0053] In some embodiments, Step 021: Mark corresponding curves on the reference virtual character model and the target texture through a wiring selection plug-in, including: Select the curve to be marked on the reference virtual character model; On the curve to be marked, select the starting point as the initial node, select intermediate nodes along the direction from the starting point to the ending point, and select the ending point as the termination node; According to the initial node, intermediate nodes, and termination nodes selected on the curve to be marked, mark the corresponding curve on the target texture for the curve to be marked.

[0054] Specifically, as Figure 4As shown, multiple curves to be marked can be selected on the reference virtual character model, and each corresponding curve is processed one by one. When selecting a curve to be marked each time, first mark the starting point of the corresponding curve on the target texture as the initial node, then mark the intermediate points of the corresponding curve as intermediate nodes until the end point of the corresponding curve is marked as the termination node. Select a curve to be marked multiple times, and according to the initial nodes, intermediate nodes, and termination nodes selected on a curve to be marked, mark a corresponding curve on the target texture until the marking of all corresponding curves to be marked is completed.

[0055] Step 022: Uniformly sample the corresponding curves marked on the reference virtual character model and the target texture to obtain the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture.

[0056] Specifically, as Figure 4 shown, after the corresponding curves to the curves to be marked are marked on the target texture, some key feature points also need to be selected on the corresponding curves marked on the reference virtual character model and the target texture. The selected feature points can represent the specific positions of the curves, usually the key feature positions of the human body. For example, the feature points on the face can include the corners of the eyes, the tip of the nose, and the corners of the mouth, etc.

[0057] In some embodiments, Step 022: Uniformly sample the corresponding curves marked on the reference virtual character model and the target texture to obtain the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture, including: dividing the corresponding curves marked on the reference virtual character model and the target texture into n pairs of equally long curve segments, where n is a positive integer greater than 1; determining n + 1 pairs of points to be marked according to the endpoints of the n pairs of equally long curve segments; calibrating n + 1 pairs of points to be marked on the reference virtual character model and the target texture to obtain the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture.

[0058] Figure 5 This is a schematic diagram of uniform sampling on the corresponding curves marked on the reference virtual character model and the target texture provided by the embodiments of the present application. As Figure 5 shown, curve A can be a marked curve on the reference virtual character model, and curve B can be the corresponding curve marked on the target texture corresponding to curve A. The starting point of curve A corresponds to the starting point of curve B, and the end point of curve A corresponds to the end point of curve B.

[0059] The curve A and curve B can be divided into n pairs of curve segments with equal lengths, where n is a positive integer greater than 1. Among them, curve A is divided into curve segments a1 to an with equal lengths, and curve B is divided into curve segments b1 to bn with equal lengths. The n + 1 endpoints of curve segments a1 to an are all feature points to be matched on the reference virtual character model, and the n + 1 endpoints of curve segments b1 to bn are all feature points to be matched on the target texture. By uniformly sampling curve A and curve B, the corresponding points on curve A and curve B can be increased to increase the number of feature points to be matched on the reference virtual character model and the number of feature points to be matched on the target texture.

[0060] Integrate the feature points to be matched on K reference virtual character models Obtain the set s of feature points to be matched on the reference virtual character model M :

[0061]

[0062] Integrate the feature points to be matched on K target textures Obtain the set T of feature points to be matched on the target texture M :

[0063]

[0064] In some embodiments, the target matching algorithm provided by the embodiments of the present application further includes: in response to an annotation event for the correspondence between the reference virtual character model and the target texture in the model preview panel, displaying in the model preview panel the correspondence annotated on the reference virtual character model and the target texture, and the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture determined according to the annotated correspondence.

[0065] Specifically, the annotation event for the correspondence between the reference virtual character model and the target texture in the model preview panel includes the event executed in step 02 above. When executing step 02 above, the correspondence annotated on the reference virtual character model and the target texture, and the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture determined according to the annotated correspondence are displayed in the model preview panel.

[0066] When using the wiring selection plug-in to annotate the corresponding curves on the reference virtual character model and the target texture, the curves to be annotated on the reference virtual character model and the corresponding curves annotated on the target texture are displayed in the model preview panel.

[0067] When uniformly sampling the corresponding curves labeled on the reference virtual character model and the target texture to obtain the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture, the model preview panel displays the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture.

[0068] Step 03: Match the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture based on the target matching algorithm to obtain the feature points after matching on the target texture. Among them, the target matching algorithm matches the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture through the labeled correspondence, and iteratively optimizes the matching result based on the target loss function. The target loss function obtained by iterative optimization realizes the minimization of the matching loss between the corresponding feature points of the reference virtual character model and the target texture, and the minimization of the matching loss between the corresponding feature points of the reference virtual character model and the corresponding feature surface of the target texture.

[0069] Specifically, the core of the target matching algorithm is to measure the matching accuracy through a loss function and adjust the positions of the feature points through iterative optimization until the best matching effect is achieved. The loss function can be designed according to factors such as the distortion degree between the reference virtual character model and the target texture and the alignment degree of the curves. Through multiple iterations, the algorithm gradually optimizes the matching result so that the texture of the subsequent obtained target virtual character model can fit the surface of the reference virtual character model as naturally as possible.

[0070] In some embodiments, Step 03: Match the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture based on the target matching algorithm to obtain the feature points after matching on the target texture, including:

[0071] Step 031: Calculate the matching loss between the corresponding feature points of the reference virtual character model and the target texture, and the matching loss between the corresponding feature points of the reference virtual character model and the corresponding feature surface of the target texture to determine the target loss function.

[0072] Specifically, the target loss function is:

[0073] E = E point2plane + E match

[0074] Among them, among them, E point2plane is the matching loss between the corresponding feature points of the reference virtual character model and the corresponding feature surface of the target texture, and E match is the matching loss between the corresponding feature points of the reference virtual character model and the target texture.

[0075] To ensure the comprehensiveness of the data, in addition to the feature points in the set of feature points to be matched on the above-mentioned reference virtual character model and the set of feature points to be matched on the target texture, random sampling can also be performed additionally on the reference virtual character model and the target texture, and n 1 sampling points x i on the reference virtual character model are combined to obtain the set of sampling points S on the reference virtual character model:

[0076] S = {x i | i = 1,..., n 1}

[0077] Combining n 2 sampling points y j on the target texture, the set of sampling points T on the target texture:

[0078] T = {y j | j = 1,..., n 2}

[0079] By calculating through the comprehensive set of sampling points T on the target texture, the normal vectors of n 2 sampling points on the target texture can be obtained to obtain the set N T :

[0080]

[0081] The definition of E point2plane is:

[0082]

[0083] where min represents the nearest point query operation. The definition of E match is:

[0084]

[0085] Step 032: Optimize the global displacement, joint poses, human body shape, and vertices of the reference virtual character model through an optimizer until the target loss function converges.

[0086] Specifically, the Adam optimizer can be used as the optimizer. The Adam optimizer is an optimization algorithm commonly used in training deep learning models, which combines the advantages of momentum and adaptive learning rate. The momentum method determines the direction of the current gradient by using the exponentially weighted average of past gradients, thereby being able to accelerate convergence and reduce oscillations.

[0087] The Adam optimizer solves the problem of using a fixed learning rate in traditional optimizers by adaptively adjusting the learning rates of each parameter in the global displacement, joint poses, body shape, and vertices of the reference virtual character model. In each update, Adam dynamically adjusts the learning rate of each parameter based on the historical gradient information of that parameter.

[0088] In some embodiments, step 032: optimizing the global displacement, joint poses, body shape, and vertices of the reference virtual character model by an optimizer until the objective loss function converges, includes: fixing the body shape of the reference virtual character model unchanged, and optimizing the global displacement and joint poses of the reference virtual character model by the optimizer until the objective loss function converges.

[0089] Specifically, first fix β unchanged and only optimize the global displacement t and the human joint poses θ until the objective loss function converges:

[0090] Minimize {θ，t} E

[0091] Optimize the global displacement, joint poses, and body shape of the reference virtual character model by an optimizer until the objective loss function converges.

[0092] Specifically, optimize the global displacement t, the human joint poses θ, and the body shape β by the optimizer until the objective loss function converges:

[0093] Minimize {θ，β，t} E

[0094] Fix the global displacement, joint poses, and body shape of the reference virtual character model unchanged, and optimize the vertices of the reference virtual character model until the objective loss function converges.

[0095] Specifically, fix t, θ, β, adjust the vertices v of the reference virtual character model, and use the Adam optimizer to optimize until the objective loss function converges:

[0096] Minimize {v} E

[0097] In some embodiments, fix the global displacement, joint poses, and body shape of the reference virtual character model unchanged, and optimize the vertices of the reference virtual character model until the objective loss function converges, includes: fix the global displacement, joint poses, and body shape of the reference virtual character model unchanged, and use the neural Jacobian field to optimize the shape Jacobian matrix of the upper triangular patches of the reference virtual character model to achieve optimizing the vertices of the reference virtual character model until the objective loss function converges.

[0098] During the process of model matching, smoothness is very important. Smoothness ensures the visual naturalness and consistency of the model. If the vertices v of the reference virtual character model are directly optimized, it may cause large changes in the shape of local areas, resulting in unnatural seams, distortions, or lack of smooth transitions. Therefore, in order to ensure the smoothness of the model during adjustment, it is necessary to avoid directly optimizing the vertex positions and instead adopt some more indirect and controlled methods.

[0099] During the optimization process, the vertices can be optimized by referring to the triangular meshes of the reference virtual character model, so as to ensure that the overall posture and structure of the human body conform to biological laws, prevent sharp changes in vertices, and thus maintain the smooth transition of the virtual character model. In the optimization of triangular meshes, the Jacobian Field is a key mathematical tool. It describes the influence of local transformations on the shape of triangular meshes. By calculating this matrix, it is possible to understand how each vertex changes after an affine or non-linear transformation, and then optimize the vertex positions to ensure the smoothness of the model. The optimization process of the Jacobian matrix usually involves how to adjust the overall shape of the model while preserving local features.

[0100] The Neural Jacobian Field is a deep learning method used to adjust the shape of triangular meshes during the optimization process. It uses a neural network to model the changes in the Jacobian matrix, especially by learning the mutual relationships and deformation patterns between each vertex, so that the optimization process is more efficient and accurate.

[0101] The set of Jacobian matrices of triangular meshes on the reference virtual character model can be denoted as:

[0102]

[0103] The optimization function for optimizing the Jacobian matrix of triangular meshes using the Neural Jacobian Field is:

[0104]

[0105] where, |t i | is the area of each triangle, is the Jacobian matrix of the Jacobian matrix triangle. After optimizing and adjusting the overall shape of the model, the vertices v of the adjusted reference virtual character model can be obtained, and then optimized through the Adam optimizer until the target loss function converges.

[0106] Minimize {v} E

[0107] In some embodiments, the target matching method further includes: in response to a registration event of a reference virtual character model and a target texture in a model preview panel, displaying in the model preview panel a matching result of matching feature points to be matched on the reference virtual character model and feature points to be matched on the target texture based on a target matching algorithm, and a result of iterative optimization of the matching result each time, wherein the final result of iterative optimization of the matching result is the feature points after matching on the target texture.

[0108] Specifically, based on the target matching algorithm, multiple matches can be performed on the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture. The model preview panel can display the matching result obtained each time of matching, and the result of iterative optimization of the matching result each time.

[0109] Step 04: Obtain a target virtual character model that matches the target texture according to the feature points after matching on the target texture.

[0110] Specifically, after registering the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture, a target virtual character model that matches the target texture can be obtained. Through multiple iterations, the algorithm gradually optimizes the matching result, and the texture of the target virtual character model can fit the surface of the reference virtual character model as naturally as possible. Even in the case where the initial reference virtual character model and the target texture differ greatly, there is no need to manually adjust the given model, and by means of iterative optimization, a target virtual character model that may match the reference virtual character model can be obtained, achieving the best matching effect.

[0111] Figure 6 This is a schematic diagram of the target virtual character model provided by the embodiments of the present application. As Figure 6 shown, in some embodiments, the matching method of the virtual character model further includes: displaying in the model preview panel a target virtual character model that matches the target texture.

[0112] In some embodiments, the method further includes: obtaining a clothing model fitted on the reference virtual character model.

[0113] Specifically, the clothing model on the reference virtual character model can be that the user tries on clothes or accessories in a virtual environment to simulate the effects of different clothes on the user, and is usually applied to scenarios such as fashion, online shopping, and virtual fitting rooms.

[0114] Determine a clothing model fitted on the target virtual character model according to the target virtual character model that matches the target texture and the clothing model fitted on the reference virtual character model.

[0115] Specifically, the clothing model on the reference virtual character model is migrated to the target virtual character model through the virtual character model migration technology, which not only ensures the adaptation of the clothing to different user body types, but also perfectly integrates the clothing with the user's virtual character. For example, based on the user's body type data and motion capture information, the virtual clothing change system can adjust the fitting degree of the clothing in real time to ensure that details such as the folds and stretches of the clothes are consistent with the user's actions.

[0116] In some embodiments, the matching method of the virtual character model provided in the embodiments of the present application can also be used for surgical navigation and precise positioning. The specific area (the organ that needs to participate in the surgical treatment) on the reference virtual character model is migrated to the target virtual character model (the model dedicated to medical equipment) through the virtual character model migration technology, ensuring that the surgeon can more accurately position the internal structure of the human body, thereby improving the success rate and efficiency of the surgery.

[0117] In some embodiments, the matching method of the virtual character model provided in the embodiments of the present application can also be used for the design of humanoid robots. In order to simulate the natural movement of the human body, the human body migration technology uses sensors and motion capture systems (such as optical sensors and inertial sensors) to capture the actions of the human body in real time. These sensors can be installed at key parts of the human body to record the movement and position changes of each joint. By collecting a large amount of motion data, the complex motion patterns of the human body can be accurately simulated. The reference virtual character model is migrated to the target virtual character model through the virtual character model migration technology, ensuring that the designed robot model and the image of the reference virtual character model are perfectly integrated.

[0118] All the above technical solutions can be combined arbitrarily to form alternative embodiments of the present application, which will not be elaborated here one by one. Through the implementation of the above technical solutions of the present invention, the accuracy of the model matching scheme in the traditional technology can be improved, ensuring that a given model can be registered to a target model with significant differences. Even when there is a large difference between the initial reference virtual character model and the target texture, there is no need to manually adjust the given model, and the target virtual character model that may match the reference virtual character model can be obtained through iterative optimization, achieving the best matching effect.

[0119] In the embodiments of the present application, a reference virtual character model and a target texture are obtained; based on the corresponding relationships marked on the reference virtual character model and the target texture, the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture are determined; based on a target matching algorithm, the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture are matched to obtain the feature points after matching on the target texture; according to the feature points after matching on the target texture, a target virtual character model matching the target texture is obtained; wherein, the target matching algorithm matches the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture through the marked corresponding relationships, and iteratively optimizes the matching result based on a target loss function, and the target loss function obtained by iterative optimization realizes the minimization of the matching loss between the corresponding feature points of the reference virtual character model and the target texture, and the minimization of the matching loss from the corresponding feature points of the reference virtual character model to the corresponding feature surface of the target texture. The matching method for the virtual character model provided by the embodiments of the present application specifically designs a target matching algorithm to improve the accuracy of the model matching scheme in the traditional technology.

[0120] To facilitate better implementation of the matching method for the virtual character model in the embodiments of the present application, the embodiments of the present application further provide a model matching device 400. Figure 7 FIG. is a schematic structural diagram of the model matching device 400 provided by the embodiments of the present application. Wherein, the model matching device 400 includes a first acquisition module 411, a first processing module 421, a second processing module 422, and a third processing module 423. The first acquisition module 411 is configured to acquire a reference virtual character model and a target texture. The first processing module 421 is configured to determine the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture based on the corresponding relationships marked on the reference virtual character model and the target texture. The second processing module 422 is configured to match the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture based on a target matching algorithm to obtain the feature points after matching on the target texture. The third processing module 423 is configured to obtain a target virtual character model matching the target texture according to the feature points after matching on the target texture; wherein, the target matching algorithm matches the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture through the marked corresponding relationships, and iteratively optimizes the matching result based on a target loss function, and the target loss function obtained by iterative optimization realizes the minimization of the matching loss between the corresponding feature points of the reference virtual character model and the target texture, and the minimization of the matching loss from the corresponding feature points of the reference virtual character model to the corresponding feature surface of the target texture.

[0121] In some embodiments, the model matching device 400 further includes a display module 430. The display module 430 is configured to: display a model preview panel; in response to an import instruction for a reference virtual character model and a target texture, display the reference virtual character model and the target texture on the model preview panel; in response to a correspondence annotation event for the reference virtual character model and the target texture on the model preview panel, display the correspondence annotated on the reference virtual character model and the target texture, and the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture determined according to the annotated correspondence; in response to a registration event for the reference virtual character model and the target texture on the model preview panel, display the matching result of matching the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture based on the target matching algorithm, and the result of iterative optimization of the matching result each time, wherein the final result of iterative optimization of the matching result is the feature points after matching on the target texture; obtain a target virtual character model that matches the target texture according to the feature points after matching on the target texture, and display the target virtual character model that matches the target texture on the model preview panel.

[0122] In some embodiments, the first processing module 421 is configured to annotate corresponding curves on the reference virtual character model and the target texture through a wiring selection plug-in; uniformly sample the corresponding curves annotated on the reference virtual character model and the target texture to obtain the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture.

[0123] In some embodiments, the first processing module 421 is configured to select a curve to be annotated on the reference virtual character model; on the curve to be annotated, select a starting point as an initial node, select intermediate nodes along the direction from the starting point to the ending point, and select the ending point as a termination node; according to the initial node, intermediate nodes, and termination node selected on the curve to be annotated, annotate a corresponding curve on the target texture that corresponds to the curve to be annotated.

[0124] In some embodiments, the first processing module 421 is configured to divide the corresponding curves annotated on the reference virtual character model and the target texture into n pairs of equally long curve segments, where n is a positive integer greater than 1; determine n + 1 pairs of points to be calibrated according to the endpoints of the n pairs of equally long curve segments; calibrate n + 1 pairs of points to be calibrated on the reference virtual character model and the target texture to obtain the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture.

[0125] In some embodiments, the second processing module 422 is configured to calculate the matching loss between the reference virtual character model and the corresponding feature points of the target texture, and the matching loss between the corresponding feature points of the reference virtual character model and the corresponding feature surface of the target texture to determine the target loss function; optimize the global displacement, joint poses, body shape, and vertices of the reference virtual character model through an optimizer until the target loss function converges.

[0126] In some embodiments, the second processing module 422 is configured to keep the body shape of the reference virtual character model fixed, and optimize the global displacement and joint poses of the reference virtual character model through an optimizer until the target loss function converges; optimize the global displacement, joint poses, and body shape of the reference virtual character model through an optimizer until the target loss function converges; keep the global displacement, joint poses, and body shape of the reference virtual character model fixed, and optimize the vertices of the reference virtual character model until the target loss function converges.

[0127] In some embodiments, the second processing module 422 is configured to keep the global displacement, joint poses, and body shape of the reference virtual character model fixed, and use the neural Jacobian field to optimize the shape Jacobian matrix of the triangular patches on the reference virtual character model, so as to optimize the vertices of the reference virtual character model until the target loss function converges.

[0128] In some embodiments, the model matching device 400 further includes a second acquisition module 412 and a fourth processing module 424. The second acquisition module 412 is configured to acquire a clothing model fitted on the reference virtual character model. The fourth processing module 424 is configured to determine the clothing model fitted on the target virtual character model according to the target virtual character model matching the target texture and the clothing model fitted on the reference virtual character model.

[0129] It should be noted that the functions of the various modules in the model matching device 400 in the embodiments of the present application can be correspondingly referred to the specific implementation manners of any of the above method embodiments, and will not be elaborated here.

[0130] Each unit in the above device can be implemented in whole or in part by software, hardware, and their combination. The above units can be embedded in the processor of the computer device in hardware form or be independent of it, or can be stored in the memory of the computer device in software form, so that the processor can call and execute the operations corresponding to the above units.

[0131] For example, the model matching device 400 can be integrated in a terminal or server with a storage and a processor and having computing capabilities, or the model matching device 400 is the terminal or server.

[0132] In some embodiments, the present application further provides a computer device, including a memory and a processor. A computer program is stored in the memory, and when the processor executes the computer program, the steps in the above method embodiments are implemented.

[0133] Figure 8 It is a schematic structural diagram of the computer device provided by the embodiments of the present application. As Figure 8 shown, the computer device 500 may include: a communication interface 501, a memory 502, a processor 503, and a communication bus 504. The communication interface 501, the memory 502, and the processor 503 communicate with each other through the communication bus 504. The communication interface 501 is used for the device 500 to perform data communication with external devices. The memory 502 can be used to store software programs and modules. The processor 503 runs the software programs and modules stored in the memory 502, such as the software programs for the corresponding operations in the foregoing method embodiments.

[0134] In some embodiments, the processor 503 may call the software programs and modules stored in the memory 502 to execute the above method for matching virtual character models.

[0135] In some embodiments, the computer device 500 may be integrated, for example, in a terminal or a server that has a storage and is equipped with a processor and has computing capabilities, or the computer device 500 is the terminal or the server.

[0136] The present application further provides a computer-readable storage medium for storing a computer program. The computer-readable storage medium includes the stored program, wherein when the program is run by the processor, it executes the method for matching the virtual character model in any of the foregoing embodiments, and the computer program enables the computer device to execute the corresponding processes in the above methods in the embodiments of the present application. For the sake of brevity, it will not be elaborated here.

[0137] The present application further provides a computer program product, including a computer program / instructions. When the computer program / instructions are executed by the processor, they implement the method for matching the virtual character model in any of the foregoing embodiments. The processor of the computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, enabling the computer device to execute the corresponding processes in the above methods in the embodiments of the present application. For the sake of brevity, it will not be elaborated here.

[0138] The present application further provides a computer program, which includes computer instructions. The computer instructions are stored in a computer-readable storage medium. The processor of the computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, enabling the computer device to execute the corresponding processes in the above methods in the embodiments of the present application. For the sake of brevity, it will not be elaborated here.

[0139] It should be understood that the processor in the embodiments of the present application may be an integrated circuit chip with signal processing capabilities. In the implementation process, the steps of the above method embodiments can be completed by the integrated logic circuit in the hardware of the processor or instructions in the form of software. The above-mentioned processor may be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. It can implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of the present application. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The steps of the method disclosed in combination with the embodiments of the present application can be directly embodied as being executed and completed by a hardware decoding processor, or executed and completed by a combination of hardware and software modules in the decoding processor. The software module may be located in a mature storage medium in the art such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. This storage medium is located in the memory, and the processor reads the information in the memory and combines its hardware to complete the steps of the above method.

[0140] It can be understood that the memory in the embodiments of the present application can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memories. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory can be a random access memory (RAM), which is used as an external cache. By way of example but not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchlink dynamic random access memory (SLDRAM), and direct rambus random access memory (DRRAM). It should be noted that the memory of the systems and methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0141] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or by a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. A professional technician can use different methods for each specific application to implement the described functions, but such implementation should not be considered to exceed the scope of the present application.

[0142] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0143] In several embodiments provided in this application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection between each other can be through some interfaces. The indirect coupling or communication connection of devices or units can be in electrical, mechanical, or other forms.

[0144] In the embodiments of this application, the term "module" or "unit" refers to a computer program with a predetermined function or a part of a computer program, which works together with other related parts to achieve a predetermined goal, and can be fully or partially implemented by using software, hardware (such as a processing circuit or a memory), or a combination thereof. Similarly, a processor (or multiple processors or memories) can be used to implement one or more modules or units. In addition, each module or unit can be a part of an overall module or unit that includes the function of that module or unit.

[0145] The units described as separate components may or may not be physically separated. The components shown as units may or may not be physical units, that is, they can be located in one place, or they can be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0146] In addition, each functional unit in the embodiments of this application can be integrated into a processing module, or each unit can exist physically alone, or two or more units can be integrated into one unit.

[0147] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of this application, in essence, or the part that contributes to the traditional technology, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to enable a computer device (which can be a personal computer, a server) to execute all or part of the steps of the methods of various embodiments of this application. The foregoing storage medium includes: USB flash drives, mobile hard disks, ROM, RAM, magnetic disks, or optical discs and other various media that can store program codes.

[0148] The above are only specific embodiments of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of changes or substitutions, which should all be covered within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims

1. A method for matching a virtual character model, characterized in that: The method comprises: Obtain a reference virtual character model and target texture; Determining feature points to be matched on the reference virtual character model and feature points to be matched on the target texture based on the correspondence relationship marked on the reference virtual character model and the target texture; Matching the feature points to be matched on the reference virtual character model with the feature points to be matched on the target texture based on a target matching algorithm to obtain matched feature points on the target texture; Obtaining a target virtual character model that matches the target texture according to the matched feature points on the target texture; Among them, the target matching algorithm matches the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture through the marked correspondence, and iteratively optimizes the matching results based on the target loss function. The target loss function obtained by iterative optimization realizes the minimization of the matching loss between the reference virtual character model and the corresponding feature points of the target texture, and the minimization of the matching loss from the corresponding feature points of the reference virtual character model to the corresponding feature surfaces of the target texture.

2. The virtual character model matching method according to claim 1, characterized in that: The method further comprises: Display the model preview panel; In response to an import instruction for the reference virtual character model and the target texture, displaying the reference virtual character model and the target texture in the model preview panel; In response to a labeling event for the correspondence between the reference virtual character model and the target texture in the model preview panel, displaying the correspondence labeled on the reference virtual character model and the target texture, and feature points to be matched on the reference virtual character model and the target texture determined according to the labeled correspondence on the model preview panel; In response to the registration event of the reference virtual character model and the target texture in the model preview panel, the matching result of matching the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture based on the target matching algorithm, and the result of each iterative optimization of the matching result are displayed in the model preview panel, wherein the final result of the iterative optimization of the matching result is the matched feature points on the target texture; A target virtual character model matching the target texture is obtained according to the matched feature points on the target texture, and the target virtual character model matching the target texture is displayed on the model preview panel.

3. The virtual character model matching method according to claim 1, characterized in that: The determining, based on the correspondence relationship marked on the reference virtual character model and the target texture, feature points to be matched on the reference virtual character model and feature points to be matched on the target texture comprises: Marking corresponding curves on the reference virtual character model and the target texture through a wiring selection plug-in; The corresponding curves marked on the reference virtual character model and the target texture are uniformly sampled to obtain feature points to be matched on the reference virtual character model and feature points to be matched on the target texture.

4. The virtual character model matching method according to claim 3, characterized in that: The step of marking corresponding curves on the reference virtual character model and the target texture by selecting a plug-in through wiring includes: Selecting a curve to be marked on the reference virtual character model; On the curve to be annotated, select the starting point as the initial node, select the middle node along the direction from the starting point to the end point, and select the end point as the end node; According to the initial node, the intermediate node and the terminal node selected on the curve to be marked, a curve corresponding to the curve to be marked is marked on the target texture.

5. The virtual character model matching method according to claim 3, characterized in that: The uniformly sampling corresponding curves marked on the reference virtual character model and the target texture to obtain feature points to be matched on the reference virtual character model and feature points to be matched on the target texture includes: Dividing the corresponding curves marked on the reference virtual character model and the target texture into n pairs of curve segments of equal length, where n is a positive integer greater than 1; According to the endpoints of n pairs of curve segments of equal length, determine n+1 points to be calibrated; The n+1 points to be calibrated are calibrated on the reference virtual character model and the target texture to obtain feature points to be matched on the reference virtual character model and feature points to be matched on the target texture.

6. The virtual character model matching method according to claim 1, characterized in that: The method of matching the feature points to be matched on the reference virtual character model with the feature points to be matched on the target texture based on the target matching algorithm to obtain the matched feature points on the target texture includes: Calculating the matching loss between the reference virtual character model and the corresponding feature points of the target texture, and the matching loss from the corresponding feature points of the reference virtual character model to the corresponding feature surfaces of the target texture to determine a target loss function; The global displacement, joint posture, human body shape and vertices of the reference virtual character model are optimized by an optimizer until the target loss function reaches convergence.

7. The virtual character model matching method according to claim 6, characterized in that: The optimizing the global displacement, joint posture, human body shape and vertices of the reference virtual character model by the optimizer until the target loss function reaches convergence includes: The human body shape of the reference virtual character model is fixed and the global displacement and joint posture of the reference virtual character model are optimized by an optimizer until the target loss function reaches convergence; Optimizing the global displacement, joint posture and human body shape of the reference virtual character model by an optimizer until the target loss function reaches convergence; The global displacement, joint posture and human body shape of the reference virtual character model are fixed unchanged, and the vertices of the reference virtual character model are optimized until the target loss function reaches convergence.

8. The virtual character model matching method according to claim 7, characterized in that: The fixing the global displacement, joint posture and human body shape of the reference virtual character model unchanged, and optimizing the vertices of the reference virtual character model until the target loss function reaches convergence, comprises: The global displacement, joint posture and body shape of the reference virtual character model are fixed unchanged, and the shape Jacobian matrix of the triangular facets on the reference virtual character model is optimized using the neural Jacobian field to optimize the vertices of the reference virtual character model until the target loss function converges.

9. The virtual character model matching method according to claim 1, characterized in that: The method further comprises: Acquire a clothing model fitted on the reference virtual character model; A clothing model fitted on the target virtual character model is determined according to the target virtual character model matched with the target texture and the clothing model fitted on the reference virtual character model.

10. A model matching device, characterized in that: The device comprises: An acquisition unit configured to acquire a reference virtual character model and a target texture; A first processing unit is configured to determine feature points to be matched on the reference virtual character model and feature points to be matched on the target texture based on the correspondence relationship marked on the reference virtual character model and the target texture; A second processing unit is configured to match the feature points to be matched on the reference virtual character model with the feature points to be matched on the target texture based on a target matching algorithm to obtain matched feature points on the target texture; A third processing unit is configured to obtain a target virtual character model matching the target texture according to the matched feature points on the target texture; Among them, the target matching algorithm matches the feature points to be matched on the reference virtual character model and the feature points to be matched on the target texture through the marked correspondence, and iteratively optimizes the matching results based on the target loss function. The target loss function obtained by iterative optimization realizes the minimization of the matching loss between the reference virtual character model and the corresponding feature points of the target texture, and the minimization of the matching loss from the corresponding feature points of the reference virtual character model to the corresponding feature surfaces of the target texture.

11. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein the program is executed by a processor to perform the method described in any one of claims 1 to 9.

12. A computer device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to execute the method according to any one of claims 1 to 9 through the computer program.

13. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the method according to any one of claims 1 to 9 is implemented.

Images