Digital human expression migration method, device, electronic device and storage medium

By filtering and migrating the expression library of the target reference model from the reference model library, the expressions are automatically migrated to the object model, and the problems of high time cost and low efficiency in the digital human expression binding process are solved, achieving efficient and accurate numbers Generate human image.

CN117541690BActive Publication Date: 2024-11-15BEIJING BAIDU NETCOM SCI & TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202311339427.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-10-16
Publication Date
2024-11-15
Estimated Expiration
2043-10-16

AI Technical Summary

Technical Problem

In the prior art, the digital human expression binding process requires professional designers to participate, and the time cost is high and the efficiency is low, making it difficult to meet the requirements of the high-quality expression driving effect of hyperrealistic digital humans.

Method used

Provides a digital human expression migration method, which can achieve automated expression binding by filtering the matching target reference model from the preset reference model library, obtaining its expression library, and migrating the last frame of the expression to the object model.

Benefits of technology

It realizes automatic, accurate and efficient migration of digital human expressions, reduces time costs, improves the efficiency of digital human image generation, and reduces dependence on professional designers.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117541690B_ABST
    Figure CN117541690B_ABST
Patent Text Reader

Abstract

The present disclosure provides a method, device, electronic device and storage medium for migrating expressions of digital humans, which relate to technical fields such as augmented reality, virtual reality, computer vision, and deep learning, and can be applied to scenes such as the metaverse and virtual digital humans. The specific implementation scheme includes: screening the identification of a target reference model that matches an object model from a preset reference model library; the reference model library includes multiple reference models; based on the identification of the target reference model, obtaining the expression library of the target reference model; migrating the last frame of the expression in the expression library of the target reference model to the object model to obtain the last frame of the expression of the object model. The technology disclosed in the present disclosure can automatically realize the expression migration of digital humans, and the entire expression migration process saves time and effort. It can not only effectively ensure the accuracy of the migrated expressions, but also effectively improve the migration efficiency of digital human expressions and the generation efficiency of digital human images.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present disclosure relates to the fields of computer technology and artificial intelligence technology, specifically to the fields of augmented reality, virtual reality, computer vision, deep learning, etc., and can be applied to scenes such as metaverse and virtual digital human. Specifically, it relates to a method, device, electronic device and storage medium for transferring expression of digital human. Background Art

[0002] Binding, as a key design link for driving the digital human image, enables the expression of the digital human to be applied to the digital human model. In the prior art, binding is usually completed by professional designers. Binding can be in the form of blendshape deformation and bone skinning. In some scenarios, a combination of the two is also used to achieve optimization.

[0003] Different digital human images require different levels of investment in binding. But it is worth noting that the time period is not short, usually more than 1 to 2 weeks. Under the requirement of high-quality expression driving effect of hyper-realistic digital humans, this time cost will be further increased significantly. Summary of the invention

[0004] The present invention provides a method, device, electronic device and storage medium for migrating expression of a digital human.

[0005] According to one aspect of the present disclosure, a method for migrating digital human expressions is provided, comprising:

[0006] Filtering an identifier of a target reference model matching the object model from a preset reference model library; the reference model library includes a plurality of reference models;

[0007] Based on the identifier of the target reference model, acquiring an expression library of the target reference model;

[0008] The last frame of the expression in the expression library of the target reference model is migrated to the object model to obtain the last frame of the expression of the object model.

[0009] According to another aspect of the present disclosure, a digital human expression migration device is provided, comprising:

[0010] A screening module, used to screen the identifier of a target reference model matching the object model from a preset reference model library; the reference model library includes a plurality of reference models;

[0011] An acquisition module, used for acquiring an expression library of the target reference model based on the identifier of the target reference model;

[0012] The last frame migration module is used to migrate the last frame of the expression in the expression library of the target reference model to the object model to obtain the last frame of the expression of the object model.

[0013] According to another aspect of the present disclosure, there is provided an electronic device, comprising:

[0014] at least one processor; and

[0015] a memory communicatively connected to the at least one processor; wherein,

[0016] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method of any possible implementation manner and the aspects described above.

[0017] According to yet another aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to cause the computer to execute the method of the above-mentioned aspect and any possible implementation manner.

[0018] According to yet another aspect of the present disclosure, a computer program product is provided, including a computer program, wherein when the computer program is executed by a processor, the computer program implements the above-mentioned aspects and any possible implementation method.

[0019] According to the technology disclosed in the present invention, the expression migration of digital humans can be automatically realized. Compared with the existing technology, the entire expression migration process saves time and effort. It can not only effectively ensure the accuracy of the migrated expression, but also effectively improve the migration efficiency of digital humans' expressions and the generation efficiency of digital human images.

[0020] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it intended to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The accompanying drawings are used to better understand the present solution and do not constitute a limitation of the present disclosure.

[0022] Figure 1 is a schematic diagram according to a first embodiment of the present disclosure;

[0023] Figure 2 is a schematic diagram according to a second embodiment of the present disclosure;

[0024] Figure 3 is a schematic diagram of an eye structure of an object model provided by the present disclosure;

[0025] Figure 4is a schematic diagram according to a third embodiment of the present disclosure;

[0026] Figure 5 is a schematic diagram according to a fourth embodiment of the present disclosure;

[0027] Figure 6 The block diagram is a block diagram of an electronic device for implementing the method of the embodiment of the present disclosure. DETAILED DESCRIPTION

[0028] The following is a description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0029] Obviously, the described embodiments are only part of the embodiments of the present disclosure, but not all of them. Based on the embodiments of the present disclosure, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present disclosure.

[0030] It should be noted that the terminal devices involved in the embodiments of the present disclosure may include but are not limited to mobile phones, personal digital assistants (PDAs), wireless handheld devices, tablet computers and other smart devices; display devices may include but are not limited to personal computers, televisions and other devices with display functions.

[0031] In addition, the term "and / or" in this article is only a description of the association relationship between the associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0032] In the prior art, the binding of expressions to digital humans is mainly done manually by professional designers. The whole process is time-consuming and laborious, resulting in very low efficiency of expression binding. Based on this, the present disclosure provides a method for migrating expressions of digital humans, which can automatically migrate the expressions of a reference model to an object model of a digital human, effectively improving the efficiency of binding expressions to the image of a digital human.

[0033] Figure 1 is a schematic diagram according to the first embodiment of the present disclosure; Figure 1 As shown, this embodiment provides a method for migrating digital human expressions, which may specifically include the following steps:

[0034] S101, selecting an identifier of a target reference model matching an object model from a preset reference model library; the reference model library includes a plurality of reference models;

[0035] The execution subject of the digital human expression migration method of this embodiment can be a digital human expression migration device, which can be an electronic entity or a software integrated application. The device can migrate the expression of any reference model in the reference model library to any object model to achieve automatic migration of the digital human expression.

[0036] The object model of this embodiment is a digital human model that uses facial expressions. The preset reference models in the reference model library are pre-configured digital human models of various forms and styles, for example, they may include male models, female models, or may also include boy models, middle-aged male models, elderly male models, girl models, middle-aged female models, elderly female models, etc. Various reference models can be preset according to specific needs, and are not limited here.

[0037] The identifier of the target reference model in this embodiment is any identifier that can identify the target reference model, and is not limited here.

[0038] S102, based on the identifier of the target reference model, obtaining an expression library of the target reference model;

[0039] In this embodiment, expression libraries can be pre-configured for various preset reference models, and the expression library of each reference model can include multiple expressions. Specifically, various expressions can be stored in the form of point cloud data. For example, the natural state of the reference model is the state where the digital human does not make any expression, and the natural state can also be the standard state. If the digital human of the reference model applies an expression, it will inevitably cause the point cloud on the reference model that executes the expression to change position, while the point cloud that does not execute the expression will not change position. Therefore, the expression of the reference model can only store the point cloud data that changes when the expression is executed.

[0040] S103, migrating the last frame of the expression in the expression library of the target reference model to the object model to obtain the last frame of the expression of the object model.

[0041] In actual application scenarios, many facial expressions have small amplitudes, and the expression library only includes the point cloud data of the last frame of the expression, that is, the expression library only records the final state of the expression. In this case, when performing expression migration, the last frame of the expression in the expression library of the target reference model can be migrated to the object model to obtain the last frame of the expression of the object model.

[0042] The expression migration of this embodiment is used to migrate the last frame of the expression of the target reference model matching the object model in the reference model library to the object model, so as to realize automatic, accurate and efficient migration of the expression.

[0043] The digital human expression migration method of the present embodiment, by adopting the above scheme, can realize the automatic migration of expressions in the expression library of the target reference model that matches the object model to the object model. Compared with the prior art, it does not require manual binding by professional designers, which saves time and effort. It can not only effectively ensure the accuracy of the migrated expressions, but also effectively improve the migration efficiency of the digital human expressions and the generation efficiency of the digital human image.

[0044] Figure 2 is a schematic diagram according to the second embodiment of the present disclosure; Figure 1 Based on the technical solutions of the embodiments shown in the figure, the technical solutions of the present disclosure are further described in more detail. Figure 2 As shown, the expression transfer method of a digital human in this embodiment may specifically include the following steps:

[0045] S201, acquiring a first characteristic curve of a preset part from an object model;

[0046] S202, obtaining a second characteristic curve of a preset part from each reference model;

[0047] S203, calculating an offset distance of a preset part between the object model and each reference model based on the first characteristic curve and each second characteristic curve;

[0048] S204, based on the offset distance of the preset position between the object model and each reference model, selecting the identifier of the target reference model matching the object model from the reference model library;

[0049] The above steps S201-S204 of this embodiment are as follows: Figure 1 A specific implementation of step S101 of the illustrated embodiment.

[0050] Since the difference between different digital human images is mainly reflected in the face, the preset part of this embodiment can refer to a part of the face, such as the eye, mouth, etc. For example, the first characteristic curve can be at least one of the six curves of upper eyelid, lower eyelid, upper eyelid outer contour, lower eyelid outer contour, inner eye corner divergence, and outer eye corner divergence. Or the first characteristic curve can also be at least one of the multiple curves of the upper lip inner line, upper lip outer line, and bite line of the mouth. For digital human image models such as object models and reference models, the number and distribution of point clouds included in the same preset part are the same. That is, for the same preset part, there are corresponding points between the two models.

[0051] In this embodiment, when steps S201-S204 are implemented, only one preset part may be selected, or more than two preset parts may be selected. For the same preset part, only one first characteristic curve may be obtained from the object model, and one corresponding second characteristic curve may be obtained from each reference model; or more than two first characteristic curves may be obtained from the object model, and more than two corresponding second characteristic curves may be obtained from each reference model.

[0052] If only one preset part is selected, a first characteristic curve and a second characteristic curve are used for the above screening, and the points on the first characteristic curve and the second characteristic curve have the same distribution and quantity. At this time, the specific implementation of step S203 may include the following steps:

[0053] (a1) obtaining the coordinates of each point on the first characteristic curve and each second characteristic curve respectively;

[0054] The coordinates of each point on the first characteristic curve and each second characteristic curve can be acquired from the model data of the object model and each reference model respectively.

[0055] (b1) calculating the point distance between the first characteristic curve and the same point markers on the second characteristic curves based on the coordinates of each point on the first characteristic curve and each second characteristic curve;

[0056] In the object model and the reference model, the same point identifiers may be used for points with corresponding distributions. For example, the first characteristic curve of a preset part of the object model may include points with identifiers 1, 2, 3, 4, 5, and 6. Correspondingly, the second characteristic curve corresponding to the first characteristic curve of the same preset part of the reference model may also include points with identifiers 1, 2, 3, 4, 5, and 6. Points with the same identifiers are points with the same distribution and corresponding positions. However, the coordinates of points with the same identifiers in two different models may not be the same.

[0057] Specifically, the point distance of the same point marker may be calculated according to the coordinates of two points of the same point marker on the first characteristic curve and each second characteristic curve.

[0058] (c1) adding the point distances of each point mark on the first characteristic curve and each second characteristic curve to obtain the sum of the point distances;

[0059] (d1) Based on the sum of the point distances and the number of points included in the first characteristic curve, an offset distance of a preset part between the object model and each reference model is obtained.

[0060] Specifically, the number of points included in the first characteristic curve is the same as the number of points included in each second characteristic curve.

[0061] For example, for the second characteristic curve of any reference model, the point distances of each point mark on the first characteristic curve and the corresponding second characteristic curve are added together to obtain the sum of the point distances, and then the sum of the point distances is divided by the number of points included in the first characteristic curve to obtain the average distance of each point as the offset distance of the preset part between the object model and the reference model. In this way, the offset distance of the preset part between the object model and the reference model can be accurately calculated.

[0062] At this time, when step S204 is specifically implemented, the identifier of the reference model with the smallest offset distance of the preset part can be selected from multiple reference models in the reference model library as the identifier of the target reference model that matches the object model. In this way, the target reference model that matches the object model can be accurately and efficiently obtained.

[0063] Optionally, in one embodiment of the present disclosure, for any reference model, in order to accurately calculate the offset distance of the preset part between the object model and the reference model, before executing step (a1), it is necessary to perform point cloud registration on the first characteristic curve and the second characteristic curve. If they are not registered, when the first characteristic curve and the second characteristic curve are far apart, even if the first characteristic curve and the second characteristic curve have exactly the same shape and distribution, the calculated offset distance of the preset part between the object model and the reference model will be large, and thus there will be a large error. In this embodiment, the point cloud registration of the first characteristic curve and the second characteristic curve is also performed by operating the second characteristic curve through at least one operation mode of displacement, rotation and scaling, so that the first characteristic curve and the second characteristic curve have the most common characteristics, such as having the same center of mass and having the most intersections. In the process of registration, the best registration position can be obtained by multiple attempts until the position where the sum of the point distances between the first characteristic curve and the second characteristic curve is found is the registration position of the first characteristic curve and the second characteristic curve.

[0064] In one embodiment of the present disclosure, if only one preset part is selected, when more than two first characteristic curves and more than two second characteristic curves participate in the above screening, for any group of first characteristic curves and corresponding second characteristic curves, the above method can be used to calculate the sum of the point distances of each point mark on the group of first characteristic curves and second characteristic curves. Then, the cumulative value of the sum of the point distances of each point mark on the first characteristic curves and second characteristic curves of all groups is taken as the offset distance of the preset part between the object model and the corresponding reference model.

[0065] In one embodiment of the present disclosure, if more than two preset parts are selected to participate in the above screening, for each preset part, an offset distance corresponding to the preset part can be calculated in the above manner. Then, the following steps are further performed:

[0066] (a2) calculating the comprehensive offset distance between the object model and each reference model based on the pre-configured weights of each preset part and the offset distance corresponding to each preset part;

[0067] The weights of the preset parts in this embodiment can be set according to actual conditions. For example, the eyes and mouth are more obvious features in the face, so the weights of the eyes and mouth can be set higher.

[0068] (b2) Based on the comprehensive offset distance between the object model and each reference model, the identification of the target reference model matching the object model is selected from the reference model library.

[0069] Specifically, the identifier of the reference model with the smallest comprehensive offset distance may be selected from multiple reference models in the reference model library as the identifier of the target reference model that matches the object model.

[0070] By adopting the above method, it is possible to very efficiently, accurately and reasonably select a target reference model that best matches the object model from the reference model library.

[0071] In addition, optionally, in one embodiment of the present disclosure, the above Figure 1 Step S101 of the illustrated embodiment may also be implemented in any of the following ways:

[0072] A. based on the attribute information of the object model and the attribute information of each reference model in the reference model library, selecting the identifier of the target reference model matching the object model from the reference model library; or

[0073] In practical applications, each digital human model can also be configured with corresponding attribute information when it is established to describe or limit the model. For example, the attribute information can be a middle-aged woman, a child, an elderly man, etc. At this time, the identifier of the reference model with the same attribute information can be selected from the model reference library according to the attribute information of the object model as the identifier of the target reference model.

[0074] B. Displaying the attribute information of each reference model in the reference model library to the user; receiving the identifier of the target reference model selected by the user that matches the object model.

[0075] In this method, the user manually selects the identifier of the target reference model that matches the object model.

[0076] In short, no matter which of the above methods is adopted, the identification of the target reference model that matches the object model can be obtained efficiently and accurately.

[0077] S205, based on the identifier of the target reference model, obtaining an expression library of the target reference model;

[0078] S206, performing overall registration of the target reference model and the object model;

[0079] In this embodiment, the object model can be overall registered with the target reference model by adjusting the target reference model to align the head position, and the target reference model and the object model coordinate system are consistent in direction and size by using displacement, rotation and / or scaling. This step is to roughly locate the corresponding position and size of the model's facial features, so as to facilitate the matching of key features in the next step.

[0080] S207, registering the key features corresponding to the expression of the target reference model with the key features of the object model;

[0081] In this embodiment, an example is taken in which only one key feature is involved in the expression. For example, the key feature may include the left eye, the right eye or the mouth. In order to achieve accurate expression migration, in the key feature registration implementation in this embodiment, the key feature corresponding to the expression and the key feature of the object model may be made consistent in key point coordinates by performing operations such as displacement, rotation and / or scaling on the target reference model. For example, when registering the left eye, it may be shown that the coordinates of the two corners of the left eye are the same. When registering the mouth, it may be shown that the coordinates of the two corners of the mouth are the same, and so on.

[0082] By adopting this step, any key features of any expression can be registered.

[0083] S208, acquiring a first size ratio of the expression based on the point cloud data of the key features corresponding to the last frame of the expression of the target reference model and the point cloud data of the key features of the target reference model in a natural state;

[0084] S209, based on the first size ratio of the expression and the point cloud data of the key features of the object model in the natural state, obtaining the point cloud data of the key features corresponding to the last frame of the expression to be migrated by the object model, and obtaining the last frame of the expression of the object model;

[0085] In this embodiment, when migrating the last frame of the expression, if it is directly migrated, there will be problems such as mismatch. For example, when normal lips are migrated to cherry-like small lips, European-style big lips, etc., the change in the width of the smiling lips and the stretching change of the lower jaw when the mouth is open will produce obvious contradictions. In order to solve this kind of problem, in this embodiment, the length, width and height of the point cloud of the key feature corresponding to the last frame of the expression of the target reference model can be calculated based on the point cloud data of the key feature corresponding to the last frame of the expression. Then, based on the point cloud of the key feature in the natural state of the target reference model, the length, width and height of the point cloud of the key feature in the natural state are calculated, and the ratio of the length of the point cloud of the key feature corresponding to the last frame of the expression to the corresponding length of the point cloud of the key feature in the natural state, the ratio of the width of the point cloud of the key feature corresponding to the last frame of the expression to the corresponding width of the point cloud of the key feature in the natural state, and the ratio of the height of the point cloud of the key feature corresponding to the last frame of the expression to the corresponding height of the point cloud of the key feature in the natural state are calculated respectively as the first size ratio of the expression.

[0086] For example, for an open mouth expression, the coordinates of the highest point of the upper lip line and the lowest point of the lower lip line of the mouth can be obtained based on the point cloud data corresponding to the last frame of the open mouth expression of the target reference model, and the width of the point cloud of the last frame of the open mouth expression of the target reference model can be obtained; the point cloud data of the target reference model in the natural state can also be obtained, and the coordinates of the highest point of the lower lip line and the coordinates of the lowest point of the lower lip line in the natural state can be obtained, and then the width of the point cloud of the target reference model in the natural state can be obtained; then the width of the point cloud of the last frame of the open mouth expression of the target reference model is divided by the width of the point cloud in the natural state to obtain the ratio of the width of the open mouth expression. In the same way, the horizontal coordinate distance between the two mouth corners can be taken as the length of the point cloud, and according to the above method, the length ratio of the open mouth expression can be obtained; and the distance between the outermost point of the lip and the height of the innermost corner of the mouth can be taken as the height of the point cloud, and according to the above method, the height ratio of the open mouth expression can be obtained; that is, the ratio of the three coordinate axis directions in the first size ratio can be obtained.

[0087] Next, based on the first size ratio of the expression and the point cloud data of the key features of the object model in the natural state, the point cloud data of the key features corresponding to the last frame when the object model makes the expression can be obtained, that is, the coordinates corresponding to each point in the point cloud of the key features corresponding to the last frame. In the updated migration result, different scales can be performed on the three coordinate axis directions, so that after migration to the object model, the point cloud of the key features corresponding to the last frame of the expression is more matched with the object model.

[0088] It has been verified through experiments that after processing by adopting steps S208 and S209, the negative impact of volume deformation during expression migration will be reduced, and the amplitude of unreasonable expressions such as cherry-like opening of the mouth, smiling, and pursing the lips will be significantly suppressed.

[0089] It should be noted that, unlike the overall scaling of linear migration, this step scales the transformation of each point differently along the three axes, ensuring that excessive changes are suppressed while similar parts are maintained.

[0090] The above steps S206-S209 are Figure 1 A specific implementation of step S103 of the illustrated embodiment.

[0091] S210, migrating the point cloud data of the key features corresponding to the last frame of the expression to be migrated by the object model to the object model;

[0092] S211, obtaining a third characteristic curve corresponding to the last frame of the expression after migration to the object model and a fourth characteristic curve corresponding to the last frame of the expression on the target reference model;

[0093] S212, adjusting the third characteristic curve using the fourth characteristic curve as a constraint;

[0094] For example, the third characteristic curve and the fourth characteristic curve of this embodiment can be characteristic curves presented when the digital human model makes an expression. For example, for closed-eye expression, the curve of the upper eyelid and the lower eyelid coinciding can be used as the third characteristic curve and the fourth characteristic curve; for pouting expression, the curve of the mouth closed and protruding can be used as the third characteristic curve and the fourth characteristic curve; for pursed-lip expression, the curve formed by the lips closed and rolled inward can be used as the third characteristic curve and the fourth characteristic curve, and so on. When performing expression migration on different models, the third characteristic curve and the fourth characteristic curve are very easy to show inconsistency. Based on this, in this embodiment, the fourth characteristic curve corresponding to the last frame of the expression on the target reference model can be used as a constraint to adjust the third characteristic curve corresponding to the last frame of the expression after migration to the object model, so that the expression can be more accurately integrated into the object model, and the expression can be more completely expressed on the object model. At the same time, it is ensured that the topology around the features of the migrated expression on the object model changes evenly with the expression, and the overall fusion degree is very high.

[0095] Specifically, in this embodiment, the third characteristic curve is adjusted with the fourth characteristic curve as a constraint. The third characteristic curve and the fourth characteristic curve can also be called dynamic curves, and the adjustment method can also be called adjustment based on dynamic curve constraints. The dynamic curve can be a mathematical relationship used in the digital human image technology pipeline. When the dynamic curve connects a group of points into a spatial curve, the change of any point on the curve will drive the change of other points on the curve. On the surface of the digital human model such as the reference model and the object model, all points are connected to the surrounding neighboring points by straight lines to form a grid structure. The third characteristic curve and the fourth characteristic curve themselves are also composed of the edges of the grid structure. The grid structure includes horizontal edges and vertical edges. The third characteristic curve and the fourth characteristic curve in this embodiment are usually horizontal edges. When the third characteristic curve is adjusted with the fourth characteristic curve as a constraint, the positions of each point in the third characteristic curve are adjusted so that the third characteristic curve is completely consistent with the fourth characteristic curve, or the distance between the third characteristic curve and the fourth characteristic curve is less than a very small preset threshold.

[0096] Figure 3 Schematic diagram of the eye structure of an object model provided by the present disclosure. Figure 3 As shown, the characteristic curve pointed by the arrow can be a third characteristic curve. Due to the characteristics of the edges and faces in the grid structure of the object model, the adjusted third characteristic curve cannot change the topological structure of the characteristic curve in the grid structure, which will inevitably lead to the change of the third characteristic curve being transmitted to the adjacent faces and adjacent points in sequence. In the marking of the facial features, this transmission will be attenuated outward from the facial features outer contour characteristic curve. In other words, in order to achieve smoother adjustment, in practical applications, the change of the third characteristic curve can be attenuated outward by a preset number of curves. The preset number can be set according to actual needs, for example, 6, 8, 10 or other values. In order to match the corresponding fourth characteristic curve in the target reference model, the adjustment amplitude of the third characteristic curve is the largest. Referring to the adjustment amplitude of the third characteristic curve, the adjustment amplitude of each characteristic curve in the preset number of characteristic curves adjacent to the third characteristic curve gradually decreases as the distance from the third characteristic curve increases, until the adjacent preset number of characteristic curves complete the last step of this round of adjustment.

[0097] Based on the above method, under the effect of the current migration expression result, the third characteristic curve of the object model will be constrained by the dynamic curve, and the characteristic line itself and the topological lines around the feature will gradually move closer to the fourth characteristic curve. At this point, the topological features of this expression can be completely migrated to the object model.

[0098] S213, smoothing the seam between the point cloud of the key features corresponding to the last frame of the expression after migration to the object model and the original point cloud of the object model;

[0099] In order to make the point cloud of the key features of the last frame of the migrated expression more closely integrated with the original point cloud of the object model, in this embodiment, it is also necessary to detect whether the joint between the two is smooth. For the uneven position, corresponding smoothing processing is required. For example, Figure 3 The AB edge shown can be a seam between the point cloud of the key features corresponding to the last frame of the migrated expression and the original point cloud of the object model, and the AB edge is located in the point cloud of the key features corresponding to the last frame of the migrated expression. During specific processing, it can be detected whether the angle between the two adjacent sides of point B in the horizontal direction is greater than the preset threshold angle. The preset threshold angle can be set according to the actual scene, for example, it can be 15 degrees, 20 degrees, or 30 degrees, etc., which is not limited here. If so, it is considered that the seam is not smooth, and the position of point B needs to be adjusted so that the angle between the two adjacent sides in the horizontal direction of point B is less than or equal to the preset threshold angle. Similarly, the angle between the two adjacent sides in the longitudinal direction of point B can also be detected in a similar manner. If it is not smooth, the position of point B is also adjusted. In short, according to the adjustment in the above manner, each seam between the point cloud of the key features corresponding to the last frame of the expression and the original point cloud of the object model should be very smooth, so that the point cloud of the key features corresponding to the last frame of the expression can be perfectly integrated with the original point cloud of the object model.

[0100] For example, when the migration of the last frame of the expression is completed, there are minor problems in the changes of the seams between the point cloud of the last frame of the migrated expression and the original point cloud, such as the intersection of multiple facial features such as the sides of the nose and the middle of the root of the nose, which may cause uneven wiring in certain cases. The smoothing processing scheme of this embodiment can fully solve such problems and ensure smooth lines.

[0101] Steps S210-S213 are supplements to the above steps S206-S209, which can further enhance the migration effect of the last frame of the expression and improve the accuracy and efficiency of expression migration.

[0102] S214, based on the point cloud data of key features corresponding to the intermediate frame of the expression in the expression library of the target reference model, the point cloud data of key features in the natural state of the target reference model, and the point cloud data of key features corresponding to the final frame of the expression of the target reference model, obtain a second size ratio of the intermediate frame of the expression relative to the final frame;

[0103] Specifically, based on the point cloud data of the key features corresponding to the middle frame of the expression in the expression library of the target reference model, the point cloud data of the key features of the target reference model in the natural state, and the point cloud data of the key features corresponding to the last frame of the expression of the target reference model, the fourth characteristic curve corresponding to the last frame of the expression of the target reference model, the fifth characteristic curve corresponding to the middle frame of the expression, and the sixth characteristic curve in the natural state are obtained. The fourth characteristic curve, the fifth characteristic curve, and the sixth characteristic curve are all characteristic curves of the key features in the target reference model.

[0104] Similarly, for the open mouth expression, the fourth characteristic curve can correspond to the upper lip line when the mouth is opened to the maximum, the fifth characteristic curve can be the upper lip line of a certain intermediate process in the mouth opening process, and the sixth characteristic curve is the closing line when the upper and lower lips are closed when the mouth is not opened. In specific calculations, the highest point on the upper lip line can be used as the key point, and the distance difference between the key point on the fifth characteristic curve and the key point on the upper lip line on the fourth characteristic curve is taken, divided by the distance difference between the key point on the fourth characteristic curve and the key point on the upper lip line on the sixth characteristic curve, to obtain the ratio of the height of the intermediate frame of the open mouth expression to the width of the last frame, as the ratio of the width in the second size ratio. Similarly, in a similar manner, and in combination with the determination method of the length and height in the first size ratio, the ratio of the length and height in the second size ratio can also be obtained.

[0105] S215 . Obtain an intermediate frame of the expression of the object model based on a second size ratio of the intermediate frame of the expression to the final frame and the final frame of the expression of the object model.

[0106] The specific implementation may include the following steps:

[0107] (a3) obtaining a corresponding reference characteristic curve of the intermediate frame of the expression of the object model based on a second size ratio of the intermediate frame of the expression relative to the final frame and a corresponding third characteristic curve of the final frame of the expression of the object model;

[0108] (b3) migrating the point cloud data of the key features corresponding to the intermediate frame of the expression of the target reference model to the object model;

[0109] (c3) obtaining a seventh characteristic curve corresponding to the intermediate frame of the expression after migration to the object model;

[0110] (d3) Using the reference characteristic curve as a constraint, adjust the seventh characteristic curve.

[0111] In this embodiment, the reference characteristic curve is the reference characteristic curve that should theoretically correspond to the expression of the intermediate frame after migration. The seventh characteristic curve is the actual reference characteristic curve after the expression of the intermediate frame migration. In order to improve the effect of the intermediate frame migration, it is necessary to adopt the method of step (d3) to perform dynamic curve adjustment.

[0112] Specifically, the implementation of step (d3) is the same as the implementation principle of step S212 in the above embodiment, and can realize the characteristic curve adjustment of the outward attenuation type. For details, please refer to the relevant records of the above embodiment, which will not be repeated here. Further, the above step S213 can also be further adopted to smooth the migrated intermediate frame, which will not be repeated here.

[0113] Step S214 and step S215 are for executing the migration of intermediate frames if the expressions in the expression library include intermediate frames.

[0114] In traditional expression binding, there are only two data, the natural state and the expression frame, and the amplitude change of the expression is the interpolation of the natural state and the target frame. This method cannot show the details of the real person's expression changes, such as the dimples when a person smiles.

[0115] From the perspective of expression migration, the middle frame of the expression has a certain nonlinear relationship with the first frame and the last frame.

[0116] In this embodiment, by sampling the characteristic curve in the intermediate frame and the last frame, the change constraints of each intermediate frame relative to the last frame are established. When migrating the intermediate frames, it is necessary to retain the characteristic curve of the intermediate frame and also add the change constraints relative to the last frame to this curve, which can effectively improve the migration efficiency of the intermediate frames.

[0117] In this embodiment, the intermediate frame of migration is an intermediate expression of the intermediate frame with several nonlinear changes added between the natural state and the last frame. In this way, the change of an expression from 0 to 1 evolves from a linear change to a nonlinear change, and detailed changes such as the appearance and disappearance of dimples and the protrusion and straightening of lip beads can be realized, making the migrated expression more natural, reasonable and efficient.

[0118] By adopting the above steps and continuously repeating the above migration steps, all expressions in the target reference model can be migrated to the object model one by one, and the expressions of the migrated object model can also reach a good level based on the above series of constraints.

[0119] The expression migration method of the digital human in this embodiment can quickly and efficiently migrate the expressions in the expression library of the target reference model to the object model through the above method. Moreover, the expression migration process is fast and has low computing power. It runs in real time in the background of the image production project, and can complete automatic binding in minutes. It can also directly perform face capture driving to preview the effect. It does not rely on the support of high-computing-power equipment, nor does it require long-term generation waiting. Expression migration is very efficient and accurate.

[0120] Furthermore, for scenes where other means such as closing eyes, pouting, and pursing lips are difficult to fully guarantee the effect, the expression transfer method of the digital human in this embodiment can completely transfer the expression and effectively ensure the effect of expression transfer.

[0121] Moreover, the expression transfer method of the digital human in this embodiment supports 4D nonlinear transition. It can also support the expression transition method of nonlinear interpolation of intermediate frames. And under the constraint, the effect of the 4D nonlinear expression after transfer can be more effectively guaranteed.

[0122] Figure 4 is a schematic diagram according to the third embodiment of the present disclosure; Figure 4 As shown, this embodiment provides a digital human expression transfer device 400, comprising:

[0123] A screening module 401 is used to screen the identifier of a target reference model that matches the object model from a preset reference model library; the reference model library includes a plurality of reference models;

[0124] An acquisition module 402 is used to acquire an expression library of the target reference model based on the identifier of the target reference model;

[0125] The last frame migration module 403 is used to migrate the last frame of the expression in the expression library of the target reference model to the object model to obtain the last frame of the expression of the object model.

[0126] The digital human expression migration device 400 of this embodiment realizes the realization principle and technical effect of digital human expression migration by adopting the above modules, which is the same as the realization of the above related method embodiments. For details, please refer to the records of the above related method embodiments, which will not be repeated here.

[0127] Figure 5 is a schematic diagram according to the fourth embodiment of the present disclosure; the expression transfer device 500 of the digital human in this embodiment, in the above Figure 4 Based on the technical solutions of the embodiments shown in the figure, the technical solutions of the present disclosure are further described in more detail. Figure 5 As shown, the digital human expression transfer device 500 of this embodiment includes: Figure 4 Modules with the same name and function: screening module 501 , acquisition module 502 and last frame migration module 503 .

[0128] In this embodiment, the screening module 501 is used to:

[0129] Based on the attribute information of the object model and the attribute information of each of the reference models in the reference model library, the identifier of the target reference model matching the object model is screened from the reference model library.

[0130] Alternatively, in one embodiment of the present disclosure, the screening module 501 is used to:

[0131] Displaying the attribute information of each reference model in the reference model library to the user;

[0132] The target reference model selected by the user to match the object model is received.

[0133] Alternatively, in one embodiment of the present disclosure, the screening module 501 is used to:

[0134] Acquire a first characteristic curve of a preset part from the object model;

[0135] Acquire a second characteristic curve of the preset part from each of the reference models;

[0136] Calculating an offset distance of the preset part between the object model and each of the reference models based on the first characteristic curve and each of the second characteristic curves;

[0137] Based on the offset distance of the preset part between the object model and each of the reference models, an identifier of a target reference model matching the object model is screened from the reference model library.

[0138] Further optionally, in an embodiment of the present disclosure, the screening module 501 is used to: respectively obtain the coordinates of each point on the first characteristic curve and each of the second characteristic curves;

[0139] Calculating the point distance between the first characteristic curve and the same point markers on the second characteristic curve based on the coordinates of each point on the first characteristic curve and each second characteristic curve;

[0140] Adding the point distances of each point mark on the first characteristic curve and each of the second characteristic curves to obtain a sum of the point distances;

[0141] Based on the sum of the point distances and the number of points included in the first characteristic curve, an offset distance of the preset part between the object model and each of the reference models is obtained.

[0142] Further optionally, in one embodiment of the present disclosure, the screening module 501 is used to:

[0143] The identifier of the reference model with the smallest offset distance of the preset part is selected from the multiple reference models in the reference model library as the identifier of the target reference model matched with the object model.

[0144] Further optionally, in an embodiment of the present disclosure, the screening module 501 is used to: if the preset parts include more than two, calculate the comprehensive offset distance between the object model and each of the reference models based on the preconfigured weights of each of the preset parts and the offset distances corresponding to each of the preset parts;

[0145] Based on the comprehensive offset distance between the object model and each of the reference models, an identifier of a target reference model matching the object model is screened from the reference model library.

[0146] Further optionally, in one embodiment of the present disclosure, the last frame migration module 503 is used to:

[0147] Performing overall registration of the target reference model with the object model;

[0148] Registering key features corresponding to the expression of the target reference model with key features of the object model;

[0149] Acquire a first size ratio of the expression based on the point cloud data of the key features corresponding to the last frame of the expression of the target reference model and the point cloud data of the key features of the target reference model in a natural state;

[0150] Based on the first size ratio of the expression and the point cloud data of the key features of the object model in the natural state, the point cloud data of the key features corresponding to the last frame of the expression to be migrated by the object model is obtained to obtain the last frame of the expression of the object model.

[0151] Further optionally, in one embodiment of the present disclosure, the last frame migration module 503 is used to:

[0152] Migrating point cloud data of key features corresponding to the last frame of the expression to be migrated by the object model to the object model;

[0153] Acquire a third characteristic curve corresponding to the last frame of the expression after migration to the object model and a fourth characteristic curve corresponding to the last frame of the expression on the target reference model;

[0154] Taking the fourth characteristic curve as a constraint, adjusting the third characteristic curve;

[0155] The point cloud of the key features corresponding to the last frame of the expression after being migrated to the object model and the seam of the original point cloud of the object model are smoothed.

[0156] Further optionally, if Figure 5 As shown, in one embodiment of the present disclosure, the digital human expression transfer device 500 further includes:

[0157] The intermediate frame migration module 504 is used to migrate the intermediate frames of the expression in the expression library of the target reference model to the object model to obtain the intermediate frames of the expression of the object model if the expression library also includes the intermediate frames of the expression.

[0158] Further optionally, in one embodiment of the present disclosure, the intermediate frame migration module 504 is used to:

[0159] Based on the point cloud data of key features corresponding to the intermediate frame of the expression in the expression library of the target reference model, the point cloud data of key features in the natural state of the target reference model, and the point cloud data of key features corresponding to the final frame of the expression of the target reference model, a second size ratio of the intermediate frame of the expression relative to the final frame is obtained;

[0160] Based on a second size ratio of the intermediate frame of the expression to the final frame and the final frame of the expression of the object model, an intermediate frame of the expression of the object model is obtained.

[0161] Further optionally, in one embodiment of the present disclosure, the intermediate frame migration module 504 is used to:

[0162] Based on the point cloud data of key features corresponding to the intermediate frame of the expression in the expression library of the target reference model, the point cloud data of key features in the natural state of the target reference model, and the point cloud data of key features corresponding to the final frame of the expression of the target reference model, a fourth characteristic curve corresponding to the final frame of the expression of the target reference model, a fifth characteristic curve corresponding to the intermediate frame of the expression, and a sixth characteristic curve in the natural state are obtained;

[0163] Based on the coordinates of the key points on the fourth characteristic curve, the fifth characteristic curve and the sixth characteristic curve, a second size ratio of the intermediate frame of the expression relative to the final frame is obtained.

[0164] Further optionally, in one embodiment of the present disclosure, the intermediate frame migration module 504 is used to:

[0165] Obtaining a corresponding reference characteristic curve of the intermediate frame of the expression of the object model based on a second size ratio of the intermediate frame of the expression relative to the final frame and a corresponding third characteristic curve of the final frame of the expression of the object model;

[0166] Migrating point cloud data of key features corresponding to the intermediate frame of the expression of the target reference model to the object model;

[0167] Acquire a seventh characteristic curve corresponding to the intermediate frame of the expression after migration to the object model;

[0168] The seventh characteristic curve is adjusted with the reference characteristic curve as a constraint.

[0169] The digital human expression migration device 500 of this embodiment realizes the realization principle and technical effect of digital human expression migration by adopting the above modules, which is the same as the realization of the above related method embodiments. For details, please refer to the records of the above related method embodiments, which will not be repeated here.

[0170] In the technical solution disclosed herein, the acquisition, storage and application of user personal information involved are in compliance with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0171] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium and a computer program product.

[0172] Figure 6 A schematic block diagram of an example electronic device 600 that can be used to implement an embodiment of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0173] like Figure 6 As shown, the device 600 includes a computing unit 601, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 602 or a computer program loaded from a storage unit 608 into a random access memory (RAM) 603. In the RAM 603, various programs and data required for the operation of the device 600 can also be stored. The computing unit 601, the ROM 602, and the RAM 603 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0174] A number of components in the device 600 are connected to the I / O interface 605, including: an input unit 606, such as a keyboard, a mouse, etc.; an output unit 607, such as various types of displays, speakers, etc.; a storage unit 608, such as a disk, an optical disk, etc.; and a communication unit 609, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 609 allows the device 600 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0175] The computing unit 601 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 601 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 601 performs the various methods and processes described above, such as the above-mentioned methods of the present disclosure. For example, in some embodiments, the above-mentioned methods of the present disclosure may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit 608. In some embodiments, part or all of the computer program may be loaded and / or installed on the device 600 via ROM 602 and / or communication unit 609. When the computer program is loaded into RAM 603 and executed by the computing unit 601, one or more steps of the above-mentioned methods of the present disclosure described above may be performed. Alternatively, in other embodiments, the computing unit 601 may be configured to perform the above-mentioned methods of the present disclosure in any other appropriate manner (e.g., by means of firmware).

[0176] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0177] The program code for implementing the method of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0178] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0179] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0180] The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0181] A computer system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The relationship of client and server is generated by computer programs running on respective computers and having a client-server relationship with each other. The server may be a cloud server, a server of a distributed system, or a server combined with a blockchain.

[0182] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this document does not limit this.

[0183] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. A method for transferring digital human expression, comprising: Filtering the identifier of the target reference model matching the object model from the preset reference model library; The reference model library includes multiple reference models; Based on the identifier of the target reference model, acquiring an expression library of the target reference model; Migrating the last frame of the expression in the expression library of the target reference model to the object model to obtain the last frame of the expression of the object model; The step of migrating the last frame of the expression in the expression library of the target reference model to the object model to obtain the last frame of the expression of the object model comprises: Based on the point cloud data of the key features corresponding to the last frame of the expression of the target reference model and the point cloud data of the key features in the natural state of the target reference model, a first size ratio of the expression is obtained; the first size ratio of the expression includes: the ratio of the point cloud of the key features corresponding to the last frame of the expression to the point cloud of the key features in the natural state in terms of length, width and height, respectively; Based on the first size ratio of the expression and the point cloud data of the key features of the object model in the natural state, the point cloud data of the key features corresponding to the last frame of the expression to be migrated by the object model is obtained to obtain the last frame of the expression of the object model.

2. The method according to claim 1, wherein: Filter the identifier of the target reference model that matches the object model from the preset reference model library, including: Based on the attribute information of the object model and the attribute information of each of the reference models in the reference model library, the identifier of the target reference model matching the object model is screened from the reference model library.

3. The method according to claim 1, wherein: Filter the identifier of the target reference model that matches the object model from the preset reference model library, including: Displaying the attribute information of each reference model in the reference model library to the user; The target reference model selected by the user to match the object model is received.

4. The method according to claim 1, wherein: Filter the identifier of the target reference model that matches the object model from the preset reference model library, including: Acquire a first characteristic curve of a preset part from the object model; Acquire a second characteristic curve of the preset part from each of the reference models; Calculating an offset distance of the preset part between the object model and each of the reference models based on the first characteristic curve and each of the second characteristic curves; Based on the offset distance of the preset part between the object model and each of the reference models, an identifier of a target reference model matching the object model is screened from the reference model library.

5. The method according to claim 4, wherein: Calculating the offset distance of the preset part between the object model and each of the reference models based on the first characteristic curve and the second characteristic curve includes: Respectively obtaining coordinates of each point on the first characteristic curve and each of the second characteristic curves; Calculating the point distance between the first characteristic curve and the same point markers on the second characteristic curve based on the coordinates of each point on the first characteristic curve and each second characteristic curve; Adding the point distances of each point mark on the first characteristic curve and each of the second characteristic curves to obtain a sum of the point distances; Based on the sum of the point distances and the number of points included in the first characteristic curve, an offset distance of the preset part between the object model and each of the reference models is obtained.

6. The method according to claim 5, wherein: Based on the offset distance of the preset part between the object model and each of the reference models, selecting an identifier of a target reference model matching the object model from the reference model library includes: The identifier of the reference model with the smallest offset distance of the preset part is selected from the multiple reference models in the reference model library as the identifier of the target reference model matched with the object model.

7. The method according to claim 5, wherein: Based on the offset distance of the preset part between the object model and each of the reference models, selecting an identifier of a target reference model matching the object model from the reference model library includes: If the preset parts include more than two, based on the pre-configured weights of the preset parts and the offset distances corresponding to the preset parts, calculating the comprehensive offset distances between the object model and the reference models; Based on the comprehensive offset distance between the object model and each of the reference models, an identifier of a target reference model matching the object model is screened from the reference model library.

8. The method according to any one of claims 1 to 7, wherein: Before obtaining the first size ratio of the expression based on the point cloud data of the key features corresponding to the last frame of the expression of the target reference model and the point cloud data of the key features of the target reference model in a natural state, the method further includes: The target reference model is globally registered with the object model.

9. The method according to claim 8, wherein: After obtaining the point cloud data of the key features corresponding to the last frame of the expression to be migrated by the object model based on the size ratio of the expression and the point cloud data of the key features of the object model in the natural state, and before obtaining the last frame of the expression of the object model, the method further includes: Migrating point cloud data of key features corresponding to the last frame of the expression to be migrated by the object model to the object model; Acquire a third characteristic curve corresponding to the last frame of the expression after migration to the object model and a fourth characteristic curve corresponding to the last frame of the expression on the target reference model; Taking the fourth characteristic curve as a constraint, adjusting the third characteristic curve; The point cloud of the key features corresponding to the last frame of the expression after being migrated to the object model and the seam of the original point cloud of the object model are smoothed.

10. The method according to any one of claims 1 to 7, wherein: After migrating the last frame of the expression in the expression library of the target reference model to the object model to obtain the last frame of the expression of the object model, the method further comprises: If the expression library also includes the intermediate frames of the expression, the intermediate frames of the expression in the expression library of the target reference model are migrated to the object model to obtain the intermediate frames of the expression of the object model.

11. The method according to claim 10, wherein: Migrating the intermediate frame of the expression in the expression library of the target reference model to the object model to obtain the intermediate frame of the expression of the object model includes: Based on the point cloud data of key features corresponding to the intermediate frame of the expression in the expression library of the target reference model, the point cloud data of key features in the natural state of the target reference model, and the point cloud data of key features corresponding to the final frame of the expression of the target reference model, a second size ratio of the intermediate frame of the expression relative to the final frame is obtained; Based on a second size ratio of the intermediate frame of the expression to the final frame and the final frame of the expression of the object model, an intermediate frame of the expression of the object model is obtained.

12. The method according to claim 11, wherein: Based on the point cloud data of key features corresponding to the intermediate frame of the expression in the expression library of the target reference model, the point cloud data of key features in the natural state of the target reference model, and the point cloud data of key features corresponding to the final frame of the expression of the target reference model, obtaining a second size ratio of the intermediate frame of the expression relative to the final frame, including: Based on the point cloud data of key features corresponding to the intermediate frame of the expression in the expression library of the target reference model, the point cloud data of key features in the natural state of the target reference model, and the point cloud data of key features corresponding to the final frame of the expression of the target reference model, a fourth characteristic curve corresponding to the final frame of the expression of the target reference model, a fifth characteristic curve corresponding to the intermediate frame of the expression, and a sixth characteristic curve in the natural state are obtained; Based on the coordinates of the key points on the fourth characteristic curve, the fifth characteristic curve and the sixth characteristic curve, a second size ratio of the intermediate frame of the expression relative to the final frame is obtained.

13. The method according to claim 11, wherein: Based on a second size ratio of the intermediate frame of the expression to the final frame and the final frame of the expression of the object model, obtaining the intermediate frame of the expression of the object model includes: Obtaining a corresponding reference characteristic curve of the intermediate frame of the expression of the object model based on a second size ratio of the intermediate frame of the expression relative to the final frame and a corresponding third characteristic curve of the final frame of the expression of the object model; Migrating point cloud data of key features corresponding to the intermediate frame of the expression of the target reference model to the object model; Acquire a seventh characteristic curve corresponding to the intermediate frame of the expression after migration to the object model; The seventh characteristic curve is adjusted with the reference characteristic curve as a constraint.

14. A digital human expression transfer device, comprising: A screening module, used to screen the identifier of the target reference model matching the object model from the preset reference model library; The reference model library includes multiple reference models; An acquisition module, used for acquiring an expression library of the target reference model based on the identifier of the target reference model; A final frame migration module, used for migrating the final frame of the expression in the expression library of the target reference model to the object model to obtain the final frame of the expression of the object model; Wherein, the last frame migration module is used to: Based on the point cloud data of the key features corresponding to the last frame of the expression of the target reference model and the point cloud data of the key features in the natural state of the target reference model, a first size ratio of the expression is obtained; the first size ratio of the expression includes: the ratio of the point cloud of the key features corresponding to the last frame of the expression to the point cloud of the key features in the natural state in terms of length, width and height, respectively; Based on the first size ratio of the expression and the point cloud data of the key features of the object model in the natural state, the point cloud data of the key features corresponding to the last frame of the expression to be migrated by the object model is obtained to obtain the last frame of the expression of the object model.

15. The device according to claim 14, wherein: The screening module is used to: Based on the attribute information of the object model and the attribute information of each of the reference models in the reference model library, the identifier of the target reference model matching the object model is screened from the reference model library.

16. The device according to claim 14, wherein: The screening module is used to: Displaying the attribute information of each reference model in the reference model library to the user; The target reference model selected by the user to match the object model is received.

17. The device according to claim 14, wherein: The screening module is used to: Acquire a first characteristic curve of a preset part from the object model; Acquire a second characteristic curve of the preset part from each of the reference models; Calculating an offset distance of the preset part between the object model and each of the reference models based on the first characteristic curve and each of the second characteristic curves; Based on the offset distance of the preset part between the object model and each of the reference models, an identifier of a target reference model matching the object model is screened from the reference model library.

18. The device according to claim 17, wherein: The screening module is used to: Respectively obtaining coordinates of each point on the first characteristic curve and each of the second characteristic curves; Calculating the point distance between the first characteristic curve and the same point markers on the second characteristic curve based on the coordinates of each point on the first characteristic curve and each second characteristic curve; Adding the point distances of each point mark on the first characteristic curve and each of the second characteristic curves to obtain a sum of the point distances; Based on the sum of the point distances and the number of points included in the first characteristic curve, an offset distance of the preset part between the object model and each of the reference models is obtained.

19. The device according to claim 18, wherein: The screening module is used to: The identifier of the reference model with the smallest offset distance of the preset part is selected from the multiple reference models in the reference model library as the identifier of the target reference model matched with the object model.

20. The device according to claim 18, wherein The screening module is used to: If the preset parts include more than two, based on the pre-configured weights of the preset parts and the offset distances corresponding to the preset parts, calculating the comprehensive offset distances between the object model and the reference models; Based on the comprehensive offset distance between the object model and each of the reference models, an identifier of a target reference model matching the object model is screened from the reference model library.

21. The device according to any one of claims 14 to 20, wherein: The last frame migration module is further used for: The target reference model is globally registered with the object model.

22. The device according to claim 21, wherein The last frame migration module is further used for: Migrating point cloud data of key features corresponding to the last frame of the expression to be migrated by the object model to the object model; Acquire a third characteristic curve corresponding to the last frame of the expression after migration to the object model and a fourth characteristic curve corresponding to the last frame of the expression on the target reference model; Taking the fourth characteristic curve as a constraint, adjusting the third characteristic curve; The point cloud of the key features corresponding to the last frame of the expression after being migrated to the object model and the seam of the original point cloud of the object model are smoothed.

23. The device according to any one of claims 14 to 20, wherein: The device also includes: The intermediate frame migration module is used to migrate the intermediate frames of the expressions in the expression library of the target reference model to the object model to obtain the intermediate frames of the expressions of the object model if the expression library also includes the intermediate frames of the expressions.

24. The device according to claim 23, wherein: The intermediate frame migration module is used to: Based on the point cloud data of key features corresponding to the intermediate frame of the expression in the expression library of the target reference model, the point cloud data of key features in the natural state of the target reference model, and the point cloud data of key features corresponding to the final frame of the expression of the target reference model, a second size ratio of the intermediate frame of the expression relative to the final frame is obtained; Based on a second size ratio of the intermediate frame of the expression to the final frame and the final frame of the expression of the object model, an intermediate frame of the expression of the object model is obtained.

25. The device according to claim 24, wherein: The intermediate frame migration module is used to: Based on the point cloud data of key features corresponding to the intermediate frame of the expression in the expression library of the target reference model, the point cloud data of key features in the natural state of the target reference model, and the point cloud data of key features corresponding to the final frame of the expression of the target reference model, a fourth characteristic curve corresponding to the final frame of the expression of the target reference model, a fifth characteristic curve corresponding to the intermediate frame of the expression, and a sixth characteristic curve in the natural state are obtained; Based on the coordinates of the key points on the fourth characteristic curve, the fifth characteristic curve and the sixth characteristic curve, a second size ratio of the intermediate frame of the expression relative to the final frame is obtained.

26. The device according to claim 24, wherein: The intermediate frame migration module is used to: Obtaining a corresponding reference characteristic curve of the intermediate frame of the expression of the object model based on a second size ratio of the intermediate frame of the expression relative to the final frame and a corresponding third characteristic curve of the final frame of the expression of the object model; Migrating point cloud data of key features corresponding to the intermediate frame of the expression of the target reference model to the object model; Acquire a seventh characteristic curve corresponding to the intermediate frame of the expression after migration to the object model; The seventh characteristic curve is adjusted with the reference characteristic curve as a constraint.

27. An electronic device comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 13.

28. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to cause the computer to execute the method according to any one of claims 1-13.

29. A computer program product comprising a computer program, which, when executed by a processor, implements the method according to any one of claims 1 to 13.