Garment solution automatic collision filtering method and device

Abstract

The invention provides an automatic collision filtering method and device for clothing solution, and the method comprises the steps: dividing a human body model into a plurality of different parts, combining a clothing model with the human body model, enabling the clothing model to be matched with the human body model, building a mapping relation between the clothing model and the human body model, and carrying out the collision filtering of the clothing model and the human body model. The method comprises the following steps: acquiring the area information of each clothing section covering the human body model, grouping the clothing sections based on the area information of each clothing section covering the human body model, defining the collision relationship between the clothing sections of different groups and the human body model of different parts, and carrying out cloth calculation based on the collision relationship. According to the scheme, grouping is automatically set for the human body and the clothing sections, and a collision grouping filtering method is used in clothing fabric calculation, so that the clothing can be processed and adjusted more accurately, the problem of clothing surface collision caused by action change can be accurately avoided, the clothing calculation efficiency and speed are greatly improved, and the clothing calculation accuracy is improved. And meanwhile, the tedious work of manually repairing the animation is saved.

Description

Technical Field

[0001] The invention belongs to the technical field of digital images, and in particular relates to a method and device for automatic collision filtering of clothing calculation. Background Art

[0002] With the development of metaverse, digital humans, and computer graphics related technologies, digital human clothing calculation has become an important means to improve the display effect of digital humans. In scenarios such as digital human live broadcasts, the character's motion data is captured by automatic capture devices or generated by AI data, and the animation quality is difficult to control.

[0003] Clothing solving refers to the process of solving by computer how clothing can reach a stable state through deformation under the influence of environmental forces. The forces on clothing include surface collision caused by changes in character posture and movement, gravity of the earth, and wind force in the environment. Among them, surface collision caused by changes in character movement has the greatest impact on clothing deformation. Clothing solving has a series of standard requirements for character movements, including that limbs cannot be interlaced and the surface spacing of limbs cannot be too small, otherwise the solving results will have problems such as penetration and adhesion.

[0004] In the prior art, when faced with the problems of limb interlacing and small limb spacing in character movements, the problem can only be solved by manually modifying the animation. This method has a large workload and low efficiency, and cannot cope with the problem of real-time solution scenes such as live broadcast. As for the problem of clothing solution penetration and adhesion caused by the interlacing of character movements or small spacing, no efficient solution that can be applied in real-time scenes has been proposed. Summary of the invention

[0005] In order to overcome the defects of the prior art, the present invention proposes an automatic collision filtering method for clothing solution, the method comprising:

[0006] Divide the human body model into different parts;

[0007] combining a clothing model with the human body model so that the clothing model fits the human body model; the clothing model includes one or more clothing patterns;

[0008] Establishing a mapping relationship between the clothing model and the human body model to obtain area information of each clothing pattern piece covering the human body model;

[0009] Grouping the clothing patterns based on the area information of the human body model covered by each clothing pattern;

[0010] The collision relationship between the garment patterns of different groups and the human body models of different parts is defined, and cloth solving is performed based on the collision relationship.

[0011] Specifically, the “segmenting the human body model into a plurality of different parts” includes:

[0012] Predicting candidate regions of various parts of the human body based on the human body model;

[0013] Each of the candidate regions is refined, and instance segmentation is performed on the human body model based on the refined candidate regions to divide the human body model into a plurality of different parts.

[0014] Optionally, the “dividing the human body model into a plurality of different parts” includes:

[0015] The human body model is divided into a torso part, a left arm part, a right arm part, a left leg part and a right leg part.

[0016] Preferably, the step of “combining the clothing model with the human body model so that the clothing model is adapted to the human body model” comprises:

[0017] The human body model is made to present a preset posture, and the human body model is made to transition from the preset posture to a specified posture through clothing solution, and the clothing model is made to adapt to the human body model in the specified posture during the transition process.

[0018] Preferably, the step of “combining the clothing model with the human body model so that the clothing model is adapted to the human body model” further comprises:

[0019] Determining whether there is an overlapping or intersecting portion between the clothing model and the human body model;

[0020] If so, the clothing model is reunited with the human body model.

[0021] Preferably, the step of "establishing a mapping relationship between the clothing model and the human body model" is achieved by acquiring vertices in the human body model corresponding to each vertex in the clothing model.

[0022] Specifically, the “grouping each of the clothing patterns based on the area information of the human body model covered by each of the clothing patterns” includes:

[0023] If the area covered by a garment pattern piece accounts for a larger proportion in one of the divided parts of the human body model than in other divided parts, the garment pattern piece is divided into corresponding groups of the divided parts;

[0024] If the area covered by a clothing pattern piece accounts for the same proportion in the multiple divided parts of the human body model and is larger than other divided parts, the clothing pattern piece is simultaneously divided into corresponding groups of the multiple divided parts;

[0025] If the area covered by a clothing pattern piece has the same proportion in all the divided parts of the human body model, the clothing pattern piece is simultaneously divided into corresponding groups of all the divided parts.

[0026] Furthermore, the method further comprises:

[0027] If a clothing pattern is divided into a plurality of groups at the same time, it is determined that the clothing pattern has a collision relationship with all the groups into which it is divided at the same time.

[0028] The present invention also proposes a clothing solution automatic collision filtering device, the device comprising:

[0029] A division module, used for dividing the human body model into a plurality of different parts;

[0030] A combining module, used for combining a clothing model with the human body model so that the clothing model is adapted to the human body model; the clothing model includes one or more clothing patterns;

[0031] An acquisition module, used for establishing a mapping relationship between the clothing model and the human body model, so as to obtain the area information of each clothing pattern piece covering the human body model;

[0032] A grouping module, configured to group each of the clothing patterns based on the area information of the human body model covered by each of the clothing patterns;

[0033] The solving module is used to define the collision relationship between the garment patterns of different groups and the human body models of different parts, and perform cloth solving based on the collision relationship.

[0034] The present invention also proposes a terminal, comprising: a processor and a memory, wherein a computer program is stored in the memory, and when the processor runs the computer program, the clothing solution automatic collision filtering method as described in any one of claims 1-8 is executed.

[0035] The present invention has at least the following beneficial effects:

[0036] The solution proposed in the present invention can effectively detect and solve the collision problem of clothing surface by automatically grouping human body and clothing patterns and using collision group filtering method to solve cloth, so as to ensure that the layout and wearing of clothing models are more reasonable and comfortable, avoid the tedious work of manual animation correction, save time and energy, and improve the efficiency and accuracy of solving.

[0037] Furthermore, this solution predicts the candidate regions of various parts of the human body, and then refines and instance-segmented to divide the human body model, which can more accurately capture various parts of the human body, provide more precise references for subsequent clothing adaptation, make the clothing fit the body curve and posture better, and the clothing model can be better adapted to various parts of the human body model, which helps to improve the realism of the clothing;

[0038] In addition, this solution can solve the interlacing and spacing problems caused by motion capture data and other reasons in scenarios such as real-time live broadcast of virtual people. It can quickly handle and avoid clothing adhesion or exposure in real-time solution, ensure the reasonable wearing of clothing in virtual scenes, and provide a better user experience.

[0039] Therefore, the present invention provides a method and device for automatic collision filtering in clothing solution. This solution can process and adjust clothing more accurately by automatically setting groups for human bodies and clothing patterns and using collision group filtering methods in clothing fabric solution. It can accurately avoid the problem of clothing surface collision caused by movement changes, greatly improve the efficiency and speed of clothing solution, and save the tedious work of manual animation correction. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0041] Figure 1 A flowchart of the overall steps of the clothing solution automatic collision filtering method provided in Example 1;

[0042] Figure 2 An example diagram for labeling vertices of different parts of a human body model;

[0043] FIG3(a) is an example diagram of a human body model wearing a clothing model, and FIG3(b) is an example diagram of obtaining an area where a clothing pattern covers the human body model;

[0044] Figure 4 A flowchart of the steps for dividing a human body model into different sections;

[0045] Figure 5 Flow chart of steps to avoid wearing out of the model when combining the clothing model with the human body model;

[0046] Figure 6 This is a module structure diagram of the clothing solution automatic collision filtering device provided in Example 2.

[0047] Reference numerals

[0048] 10-division module; 11-prediction unit; 12-division unit; 20-combination module; 21-judgment unit; 22-correction unit; 30-acquisition module; 40-grouping module; 50-solution module. DETAILED DESCRIPTION

[0049] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0050] Hereinafter, various embodiments of the present invention will be described more fully. The present invention may have various embodiments, and adjustments and changes may be made therein. However, it should be understood that there is no intention to limit the various embodiments of the present invention to the specific embodiments disclosed herein, but the present invention should be understood to cover all adjustments, equivalents and / or alternatives falling within the spirit and scope of the various embodiments of the present invention.

[0051] Hereinafter, the term "include" or "may include" used in various embodiments of the present invention indicates the presence of disclosed functions, operations or elements, and does not limit the addition of one or more functions, operations or elements. In addition, as used in various embodiments of the present invention, the terms "include", "have" and their cognates are intended only to indicate specific features, numbers, steps, operations, elements, components or combinations of the foregoing items, and should not be understood as first excluding the presence of one or more other features, numbers, steps, operations, elements, components or combinations of the foregoing items or the possibility of adding one or more features, numbers, steps, operations, elements, components or combinations of the foregoing items.

[0052] In various embodiments of the present invention, the expression "or" or "at least one of A or / and B" includes any combination or all combinations of the words listed at the same time. For example, the expression "A or B" or "at least one of A or / and B" may include A, may include B, or may include both A and B.

[0053] The expressions (such as "first", "second", etc.) used in various embodiments of the present invention may modify the various constituent elements in various embodiments, but may not limit the corresponding constituent elements. For example, the above expressions do not limit the order and / or importance of the elements. The above expressions are only used for the purpose of distinguishing an element from other elements. For example, the first user device and the second user device indicate different user devices, although both are user devices. For example, without departing from the scope of various embodiments of the present invention, the first element may be referred to as the second element, and similarly, the second element may also be referred to as the first element.

[0054] It should be noted that in the present invention, unless otherwise clearly specified and defined, the terms such as "installation", "connection", "fixation" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium; it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0055] In the present invention, those skilled in the art need to understand that the terms indicating orientation or positional relationship herein are based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention.

[0056] The terms used in various embodiments of the present invention are only used for the purpose of describing specific embodiments and are not intended to limit various embodiments of the present invention. As used herein, the singular form is intended to also include the plural form, unless the context clearly indicates otherwise. Unless otherwise limited, all terms used here (including technical terms and scientific terms) have the same meaning as the meaning generally understood by those of ordinary skill in the art to which the various embodiments of the present invention belong. The terms (such as the terms defined in the dictionary generally used) will be interpreted as having the same meaning as the contextual meaning in the relevant technical field and will not be interpreted as having an idealized meaning or an overly formal meaning, unless clearly defined in various embodiments of the present invention.

[0057] Example 1

[0058] See also Figure 1 This embodiment proposes a clothing solution automatic collision filtering method, the method comprising:

[0059] S100: Divide the human body model into a plurality of different parts.

[0060] In this embodiment, step S100 is implemented through a 3D instance segmentation algorithm; a 3D instance segmentation algorithm (3DInstance Segmentation Algorithm) is an algorithm in the field of computer vision. 3D instance segmentation is used to represent a 3D point cloud or a 3D model as discrete points or grids, and to identify different object instances through an algorithm, so that different objects in a 3D scene can be accurately segmented and identified.

[0061] In this embodiment, the parts into which the human body model is divided in step S100 include five different parts: a torso human body model (Body), a left arm human body model (LeftArm), a right arm human body model (RightArm), a left leg human body model (LeftLeg), and a right leg human body model (RightLeg).

[0062] Preferably, please refer to Figure 2 ,After the human body model is divided into a number of different parts, the vertices of the human body model in different parts can be ,marked with different colors to facilitate the user to view it.

[0063] S200: combining the clothing model with the human body model to make the clothing model fit the human body model.

[0064] Specifically, step S200 includes:

[0065] The human body model is made to present a preset pose (Apose), and the human body model is made to transition from the preset pose to the specified pose through clothing solving, and at the same time, the clothing model is made to adapt to the human body model in the specified pose during the transition process.

[0066] It should be noted that the human body model and clothing model used in this embodiment are both three-dimensional models; in this embodiment, the clothing model includes one or more clothing patterns.

[0067] S300: Establishing a mapping relationship between a clothing model and a human body model to obtain information about the area where each clothing pattern covers the human body model.

[0068] Specifically, referring to FIG. 3 , step S300 is implemented by obtaining vertices in the human body model corresponding to each vertex in the clothing model. In this embodiment, the method for obtaining vertices in the human body model corresponding to each vertex in the clothing model is the Minimum Distance Skinning Method.

[0069] It should be noted that the minimum distance skinning method is a technology used in computer graphics for character animation. It is a method for applying the skeletal animation of a character model (such as a human character) to its surface mesh model. The goal of the minimum distance skinning method is to apply the skeletal animation to the surface mesh model so that it can deform with the movement of the bones and achieve a natural animation effect. This is achieved by calculating the association between each surface mesh vertex and the nearest bone. The minimum distance skinning method can quickly calculate the animation deformation of the character model at runtime, and has good realism and stability.

[0070] S400: Grouping the clothing patterns based on the area information of the human body model covered by each clothing pattern.

[0071] Specifically, step S400 includes:

[0072] If the area covered by a garment pattern piece accounts for a larger proportion in one divided part of the human body model than in other divided parts, the garment pattern piece is divided into corresponding groups of the divided part;

[0073] If the area covered by a clothing pattern piece accounts for the same proportion in the multiple divided parts of the human body model and is larger than the other divided parts, the clothing pattern piece is simultaneously divided into corresponding groups of the multiple divided parts;

[0074] If the area covered by a clothing pattern piece accounts for the same proportion in all the divided parts of the human body model, the clothing pattern piece is simultaneously divided into corresponding groups of all the divided parts.

[0075] Optionally, the grouping of the clothing patterns is into a torso grouped clothing pattern (Body Cloth), a left arm grouped clothing pattern (LeftArm Cloth) and a right arm grouped clothing pattern (RightArm Cloth), a left leg grouped clothing pattern (LeftLegCloth) and a right leg grouped clothing pattern (RightLeg Cloth).

[0076] Specifically, the calculation formula based on the proportion of the area covered by each clothing pattern piece in each divided part of the human body model is:

[0077] V all =V Body +V LeftArm +V RightArm +V LeftLeg +V RightLeg

[0078]

[0079]

[0080]

[0081]

[0082]

[0083] Among them, V represents the area covered by the selected clothing pattern, and P represents the proportion of the area covered by the clothing pattern in a certain divided part of the human body model;

[0084] For example, if the area covered by a garment pattern accounts for P of the torso human body model Body The proportion of the human body model larger than the left arm is P LeftArm , the proportion of the right arm human body model P RightArm , the proportion of the left leg part of the human body model P LeftLeg , the proportion of the right leg part of the human body model P RightLeg , the garment pattern is divided into garment patterns of torso grouping;

[0085] If the area covered by a clothing pattern is P in the left leg part of the human body model LeftLeg , the proportion of the right leg part of the human body model P RightLeg The same as, and greater than the proportion of the human body model P Body , the proportion of the left arm human body model P LeftArm , the proportion of the right arm human body model P RightArm , then the clothing pattern is divided into clothing patterns for the left leg group and clothing patterns for the right leg group.

[0086] S500: defining collision relationships between clothing patterns of different groups and different parts of human body models, and performing cloth solving based on the collision relationships.

[0087] By setting reasonable collision relationship rules in step 500, the problem of most limb interlacing or too small spacing between limb movements in character animation movements can be solved.

[0088] Specifically, if a clothing pattern is divided into multiple groups at the same time, it is determined that the clothing pattern has a collision relationship with all the groups it is divided into. Exemplarily, if a clothing pattern is divided into a first group and a second group at the same time, the clothing pattern has a collision relationship with the first group and a collision relationship with the second group at the same time.

[0089] Through the cloth solving step, when limbs intersect, the clothing will not produce additional adhesion. For example, when the hand extends into the body, the clothing pattern of the torso group will not follow the hand into the body, but stay outside the torso human body model; the clothing pattern of the torso group will not collide with the clothing pattern of the left arm group or the clothing pattern of the right arm group, but will be independent of each other. Therefore, this solution can better solve the problem of clothing adhesion and exposure caused by the intersect of the character's movements or the small spacing between the limbs.

[0090] Specifically, see Figure 4 , step S100 includes:

[0091] S110: predicting candidate regions of various parts of the human body based on the human body model.

[0092] In this embodiment, step S110 predicts the target candidate region (Object Proposal) through the Generic Structure Propagation Network (GSPN) algorithm.

[0093] It should be noted that GSPN is an algorithm for generating target candidate regions. The GSPN algorithm uses a convolutional neural network (CNN) to extract features from the input image and generates candidate regions by modeling the structure and context information of the target object. The main idea of ​​GSPN is to introduce multi-scale anchor boxes on the extracted feature map, and use the predicted bounding box offset and confidence score to screen and rank these anchor boxes. Finally, GSPN will output a set of high-quality target candidate regions.

[0094] The target candidate region refers to the candidate target region in the image generated by GSPN. These target candidate regions are obtained by estimating and predicting the positions in the image that may contain the target object based on the GSPN model. The candidate region is usually represented by a bounding box, and each bounding box represents a possible target region. The role of predicting the target candidate region is to narrow the search space in the instance segmentation task and provide the initial position that may contain the target object;

[0095] S120: Refine each candidate region, and perform instance segmentation on the human body model based on the refined candidate regions to divide it into a plurality of different parts.

[0096] In this embodiment, a refinement point cloud network (Refinement PointNet, R-PointNet) is used to refine each candidate region and perform instance segmentation.

[0097] It should be noted that the refined point cloud network is an algorithm used to refine target candidate areas and obtain instance segmentation results. It uses point cloud data and deep learning technology to further process and analyze target candidate areas. The refined point cloud network takes the point cloud data in each target candidate area as input and processes it through a multi-layer perceptron (MLP) and a point cloud feature extractor to obtain more accurate target segmentation results. The refined point cloud network can capture detailed information of the target, perform shape reasoning, and output the semantic category or instance segmentation results of each point, thereby achieving precise positioning and segmentation of the target.

[0098] Specifically, see Figure 5 , step S200 further includes:

[0099] S210: Determine whether there are overlapping or intersecting parts between the clothing model and the human body model.

[0100] It should be noted that step S210 can be completed through intelligent detection and judgment of the device or manual judgment.

[0101] If there are overlapping or intersecting parts between the clothing model and the human body model, step S220 is executed.

[0102] S220: Recombining the clothing model with the human body model.

[0103] In this way, the occurrence of penetration during the combination of the clothing model and the human body model can be effectively avoided.

[0104] Exemplarily, the collision relationship between the garment patterns of different groups and the human body models of different parts may be:

[0105] The torso grouped clothing pattern has collision relationships with the torso human model, the left arm human model, the right arm human model, the left leg human model, the right leg human model, the torso grouped clothing pattern itself, the left arm grouped clothing pattern, the right arm grouped clothing pattern, the left leg grouped clothing pattern, and the right leg grouped clothing pattern;

[0106] The left arm clothing pattern has a collision relationship with the torso human model, the left arm human model, the torso grouped clothing pattern and the left arm clothing pattern itself, but does not have a collision relationship with the right arm human model, the left leg human model, the right leg human model, the right arm grouped clothing pattern, the left leg grouped clothing pattern and the right leg grouped clothing pattern;

[0107] The right arm clothing pattern has a collision relationship with the torso human model, the right arm human model, the torso grouped clothing pattern and the right arm clothing pattern itself, but does not have a collision relationship with the left arm human model, the left leg human model, the right leg human model, the left arm grouped clothing pattern, the left leg grouped clothing pattern and the right leg grouped clothing pattern;

[0108] The left leg clothing pattern has a collision relationship with the torso human model, the left leg human model, the torso grouped clothing pattern, the left arm grouped clothing pattern, the right arm grouped clothing pattern and the left leg grouped clothing pattern itself, but does not have a collision relationship with the left arm human model, the right arm human model, the right leg human model and the right leg grouped clothing pattern;

[0109] The right leg partial clothing pattern has a collision relationship with the torso partial human body model, the right leg partial human body model, the torso grouped clothing pattern, the left arm grouped clothing pattern, the right arm grouped clothing pattern and the right leg grouped clothing pattern itself, but does not have a collision relationship with the left arm partial human body model, the right arm partial human body model, the left leg partial human body model and the left leg grouped clothing pattern.

[0110] Therefore, the automated collision filtering method for clothing solution proposed in this embodiment automatically groups the human body and clothing patterns, and uses the collision group filtering method in clothing fabric solution, thereby bypassing the tedious work of manual animation correction, and bringing new ideas and methods to solve the problems of clothing adhesion or exposure caused by limb interlacing or small spacing in animation.

[0111] In addition, this solution can be applied in real-time scenarios such as virtual human live broadcast. In the real-time live broadcast of virtual humans, the movements are generally generated in real time by the motion capture of the real actor, and the clothing solution program needs to solve the clothing according to the motion capture data. However, due to the influence of factors such as equipment accuracy and motion generation algorithm, the motion capture data often has abnormal situations such as the interlacing and jumping of human limbs, which leads to direct adhesion of fabrics during the real-time solution of clothing. In severe cases, the clothing will be dragged off the human body, causing exposure. The automatic collision filtering method for clothing solution proposed in this embodiment is adopted. After setting the collision grouping of the human body and the clothing pattern, the problem of interlacing of movements or too small spacing in the motion capture data can be avoided in real time according to the configured collision rules, thereby avoiding adhesion or exposure of clothing during live broadcast.

[0112] Example 2

[0113] This embodiment proposes a clothing solution automatic collision filtering device for implementing the method proposed in embodiment 1, see Figure 6 , the device comprises:

[0114] A division module 10, for dividing the human body model into a plurality of different parts;

[0115] A combining module 20, for combining the clothing model with the human body model so that the clothing model is adapted to the human body model;

[0116] An acquisition module 30 is used to establish a mapping relationship between a clothing model and a human body model to obtain information about the area of ​​the human body model covered by each clothing pattern;

[0117] A grouping module 40, for grouping each garment pattern piece based on the area information of the human body model covered by each garment pattern piece;

[0118] The solving module 50 is used to define the collision relationship between the garment patterns of different groups and the human body models of different parts, and perform cloth solving based on the collision relationship.

[0119] Specifically, a clothing model includes one or more clothing patterns.

[0120] In this embodiment, the division module 10 is used to divide the human body model into five different parts, namely, a torso human body model (Body), a left arm human body model (LeftArm), a right arm human body model (RightArm), a left leg human body model (LeftLeg), and a right leg human body model (RightLeg); the grouping module 40 groups the clothing patterns into torso grouped clothing patterns (Body Cloth), left arm grouped clothing patterns (LeftArm Cloth), right arm grouped clothing patterns (RightArm Cloth), left leg grouped clothing patterns (LeftLeg Cloth), and right leg grouped clothing patterns (RightLeg Cloth).

[0121] Specifically, the division module 10 includes:

[0122] A prediction unit 11, used for predicting candidate regions of various parts of the human body based on the human body model;

[0123] The segmentation unit 12 is used to refine each candidate region and perform instance segmentation on the human body model based on the refined candidate regions to divide it into a plurality of different parts.

[0124] The combination module 20 comprises:

[0125] A judging unit 21, used for judging whether there is an overlapping or intersecting part between the clothing model and the human body model;

[0126] The correction unit 22 is used to re-combine the clothing model with the human body model when the judgment unit 21 judges that there is an overlap or intersection between the clothing model and the human body model.

[0127] Example 3

[0128] This embodiment proposes a terminal, which includes a processor and a memory. The memory stores a computer program. When the processor runs the computer program, the clothing solution automatic collision filtering method proposed in Embodiment 1 is executed.

[0129] In summary, the present invention provides a method and device for automated collision filtering in clothing solution. This solution can process and adjust clothing more accurately by automatically grouping the human body and clothing patterns and using the collision group filtering method in clothing fabric solution. It can accurately avoid the problem of clothing surface collision caused by movement changes, greatly improve the efficiency and speed of clothing solution, and save the tedious work of manual animation correction.

[0130] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for automated collision filtering of clothing, characterized in that: The method comprises: Divide the human body model into different parts; combining a clothing model with the human body model so that the clothing model fits the human body model; the clothing model includes one or more clothing patterns; Establishing a mapping relationship between the clothing model and the human body model to obtain area information of each clothing pattern piece covering the human body model; grouping the clothing patterns based on the area information of the human body model covered by each clothing pattern; The collision relationship between the garment patterns of different groups and the human body models of different parts is defined, and cloth solving is performed based on the collision relationship.

2. The clothing solving automatic collision filtering method according to claim 1, characterized in that: The “segmenting the human body model into a plurality of different parts” includes: Predicting candidate regions of various parts of the human body based on the human body model; Each of the candidate regions is refined, and instance segmentation is performed on the human body model based on the refined candidate regions to divide the human body model into a plurality of different parts.

3. The clothing solving automatic collision filtering method according to claim 1 or 2, characterized in that: The “dividing the human body model into a plurality of different parts” includes: The human body model is divided into a torso part, a left arm part, a right arm part, a left leg part and a right leg part.

4. The clothing solving automatic collision filtering method according to claim 1, characterized in that: The “combining the clothing model with the human body model so that the clothing model is adapted to the human body model” includes: The human body model is made to present a preset posture, and the human body model is made to transition from the preset posture to a specified posture through clothing solution, and the clothing model is made to adapt to the human body model in the specified posture during the transition process.

5. The clothing solving automatic collision filtering method according to claim 1 or 4, characterized in that: The “combining the clothing model with the human body model so that the clothing model is adapted to the human body model” further includes: Determining whether there is an overlapping or intersecting portion between the clothing model and the human body model; If so, the clothing model is reunited with the human body model.

6. The clothing solving automatic collision filtering method according to claim 1, characterized in that: The step of "establishing a mapping relationship between the clothing model and the human body model" is achieved by obtaining vertices in the human body model corresponding to each vertex in the clothing model.

7. The clothing solving automatic collision filtering method according to claim 1, characterized in that: The “grouping each of the clothing patterns based on the area information of the human body model covered by each of the clothing patterns” includes: If the area covered by a garment pattern piece accounts for a larger proportion in one of the divided parts of the human body model than in other divided parts, the garment pattern piece is divided into corresponding groups of the divided parts; If the area covered by a clothing pattern piece accounts for the same proportion in the multiple divided parts of the human body model and is larger than other divided parts, the clothing pattern piece is simultaneously divided into corresponding groups of the multiple divided parts; If the area covered by a clothing pattern piece has the same proportion in all the divided parts of the human body model, the clothing pattern piece is simultaneously divided into corresponding groups of all the divided parts.

8. The clothing solving automatic collision filtering method according to claim 1 or 7, characterized in that: The method further comprises: If a clothing pattern is divided into a plurality of groups at the same time, it is determined that the clothing pattern has a collision relationship with all the groups into which it is divided at the same time.

9. A clothing solution automatic collision filtering device, characterized in that: The device comprises: A division module, used for dividing the human body model into a plurality of different parts; A combining module, used for combining a clothing model with the human body model so that the clothing model is adapted to the human body model; the clothing model includes one or more clothing patterns; An acquisition module, used for establishing a mapping relationship between the clothing model and the human body model, so as to obtain the area information of each clothing pattern piece covering the human body model; A grouping module, configured to group each of the clothing patterns based on the area information of the human body model covered by each of the clothing patterns; The solving module is used to define the collision relationship between the garment patterns of different groups and the human body models of different parts, and perform cloth solving based on the collision relationship.

10. A terminal, characterized in that: include: A processor and a memory, wherein a computer program is stored in the memory, and when the processor runs the computer program, the method for automated collision filtering of clothing solution according to any one of claims 1 to 8 is executed.

Images