A method for generating personalized patterns for plus-size yoga wear

Abstract

This application provides a method for generating personalized patterns for plus-size yoga wear, belonging to the field of digital pattern making and personalized customization technology for clothing. Using a plus-size 3D human body model as input, personalized 3D human body models of different obesity levels are first constructed based on continuous deformation parameters. Then, based on human anatomical characteristics and the structural design principles of women's clothing patterns, the target area is segmented to extract 3D components related to the structure of the yoga wear. Subsequently, an improved mass-spring physical model is used to independently flatten each component, outputting a prototype pattern. Based on the prototype pattern, size pre-shrunk and style structure transformations are performed to form a personalized yoga wear pattern. The rationality of the pattern is verified through virtual try-on and pressure distribution evaluation. This method retains the local geometric features of plus-size body types while improving flattening stability and accuracy, making it suitable for the rapid generation of personalized patterns for yoga wear and other tight-fitting sportswear for overweight and obese individuals.

Description

Technical Field

[0001] This application relates to a method for generating personalized patterns for plus-size yoga wear, belonging to the field of digital pattern making and personalized customization technology for clothing. Background Technology

[0002] Tight-fitting sportswear (such as yoga wear) requires a high degree of precision in matching the pattern structure with the human body shape, and its fit, support, and pressure comfort directly affect the wearing experience.

[0003] For overweight and obese individuals, traditional clothing patterns are usually adjusted based on standard sizes, which makes it difficult to accurately reflect the local shape differences in areas such as the abdomen, chest, and buttocks. In areas with complex curvature, problems such as local stretching, wrinkles, or excessive pressure can easily occur, thus affecting wearing comfort and freedom of movement.

[0004] Existing technologies for generating 2D paper patterns from 3D human body models mainly include overall flattening methods, reverse modeling methods, and flattening methods based on geometric analysis and physical simulation. Among them, the overall flattening method is prone to large deformation errors when dealing with complex curved surfaces of large-sized bodies; although the reverse modeling method has higher accuracy, the modeling and flattening process is complex and inefficient, making it difficult to meet the needs of rapid personalization for multiple body types.

[0005] Therefore, it is necessary to propose a pattern generation method that can improve the stability and efficiency of flattening while maintaining the local geometric features of the human body, and support the rapid personalized design of plus-size women's yoga clothing. Summary of the Invention

[0006] In view of this, this application provides a method for generating personalized patterns for plus-size yoga wear. This method is based on three-dimensional geometric analysis and a physical surface flattening algorithm. By constructing a continuously deformable human body model, implementing human body segmentation processing, and independently physical flattening each part, it realizes the rapid generation and virtual verification of plus-size yoga wear patterns for different obesity levels. It completes the generation of two-dimensional plus-size yoga wear personalized patterns from a plus-size three-dimensional human body model, thereby solving the problems of large flattening errors, insufficient fit, and low generation efficiency of existing methods under complex curved surface conditions for plus-size yoga wear.

[0007] Specifically, this application is implemented through the following scheme: A method for generating personalized patterns for plus-size yoga wear includes the following steps: Step 1: Construction of a continuously deformable human body model.

[0008] Step two: Human body size extraction and body shape feature analysis.

[0009] Step 3: Human body model segmentation.

[0010] Step four, the physical component flattening process.

[0011] Step 5: Transformation of the prototype pattern into a personalized yoga wear pattern.

[0012] Step six: Virtual try-on and performance verification.

[0013] Furthermore, as a preferred option: In step one, based on three basic body types, the human body 3D models are seamlessly topologically deformed by setting continuous deformation parameters to generate multiple large-size 3D human body models with different obesity levels, so as to maintain topological consistency between different body types.

[0014] In step two, multiple anthropometric parameters related to the generation of yoga garment patterns are extracted from the human models of different obesity levels for subsequent pattern structure design and size control.

[0015] In step three, combining human anatomical features and the design principles of women's clothing patterns, the three-dimensional human body model is divided into multiple independent three-dimensional components according to key areas such as the chest, waist, hips, and lower limbs, in order to reduce the complexity of a single curved surface. For example, the segmentation is completed in Geomagic Wrap software. First, the human body areas related to the yoga clothing pattern design are trimmed out, and then the area is segmented using the scribing tool and the components are extracted using the cutting tool. The segmentation divides the human body model into 27 areas, of which the upper body front piece is divided into 8 areas, the upper body back piece into 5 areas, and the lower body front and back pieces are each divided into 7 areas.

[0016] In step four, a surface flattening algorithm based on a mass-spring model is used to flatten each human body segment independently. The side lengths are kept consistent by spring constraints, and the projection plane is determined by principal component analysis. An adaptive time step strategy is combined to improve the flattening convergence stability and obtain a two-dimensional prototype paper pattern.

[0017] Specifically: Model discretization and boundary extraction, principal component analysis (PCA) projection plane determination: By reading the 3D mesh model OBJ file of the component, the vertices (particles) of the model are defined as points in 3D space. .

[0018] During this process: The static equilibrium length can be taken as the corresponding side length in the three-dimensional mesh, so as to keep the boundary length consistent during the flattening process.

[0019] The covariance matrix of the point basis is used to perform eigenvalue decomposition to determine the main planar orientation of the component and establish a two-dimensional projected coordinate system.

[0020] (2) Spring force model and energy minimization: In two-dimensional projection space, the goal of a node is to adjust its position using spring forces to reduce curvature and maintain reasonable deformation while preserving boundary length. In each iteration, the force applied by the spring is described by Hooke's Law: .

[0021] Where: Xa and Xb are the two-dimensional coordinates of the two points connected by the spring in the plane; K is the spring stiffness coefficient; The distance between the two points.

[0022] Simultaneously, the spring force is calculated according to Hooke's law in the two-dimensional projection space. The spring force is applied to the nodes, and the damping effect is considered. The two-dimensional coordinates of the particles are iteratively updated using Newton's laws of motion, so that the system tends to a stable equilibrium state. .

[0023] In the numerical simulation, the finite difference method is used to adjust the node positions in time steps, gradually approaching the energy minimum state.

[0024] (3) Adaptive time step control The time step is dynamically adjusted based on the target optimization values ​​from the two mesh expansions. The target optimization value is taken as the average relative error of the mesh edge lengths (Chen et al., 2007). Let Li be the length of the n edges connected to the j-th vertex, and its corresponding length in the k-th iteration be... Then the relative error of the j-th point is: .

[0025] Among them, Li, These are the lengths of the i-th edge connected to the j-th vertex on the spatial grid and the planar grid, respectively.

[0026] If the current optimization target value is smaller than the previous optimization value, then increase the time step. Conversely, the time step is reduced. The formula for dynamically adjusting the time step is: .

[0027] Here, Ecurrent and ELast represent the target optimization values ​​of the current and previous iterations, respectively.

[0028] To avoid the time step from oscillating between two extreme values ​​in practical applications, the algorithm was modified as follows: when the current optimized value is detected to be smaller than the previous optimized value, the step size is not increased immediately, but the iteration continues at the current time step size; if the optimized value has not increased or diverged in 5 consecutive iterations, the time step size is increased; once the optimized value increases, the time step size is immediately reduced.

[0029] (4) Back projection and model reconstruction: After spring optimization in the two-dimensional plane, the two-dimensional coordinates are mapped back to three-dimensional space, and the z-axis coordinate is fixed (usually 0) to obtain the unfolded three-dimensional model: .

[0030] This algorithm combines physical simulation and principal component analysis (PCA) to implement a flattening method that can unfold complex 3D curved surface meshes into 2D planes. While preserving the original mesh topology and local geometric properties, a dynamically adjusted time step strategy is employed, effectively improving convergence speed and algorithm stability, making it suitable for surface flattening applications in the clothing industry.

[0031] In step five, Based on the obtained prototype pattern, and according to the style structure of yoga wear and the characteristics of knitted elastic fabric, differential reduction and structural adjustment are implemented in key parts. Taking the long sleeve of the basic three-dimensional human body model as an example, the style modification is to change the top to a sleeveless wide shoulder strap, and the side of the top is divided by two intersecting lines to achieve a slimming effect. The bottom is a high-waisted close-fitting shorts, generating a personalized yoga wear pattern.

[0032] In step six, virtual try-on technology is used to simulate wearing the generated personalized yoga clothing pattern in standing postures and various typical yoga poses. The surface smoothness and pressure distribution of the clothing are analyzed to verify the fit and comfort of the pattern. Attached Figure Description

[0033] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application.

[0034] Figure 1 This is a flowchart illustrating the process of this application.

[0035] Figure 2 The images show the front and back views of the tops for different degrees of obesity in this application. Part (a) is the front and back view of personalized model one, part (b) is the front and back view of personalized model two, part (c) is the front and back view of personalized model three, and part (d) is the front and back view of personalized model four.

[0036] Figure 3 The figures show the breakdown of the lower garments under different degrees of obesity in this application. Part (a) is the front and back breakdown of personalized model one, part (b) is the front and back breakdown of personalized model two, part (c) is the front and back breakdown of personalized model three, and part (d) is the front and back breakdown of personalized model four.

[0037] Figure 4 The diagrams are model diagrams with overweight parameters of 0.4, 0.6, 0.8, and 1.0. Part (a) in the diagram is global flattening, and part (b) is segmented folding and assembly.

[0038] Figure 5 The diagram shows the flattening and extraction process of the original pattern for the upper body in this application. Part (a) shows the flattening and extraction process of the original pattern for the front piece of the upper body, part (b) shows the flattening and extraction process of the original pattern for the back piece of the upper body, part (c) shows the original pattern for the upper body of personalized model one, part (d) shows the original pattern for the upper body of personalized model two, part (e) shows the original pattern for the upper body of personalized model three, and part (f) shows the original pattern for the upper body of personalized model four.

[0039] Figure 6 The diagram shows the flattening and extraction process of the prototype pattern for the lower garment in this application. Part (a) shows the flattening and extraction process of the prototype pattern for the front piece of the lower body, part (b) shows the flattening and extraction process of the prototype pattern for the back piece of the lower body, part (c) shows the prototype pattern for the lower garment of personalized model one, part (d) shows the prototype pattern for the lower garment of personalized model two, part (e) shows the prototype pattern for the lower garment of personalized model three, and part (f) shows the prototype pattern for the lower garment of personalized model four.

[0040] Figure 7 This is a schematic diagram of the assembly process of the yoga garment pattern in this application. Part (a) in the diagram is the yoga top, and part (b) is the yoga shorts.

[0041] Figure 8 The diagram shows the dimensions of each measurement point on the paper pattern. In the diagram, part (a) is the front of the upper garment, part (b) is the north side of the upper garment, part (c) is the front of the lower garment, and part (d) is the back of the lower garment.

[0042] Figure 9 The diagram shows the wearing status under different overweight parameters. Part (a) is the front view, part (b) is the back view, and part (c) is the side view. Detailed Implementation

[0043] To make the technical problems, technical solutions, and beneficial effects to be solved by this application clearer, the technical solutions in the embodiments of this application will be further described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit the technical solutions of this application. All other embodiments obtained by those skilled in the art based on the embodiments in this application without creative effort are within the scope of protection of this application.

[0044] It should be noted that when a component is referred to as "fixed to" or "set on" another component, it can be located directly or indirectly on that other component. When a component is referred to as "connected to" another component, it can be directly or indirectly connected to that other component. The terms "upper," "lower," "left," "right," "front," "rear," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicate the orientation or position based on the orientation or position shown in the accompanying drawings, and are only for ease of description and should not be construed as limiting the present technical solution.

[0045] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or specifying the number of technical features. "A plurality of" means two or more, unless otherwise explicitly defined.

[0046] This embodiment provides a method for generating personalized patterns for plus-size yoga wear, specifically using plus-size women's yoga wear as the object of generation, combined with... Figure 1 Perform the following operations:

[0047] Part 1: Obtaining Personalized 3D Models

[0048] First, 3D body shape data is collected.

[0049] Three basic female 3D human body models were obtained. These three models maintained consistency in height, skeleton, proportions, and posture, differing only in the degree of fat distribution, corresponding to thin, medium, and obese body types, respectively. Based on this, to achieve continuous generation of body types with different obesity levels, continuous deformation parameters were set for the basic human body models in the 3D modeling environment. These parameters were used to adjust the surface morphology of the human body model in areas such as the abdomen, chest, buttocks, and thighs, with values ​​ranging from 0 to 1, such as 0.4, 0.6, 0.8, and 1.0.

[0050] When the continuous deformation parameter is at an intermediate value, the human body model appears to be of medium build; as the parameter value increases, the local fat accumulation characteristics of the human body gradually intensify. By adjusting the continuous deformation parameter while maintaining the topology of the human body mesh, multiple plus-size female 3D human body models with different obesity levels are generated, thereby ensuring topological consistency and morphological continuity between models of different body types.

[0051] Secondly, human body dimensions are extracted and body shape features are analyzed.

[0052] After obtaining 3D human body models of different obesity levels, the body dimensions are extracted from the models, such as... Figure 2 As shown.

[0053] Specifically, key measurement locations related to the generation of yoga garment patterns are determined on a 3D human body model, and circumference, height, and length parameters are extracted. Circumference parameters include bust, waist, hip, and thigh circumference; height parameters include bust height, waist-to-hip height, and crotch length; and length parameters include front and back width, shoulder length, and pant length.

[0054] Table 1: Personalized Female Model Body Measurement Data .

[0055] By extracting and organizing the human body size parameters, a quantitative basis is provided for subsequent human body model segmentation, paper pattern structure design, and size control.

[0056] Part Two: Design and Segmentation.

[0057] To reduce the deformation error introduced by flattening complex curved surfaces, the 3D human body model is segmented.

[0058] Specifically, combining the anatomical features of the human body with the design principles of women's clothing patterns, dividing lines are drawn on the surface of the 3D human body model, dividing the model into multiple independent 3D parts. These divisions include the upper body front piece, upper body back piece, and lower body front and back pieces. The upper body front piece is divided around the key points of the chest, while the upper body back piece is divided around the scapula. The lower body is divided with reference to the protruding abdominal area of ​​the front piece and the protruding buttocks area of ​​the back piece. This division method divides the human body model into multiple 3D parts with relatively small curvatures, reducing the surface complexity of any single part.

[0059] The human body model was divided into 27 regions, with the front upper body divided into 8 regions and the back upper body divided into 5 regions (see...). Figure 2 The lower body is divided into 7 sections each for the front and back pieces (see...). Figure 3 ).

[0060] Part Three: Pattern Flattening and Optimization The main steps include four steps: 3D surface analysis of components, mass spring algorithm, large-size prototype paper pattern, and quantitative and qualitative index analysis.

[0061] Among them, the core idea of ​​3D clothing flattening based on the mass (point mass)-spring algorithm model is to discretize a continuous object into a set of point masses (nodes) and connect them with springs (connectors) to simulate its elasticity and deformation characteristics (Xiao et al., 2024). The principle of 3D mesh flattening based on the point mass-spring model is as follows: Figure 4 As shown, the global flattening results are obtained for different overweight parameters (e.g., 0.4, 0.6, 0.8, 1.0) (see...). Figure 4 (a) and segmented folding and assembly (see section (a)) and segmented folding and assembly (see section (a)). Figure 4 (part (b) of the text).

[0062] The algorithm in this embodiment aims to unfold a 3D model (OBJ file format) or its projected form into a plane. It uses a spring-mass model to elastically constrain and optimize the model's edge points, ensuring that the side lengths of the unfolded model remain consistent and achieving the most natural possible layout in a 2D plane. Its core principles can be summarized as follows: (1) Model discretization and boundary extraction: By reading the OBJ file, the vertices (masses) of the model are defined as three-dimensional. .

[0063] (2) Principal Component Analysis (PCA) Projection Plane: The covariance matrix of the point set is used to perform eigenvalue decomposition to obtain the principal plane orientation of the model, thereby defining a two-dimensional projection coordinate system.

[0064] The corresponding mathematical expression is: .

[0065] Its characteristic decomposition: .

[0066] (3) Spring Force Model and Energy Minimization: In the two-dimensional projected space, the goal of a node is to adjust its position through the spring force, thereby reducing the degree of bending and ensuring reasonable deformation while maintaining the boundary length. In each iteration, the force F applied by the spring is... ab Described by Hooke's Law: .

[0067] Where: Xa and Xb are the two-dimensional coordinates of the two points connected by the spring in the plane; K is the spring stiffness coefficient; The distance between the two points.

[0068] Simultaneously, by applying spring force to the nodes and considering damping, iterative updates are performed using Newton's laws of motion: .

[0069] In the numerical simulation, the finite difference method is used to adjust the node positions in time steps, gradually approaching the energy minimum state.

[0070] (4) Adaptive Time Step Control: The time step is dynamically adjusted based on the target optimization values ​​of the mesh expansions in the previous and subsequent expansions. The target optimization value is taken as the average relative error of the mesh edge lengths (Chen et al., 2007). Let the length of the n edges connected to the j-th vertex be Li, and the corresponding length in the k-th iteration be... The relative error EL at the j-th point j for: .

[0071] Among them, Li, These are the lengths of the i-th edge connected to the j-th vertex on the spatial grid and the planar grid, respectively.

[0072] If the current optimization target value is smaller than the previous optimization value, then increase the time step. Conversely, the time step is reduced. The formula for dynamically adjusting the time step is: .

[0073] Among them, E current and E Last These represent the target optimization values ​​for the current and previous iterations, respectively.

[0074] To avoid the time step from oscillating between two extreme values ​​in practical applications, the algorithm was modified as follows: when the current optimized value is detected to be smaller than the previous optimized value, the step size is not increased immediately, but the iteration continues at the current time step size; if the optimized value has not increased or diverged in 5 consecutive iterations, the time step size is increased; once the optimized value increases, the time step size is immediately reduced.

[0075] (5) Back projection and model reconstruction: After spring optimization in the two-dimensional plane, the two-dimensional coordinates are mapped back to three-dimensional space, and the z-axis coordinate is fixed (usually 0) to obtain the unfolded three-dimensional model: .

[0076] The algorithm combines physical simulation and principal component analysis (PCA) to achieve a flattening process that unfolds complex 3D curved surface meshes into 2D planes. While preserving the original mesh topology and local geometric properties, a dynamically adjusted time step strategy is employed to effectively improve convergence speed and algorithm stability, making it suitable for surface flattening applications in the clothing industry.

[0077] To achieve surface flattening of segmented parts of a 3D human body model, this application employs a deformation algorithm based on a mass-spring model. The system hardware configuration consists of an Intel CPU (3.40 GHz) and 8 GB of memory. In this algorithm, the deformation process aims to minimize the energy within the system, ensuring the integrity of the geometric structure and minimizing deformation. Specific steps include: resetting the spring parameters in two-dimensional space, adjusting the spring length, and gradually adjusting the positions of all mass points using an iterative optimization method until the system reaches a stable equilibrium state. After this processing, the flattening of the 3D human body part model was successfully achieved, and the relevant results are as follows: Figures 5 to 7 As shown.

[0078] Figure 5 Parts (a) and (b) in the text Figure 7 Part (a) shows the process of flattening and extracting the pattern for the top; Figure 5 Parts (c) to (f) are, respectively, the original patterns of the tops under different degrees of obesity (corresponding to overweight parameters = 0.4, 0.6, 0.8, and 1.0) according to Personalized Model 1 to Personalized Model 4. Figure 6 Parts (a) and (b) in the text Figure 7 (b) shows the process of flattening and extracting the prototype paper pattern for the lower body; Figure 6 Parts (c) to (f) are, in order, the lower body prototype paper patterns of Personalized Model 1 to Personalized Model 4 under different degrees of obesity (corresponding to overweight parameters = 0.4, 0.6, 0.8, and 1.0 respectively).

[0079] To verify the dimensional accuracy of the paper pattern generated based on the 3D flattening algorithm, this application conducted key dimensional measurements on the flattened paper pattern of a plus-size female human body model. Combined with... Figure 8Comparing these paper pattern measurements with the original 3D scan data revealed that the difference between the measured bust circumference ((front bust + back bust) × 2) and the original bust circumference was mainly concentrated within the range of -1.6 to 4.3 cm, which is within a reasonable range. The deviations of major circumferences such as waist circumference ((front waist + back waist) × 2) and hip circumference ((front hip + back hip) × 2) were all controlled within ±3 cm, meeting the industrial allowable tolerances specified in GB / T 1335.2-2008 "Women's Clothing Sizes". Further analysis revealed that dimensions such as back width, front and back crotch width, back center length, and trouser length showed a significant linear increasing trend with increasing size, thus verifying the rationality of the proposed paper pattern scaling system from both mathematical logic and engineering implementation perspectives.

[0080] For the specific body characteristics of plus-size women, it was found that as the overweight parameter gradually increased from 0.4 to 1.0, the pattern structure lines in the abdominal protrusion and gluteal curvature areas showed significant changes in convexity. This indicates that the pattern acquisition method of first segmenting and then flattening the pattern, adopted in this application, can better preserve the expression and changes of body characteristics in the pattern structure.

[0081] Table 2: Key Dimensions of Flattened Paper Patterns .

[0082] To verify the superiority of the "segmentation before flattening" method in handling the geometric features of garment patterns, its effects were compared with the "overall flattening" method. Taking the upper body right front piece as an example: while direct overall flattening is simple, it easily causes local deformation and stretching in complex curved areas, resulting in wrinkles. Furthermore, this rigid flattening process is not conducive to subsequent pattern optimization. The segmentation before flattening method, however, effectively controls deformation, better conforms to the human body contour, has clear lines and structure, and results in a reasonable overall structure after splicing, without obvious deformation or wrinkles. When dealing with the complex curves of plus-size women, the segmentation and splicing method better preserves local details and reflects body shape changes.

[0083] Part Four: Personalized Pattern Generation and Simulated Fitting Based on the obtained two-dimensional prototype paper pattern, the paper pattern is structurally adjusted according to the style and structure of yoga clothing and the characteristics of knitted elastic fabric.

[0084] Specifically, differentiated weight reduction is implemented for key areas such as the bust, waist, hips, and thighs to improve wearing comfort while ensuring a good fit. By adjusting the structural lines of the prototype pattern, personalized two-dimensional patterns for yoga clothing suitable for different obesity levels are generated.

[0085] The generated personalized yoga outfit 2D pattern (see) Figure 7 Import Style 3D for virtual stitching and apply it to the corresponding 3D human body model.

[0086] By simulating the human standing posture and various typical yoga poses, the wearing status of yoga clothing in different postures was analyzed (e.g. Figure 9 As shown in the figure, the flatness of the garment surface and the pressure distribution in the garment-human contact area are evaluated to verify the fit and comfort of the generated paper pattern.

[0087] The above-described embodiments merely illustrate several feasible implementations of the present invention, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of the present invention, nor are the embodiments intended to limit the scope of protection in the claims of the present invention. For those skilled in the art, various modifications and improvements can be made without departing from the concept of the present invention. All equivalent implementations or changes that do not depart from the present invention should be included in the technology of the present invention.

Claims

1. A method for generating personalized patterns for plus-size yoga wear, characterized in that, The steps are as follows: Step 1: Obtain three basic 3D human body models: thin, medium, and obese. Then, by establishing continuous deformation parameters, perform seamless topological deformation on these basic 3D human body models. The continuous deformation parameters include ultra-thin parameters and ultra-heavy parameters, and the parameter values ​​range from 0 to 1. The seamless topological deformation satisfies the requirement of generating a personalized three-dimensional human body model with four levels of obesity gradient under the condition of only adjusting the overweight parameter. The personalized three-dimensional human body model includes a slightly overweight model and three personalized body shape models. Step 2: Based on the personalized 3D human body model, perform 3D human body size measurement and extract human body measurement parameters related to the generation of plus-size yoga clothing patterns; Step 3: Based on the principles of human anatomy and paper pattern design, the human body area covered by the yoga clothes is divided into sections using line drawing and cutting tools in 3D software, and multiple 3D component models are extracted. Step four: Apply the improved mass-spring algorithm to flatten the surfaces of each of the three-dimensional component models to obtain the corresponding two-dimensional prototype paper patterns. The improved mass-spring algorithm refers to: reading the three-dimensional mesh model of each component, extracting vertices as a set of points in three-dimensional space, taking the vertices as mass points and establishing a spring connection set based on the adjacency relationship of the face patches, so that each spring connects two mass points and its static equilibrium length is taken as the corresponding side length in the three-dimensional mesh, ensuring that the boundary length remains consistent during the flattening process; Step 5: Based on the two-dimensional prototype paper pattern, perform size pre-shrunk and style structure transformation to generate a personalized plus-size women's yoga clothing two-dimensional paper pattern. Step six: The personalized plus-size women's yoga clothing two-dimensional paper pattern is virtually sewn and tried on in a virtual fitting system, and the pressure distribution under standing posture and two typical yoga postures is analyzed to verify the fit and comfort of the paper pattern.

2. The method for generating personalized patterns for plus-size yoga wear according to claim 1, characterized in that: In step one, both the ultra-thin and overweight parameters are 0, corresponding to the medium type of the basic three-dimensional human body model.

3. The method for generating a personalized pattern for plus-size yoga clothing according to claim 1, characterized in that: In step one, the overweight parameters are set to 0.4, 0.6, 0.8 and 1.

0.

4. A method for generating personalized patterns for plus-size yoga wear according to claim 1, characterized in that, In step three, the segmentation is completed in Geomagic Wrap software: first, the human body area related to the yoga clothing pattern design is trimmed out, then the area is segmented by the scribing tool and the parts are extracted by the cutting tool, dividing the human body model into 27 areas, of which the upper body front piece is divided into 8 areas, the upper body back piece is divided into 5 areas, and the lower body front and back pieces are each divided into 7 areas.

5. The method for generating a personalized pattern for plus-size yoga wear according to claim 1, characterized in that, In step four, the covariance matrix of each point set of the three-dimensional component model is used to perform eigenvalue decomposition, determine the main planar orientation of the component, and establish a two-dimensional projected coordinate system.

6. The method for generating a personalized pattern for plus-size yoga clothing according to claim 1, characterized in that: In step four, the spring force is calculated in the two-dimensional projection space according to Hooke's law, and combined with the damping effect, the two-dimensional coordinates of the mass point are iteratively updated using Newton's laws of motion.

7. The method for generating a personalized pattern for plus-size yoga clothing according to claim 1, characterized in that: In step four, the time step is dynamically adjusted based on the target optimization values ​​of the two mesh expansions. The target optimization value is the average relative error of the mesh side length. When it is detected that the current optimization value is smaller than the previous optimization value, the time step is not immediately increased, but the iteration continues at the current time step. If the optimization value does not increase or diverge after 5 consecutive iterations, the time step is increased; if the optimization value increases, the time step is immediately reduced.

8. The method for generating a personalized pattern for plus-size yoga clothing according to claim 1, characterized in that: In step five, the size pre-shrinking, while adhering to ergonomic principles and the elastic properties of the knitted fabric, involves differentiated reduction of key areas.

9. The method for generating a personalized pattern for plus-size yoga wear according to claim 1, characterized in that, In step five, the style modification refers to: the top being sleeveless with wide shoulder straps, and the top having a slimming effect achieved through two intersecting dividing lines on the side; and the bottom being high-waisted fitted shorts.

10. The method for generating a personalized pattern for plus-size yoga clothing according to claim 1, characterized in that: In step six, the virtual try-on includes standing postures as well as two typical yoga poses: Tree Pose and Warrior I.

Images