Virtual fitting method based on color three-dimensional model

Abstract

The invention provides a virtual fitting method based on a color three-dimensional model, and the method comprises the steps: obtaining the body point cloud data of a user through a three-dimensional scanning device, constructing a color three-dimensional human body model, and setting skeleton binding parameters; dividing a pressure sensitive area based on model surface curvature distribution, and generating a dynamic pressure map; performing cloth physical simulation on the model in a physical engine, and extracting deformation response data of the top points of the clothing grids; real-time motion posture data of a user is collected through an inertial sensor, motion distortion correction is conducted, and a joint motion track is generated. Dynamically comparing the joint movement track with the deformation response data to generate a clothing fitting degree deviation report; and optimizing the clothing pattern parameters and the fabric physical attributes based on the deviation report, and outputting a clothing adaptation proposal. According to the invention, the suitability and comfort of clothes can be improved, and the production efficiency and the product quality are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of virtual fitting, and more particularly to a virtual fitting method based on a color three-dimensional model. Background Art

[0002] In the field of clothing production and fitting, traditional fitting methods have always had many limitations. When purchasing clothing, consumers often need to try on clothing of different sizes multiple times to find the style that suits them best. This process is not only time-consuming and labor-intensive, but also particularly difficult for people with special body shapes to find suitable clothing. For clothing manufacturers, the traditional clothing production process, from design to pattern making to sample production, requires repeated revisions and adjustments to ensure that the clothing pattern and size meet the needs of different consumers. This process is not only costly but also inefficient.

[0003] While existing virtual fitting technology has addressed these issues to a certain extent, it still has many shortcomings. Current virtual fitting systems are mostly based on two-dimensional images or simple three-dimensional models, which cannot accurately simulate the actual wearing experience of clothing on the human body. These systems typically only provide static clothing displays and fail to account for the dynamic changes of clothing during human movement, resulting in significant deviations from actual fitting results. Furthermore, existing virtual fitting technology is also lacking in its ability to assess clothing fit and comfort, failing to provide consumers with accurate information about the wearing experience.

[0004] In the process of implementing the embodiments of the present invention, there are at least the following problems or defects in the existing technology: the existing virtual fitting technology cannot accurately simulate the dynamic wearing effect of clothing on the human body, and cannot take into account the deformation and fit changes of clothing during human movement; the existing virtual fitting system has deficiencies in the evaluation of clothing fit and comfort, and cannot provide consumers with accurate wearing experience information; the existing virtual fitting technology lacks effective means for optimizing clothing patterns, and cannot accurately adjust clothing patterns according to the consumer's body shape and movement posture. Summary of the Invention

[0005] The present invention provides a virtual fitting method based on a color three-dimensional model, comprising:

[0006] Step 1: Obtain user body point cloud data through a 3D scanning device, build a color 3D human body model based on the point cloud data, and set bone binding parameters;

[0007] Step 2: Based on the surface curvature distribution of the color three-dimensional human body model, divide the pressure sensitive area and generate a dynamic pressure map;

[0008] Step 3: Based on the garment draping dynamics parameters, perform cloth physics simulation on the colored three-dimensional human body model in a physics engine to extract deformation response data of garment mesh vertices;

[0009] Step 4: collecting real-time motion posture data of the user through an inertial sensor, performing motion distortion correction on the motion posture data, and generating a joint motion trajectory;

[0010] Step 5: Dynamically compare the joint motion trajectory with the deformation response data to generate a garment fit deviation report;

[0011] Step 6: Optimize clothing pattern parameters and fabric physical properties based on the clothing fit deviation report, and output a clothing fit recommendation.

[0012] Furthermore, the step 1 includes:

[0013] Step 1.1: Obtain the user's full body point cloud data through multi-angle depth scanning;

[0014] Step 1.2: constructing a human body surface model based on the point cloud data and extracting key anatomical landmarks;

[0015] Step 1.3, mapping the skin layer compression modulus parameter from the biomechanical property library to the human body surface model;

[0016] Step 1.4, setting the rotational freedom constraint conditions of the main joint points based on the key anatomical landmarks;

[0017] Step 1.5: Verify the displacement range of the main joint points through the posture interpolation algorithm to generate model deformation tolerance data.

[0018] Furthermore, the step 2 includes:

[0019] Step 2.1: using a finite element stress analysis tool to identify high-friction areas of the color 3D human body model, wherein the high-friction areas include the axillary curved surface and the knee flexion surface;

[0020] Step 2.2, setting a density gradient distribution scheme of the pressure sensor based on the contour of the high friction area;

[0021] Step 2.3: Verify whether the pressure sensor covers the seam area of ​​the garment, and complete the sensor deployment in the uncovered area;

[0022] Step 2.4: discretize the pressure sensitive area into a set of triangular facets using a triangular mesh partitioning algorithm;

[0023] Step 2.5: Dynamically adjust the sensor density based on the curvature change of the triangular facet set to generate a dynamic pressure map.

[0024] Furthermore, the step 3 includes:

[0025] Step 3.1. Load the garment pattern vector diagram into the physics engine and input the warp yarn elastic modulus and weft yarn bending stiffness parameters.

[0026] Step 3.2, simulating the draping state of the clothing on the surface of the colored three-dimensional human body model based on the physical engine, and calculating the displacement vectors of the mesh vertices;

[0027] Step 3.3, generating a vertex normal vector offset map based on the displacement vector;

[0028] Step 3.4, mapping the stress scalar value in the vertex normal vector offset map to the HSV color wheel, where blue represents low pressure areas and red represents high pressure areas;

[0029] Step 3.5: Project the three-dimensional stress distribution into a two-dimensional thermal map using UV unfolding technology.

[0030] Furthermore, the step 4 includes:

[0031] Step 4.1, collect six-degree-of-freedom motion data of joint points through a multi-axis inertial sensor group;

[0032] Step 4.2: Using a Kalman filter to eliminate drift errors of the six-degree-of-freedom motion data;

[0033] Step 4.3: Calculate the joint rotation angle based on the kinematic chain model;

[0034] Step 4.4: Construct a time series dataset of joint motion trajectories;

[0035] Step 4.5: Fill in the missing frames in the time series dataset using an interpolation algorithm.

[0036] Furthermore, the step 5 includes:

[0037] Step 5.1, extracting the clothing mesh deformation rate of the key frame in the physical simulation;

[0038] Step 5.2: Perform matrix difference operation on the real-time joint rotation angle and the simulated joint rotation angle to generate a rotation deviation matrix;

[0039] Step 5.3: Calculate the deviation gradient of the deformation rate and the pressure threshold based on the rotation deviation matrix;

[0040] Step 5.4: Set multiple deformation rate thresholds, including a visual distortion warning threshold and a physical deformation exceeding threshold.

[0041] Step 5.5: Generate a fit deviation report with a timestamp.

[0042] Furthermore, the step 6 includes:

[0043] Step 6.1, analyzing the curvature deviation of the garment underarm triangle panel;

[0044] Step 6.2: Adjust the fabric Poisson's ratio parameter based on the fabric stretch in the knee flexion area.

[0045] Step 6.3, optimize the contour line of the cutting piece by using the B-spline curve reconstruction algorithm;

[0046] Step 6.4: Inject curvature redundancy into the knee bend area and superimpose warp elastic compensation allowance on the side seams;

[0047] Step 6.5: Convert the optimized template parameters into JSON structured data and output it.

[0048] Furthermore, the density gradient distribution scheme includes:

[0049] Step 8.1, deploying a ring sensor array in the axillary area of ​​the color 3D human body model, with each sensor being distributed at a preset interval;

[0050] Step 8.2: Using a diamond-shaped sensor array on the dorsal region, with each sensor being distributed at a first preset spacing;

[0051] Step 8.3: Using a spiral sensor array in the limb area, with each sensor distributed at a second preset spacing;

[0052] Step 8.4: Dynamically increase the distance between adjacent sensors based on the density of clothing seams.

[0053] Step 8.5: Adjust the sensor sampling frequency based on real-time pressure feedback.

[0054] Furthermore, the generation of the vertex normal vector offset map includes:

[0055] Step 9.1. Calculate the stress distribution within the triangular mesh using thin plate theory.

[0056] Step 9.2: Map the stress scalar value to the continuous gradient interval of the HSV color circle;

[0057] Step 9.3, use UV parameterization to unfold the three-dimensional stress distribution surface;

[0058] Step 9.4: Mark the spatial correlation between the high-pressure area and the seam line of the garment in the two-dimensional heat map;

[0059] Step 9.5: Generate a clothing pattern optimization weight coefficient matrix based on the two-dimensional heat map.

[0060] Furthermore, the setting of the curvature redundancy includes:

[0061] Step 10.1, measuring the tensile deformation of the rear panel of the trousers when the knee is flexed at a preset angle;

[0062] Step 10.2, inversely calculating the back crotch line curvature compensation value based on the stretching deformation;

[0063] Step 10.3: Create a dynamic pattern parameter template in the cutting piece CAD file;

[0064] Step 10.4, injecting curvature redundancy into the knee bend area;

[0065] Step 10.5: Add warp elastic compensation allowance to the side seams.

[0066] The above embodiments of the present invention have at least the following beneficial effects:

[0067] 1. By acquiring point cloud data of the user's body through 3D scanning equipment and constructing a color 3D human model, and simultaneously setting skeletal binding parameters, the system accurately reflects the user's physical characteristics and motion state, providing an accurate foundation for subsequent clothing simulation and adaptation. This enables the virtual fitting system to provide a personalized fitting experience for users of different body shapes and athletic requirements, resolving the problem that traditional fitting methods struggle to meet diverse needs.

[0068] 2. Based on the surface curvature distribution of the color 3D human body model, the system divides pressure-sensitive areas and generates dynamic pressure maps. Combined with fabric physics simulation within the physics engine, this system simulates the deformation and pressure distribution of clothing during human movement in real time. This technical feature enables the system to accurately assess the fit and comfort of clothing, providing users with detailed wearing experience information and providing manufacturers with a basis for optimizing clothing patterns and fabric properties. This addresses the shortcomings of existing virtual fitting technologies in terms of dynamic simulation and comfort assessment.

[0069] 3. Inertial sensors collect real-time user motion data and perform motion distortion correction to generate joint motion trajectories. This data is dynamically compared with the deformation response data of the garment mesh vertices to generate a garment fit deviation report. Based on this report, garment pattern parameters and fabric physical properties are optimized, and a garment fit recommendation is output. This series of technical features establishes a closed-loop feedback mechanism from user motion data collection to garment pattern optimization, effectively improving garment fit and wearing comfort, and addressing the inefficient and costly pattern adjustment issues in traditional garment production processes. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] The above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily apparent by reading the following detailed description with reference to the accompanying drawings, in which several embodiments of the present invention are shown by way of example and not limitation, in which:

[0071] Figure 1 A flowchart of a virtual fitting method based on a color 3D model provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0072] The technical solutions of this application will be described clearly and completely below, in conjunction with the accompanying drawings. It should be understood that the described embodiments represent only a portion of the embodiments of this application, and not all of them. The components of this application, generally described and illustrated in the drawings herein, may be arranged and designed in a variety of different configurations. Therefore, the following detailed description of the embodiments of this application provided in the drawings is not intended to limit the scope of the claimed application, but rather merely represents selected embodiments of this application. All other embodiments derived by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application. It should be noted that similar reference numerals and letters represent similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined or explained in subsequent drawings. Furthermore, in the description of this application, the terms "first," "second," etc., are used solely to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0073] This application proposes that Figure 1 As shown, the following steps are included:

[0074] Step 1: Obtain user body point cloud data through a 3D scanning device, build a color 3D human body model based on the point cloud data, and set bone binding parameters;

[0075] Step 2: Based on the surface curvature distribution of the color three-dimensional human body model, divide the pressure sensitive area and generate a dynamic pressure map;

[0076] Step 3: Based on the garment draping dynamics parameters, perform cloth physics simulation on the colored three-dimensional human body model in a physics engine to extract deformation response data of garment mesh vertices;

[0077] Step 4: collecting real-time motion posture data of the user through an inertial sensor, performing motion distortion correction on the motion posture data, and generating a joint motion trajectory;

[0078] Step 5: Dynamically compare the joint motion trajectory with the deformation response data to generate a garment fit deviation report;

[0079] Step 6: Optimize clothing pattern parameters and fabric physical properties based on the clothing fit deviation report, and output a clothing fit recommendation.

[0080] A color 3D human body model is a digital model of the human body with biomechanical properties constructed using point cloud data acquired by a 3D scanning device. This can be achieved using multi-angle depth scanning combined with a surface reconstruction algorithm to create an anatomical model that includes skin compressive modulus and skeletal motion constraints. Dynamic pressure mapping refers to the distribution of pressure-sensitive areas based on the curvature of the human body surface. This can be achieved using finite element stress analysis combined with a triangular meshing algorithm to identify stress concentration areas when clothing contacts the body. Fabric physics simulation refers to the dynamic process of simulating the draping state of clothing within a physics engine. This can be achieved using the warp yarn elastic modulus and weft yarn bending stiffness parameters as input variables to extract deformation response data of the clothing mesh vertices during motion. Joint motion trajectory refers to the motion path data of joint points collected and corrected by inertial sensors. This can be achieved using a multi-axis inertial sensor array combined with a Kalman filter algorithm to eliminate posture drift errors during motion capture. The garment fit deviation report refers to a quantitative assessment result generated by comparing joint motion trajectories with garment deformation response data. This can be achieved by using a matrix difference operation combined with a deformation rate threshold classification algorithm to identify areas of garment fit defects during dynamic movement. Garment pattern parameter optimization refers to the process of adjusting the contours of the cut pieces and fabric properties based on fit deviation data. This can be achieved by using a B-spline curve reconstruction algorithm combined with Poisson's ratio parameter adjustment to inject curvature redundancy and elastic compensation margins into the knee bend and side seam areas.

[0081] The core innovation of this application lies in the construction of a closed-loop feedback system from 3D human body modeling to garment pattern optimization, integrating biomechanical modeling, dynamic pressure sensing, and physics engine simulation technologies. This system enables precise assessment and adaptive adjustment of garment fit based on dynamic motion data. By comparing joint motion trajectories captured by inertial sensors with garment deformation data generated by the physics engine in real time, a correlation model between human kinematic parameters and garment mechanical response is established, addressing the technical difficulty of existing technologies in quantitatively assessing dynamic fit.

[0082] The working process and principle of this application are as follows: first, point cloud data of the user's body is acquired through a 3D scanning device. Based on this point cloud data, a color 3D human body model is constructed and skeletal binding parameters are set. This step establishes a digital human body foundation with anatomical features and biomechanical properties. Next, based on the surface curvature distribution of the color 3D human body model, pressure-sensitive areas are divided and a dynamic pressure map is generated. This step can accurately identify stress concentration areas when the garment contacts the human body.

[0083] Furthermore, based on the garment's draping dynamics parameters, a cloth physics simulation is performed on the color 3D human model in a physics engine, extracting deformation response data from the garment mesh vertices. This step achieves physically accurate simulation of the garment's dynamic deformation. Subsequently, inertial sensors are used to collect the user's real-time motion posture data, which is then corrected for motion distortion to generate joint motion trajectories. This step addresses the issue of posture distortion caused by drift in motion capture data.

[0084] Next, the joint motion trajectory is dynamically compared with the deformation response data to generate a garment fit deviation report. This step establishes a correlation model between human kinematic parameters and garment deformation, providing data support for quantitative evaluation of garment dynamic fit. Finally, based on the garment fit deviation report, the garment pattern parameters and fabric physical properties are optimized, and a garment fit recommendation is output. This step establishes a positive feedback mechanism from dynamic fitting to pattern optimization, enabling personalized and precise adjustment of garment fit.

[0085] Therefore, this application realizes closed-loop control from human body modeling to clothing optimization by constructing a virtual fitting system that combines biomechanical properties with dynamic motion.

[0086] As a preferred embodiment, the solution of this application is specifically implemented as follows:

[0087] First, a high-precision 3D laser scanner is used to scan the user's entire body, generating point cloud data containing geometric shape and surface texture information. Based on this point cloud data, a mesh reconstruction algorithm is used to construct a color 3D human model. Simultaneously, a skeleton detection algorithm is used to identify key skeletal points and set the corresponding skeletal binding parameters.

[0088] Next, the color 3D human body model is subjected to surface curvature analysis to identify areas of high curvature as potential pressure-sensitive regions. Based on the curvature distribution characteristics, an adaptive meshing algorithm is used to generate a dynamic pressure map for subsequent accurate pressure distribution simulation.

[0089] Furthermore, the physics engine imports the garment's draping dynamics parameters, including the fabric's elastic modulus, Poisson's ratio, and coefficient of friction. Fabric physics simulation is performed on a color 3D human body model, simulating the garment's deformation under the effects of gravity and human motion. Deformation response data is extracted by tracking the displacement and deformation of the garment mesh's vertices.

[0090] Subsequently, a multi-axis inertial sensor is used to collect real-time motion data of the user during different movements. This data is processed using a Kalman filter to eliminate the accumulated errors caused by sensor drift. This corrected data is then used to generate accurate joint motion trajectories.

[0091] Next, the joint motion trajectory data is time-series aligned and spatially registered with the previously acquired garment deformation response data. A dynamic comparison algorithm is used to calculate the garment's fit deviation under different motion states, generating a fit deviation report containing timestamps and spatial position information.

[0092] Finally, based on the fit deviation report, an optimization algorithm is used to adjust the garment pattern parameters. Adjustments include correcting the dimensions of key areas and optimizing the placement of seams. Furthermore, fine-tuning the fabric's physical properties is performed based on the deviations, such as adjusting the elastic modulus or adding local reinforcements. Finally, a detailed garment fit recommendation is generated to guide subsequent customized garment production.

[0093] This application further proposes to obtain the user's full-body point cloud data through multi-angle depth scanning; construct a human body surface model based on the point cloud data and extract key anatomical landmarks; map the skin layer compression modulus parameters from the biomechanical property library to the human body surface model; set the main joint rotational freedom constraint conditions based on the key anatomical landmarks; verify the displacement range of the main joints through the posture interpolation algorithm, and generate model deformation tolerance data.

[0094] Among them, multi-angle depth scanning can be achieved using a circular scanning array or a movable scanning device, for example, using six depth sensors distributed at equal angles on a ring bracket to synchronously collect data. The extraction of key anatomical landmarks can be based on the identification of areas with sudden changes in curvature in the point cloud, such as bony landmarks such as the anterior superior iliac spine and the acromion process, and the positioning error can be controlled within the range of ±1.5 mm. The mapping of the skin layer compression modulus parameters is achieved through the age-sex-body mass index correlation model in the biomechanical property library, for example, mapping the standard parameters of a 25-year-old male to the corresponding surface mesh. The setting of the main joint rotational freedom constraint conditions includes sagittal plane flexion and extension angle restrictions and coronal plane abduction angle restrictions, for example, the hip joint flexion range is set to 0-120 degrees. The posture interpolation algorithm can use cubic spline interpolation to generate deformation tolerance data for three key frames of knee flexion 30 degrees, 60 degrees, and 90 degrees.

[0095] Specifically, full-body point cloud data is acquired through multi-angle depth scanning. Six depth sensors are arranged in a circular pattern to collect data simultaneously, eliminating blind spots in single-view scanning and ensuring a point cloud coverage rate exceeding 98%. The human body surface model is constructed using a Poisson surface reconstruction algorithm. Key anatomical landmarks are automatically identified using a curvature gradient detection algorithm. For example, in the scapular spine region, points with a curvature change exceeding 0.3 / mm² are used as markers. Skin compressive modulus parameters are mapped to the user's body fat percentage from a biomechanical property library. For example, a body fat percentage of 18-22% maps to a soft tissue elastic modulus of 25-35 kPa. Rotational constraints on major joints are based on International Society of Biomechanics standards, for example, limiting elbow pronation to a range of 0-80 degrees. A posture interpolation algorithm generates intermediate posture data along the joint motion trajectory. Inverse kinematics is used to verify that the compression of the posterior thigh soft tissue does not exceed 12% of its original length at 90 degrees of knee flexion. This generates deformation tolerance threshold data. This process controls the skeletal binding position deviation within 2 mm through the anatomical landmark positioning error compensation mechanism. At the same time, the dynamic mapping of biomechanical parameters reduces the clothing pressure simulation error to within ±8%.

[0096] As a preferred embodiment, the solution of this application is specifically implemented as follows:

[0097] First, a multi-angle depth scan is performed to obtain full-body point cloud data. Multiple depth cameras are used to perform a 360-degree scan around the user, with each camera capturing data at 30 frames per second, for a scan time of 10 seconds. During the scan, the user maintains a T-pose to ensure that every part of the body is fully captured.

[0098] Next, a human surface model was constructed based on the point cloud data, and key anatomical landmarks were extracted. A NURBS surface fitting algorithm was used to reconstruct the point cloud data, generating a highly accurate human surface model. Image processing and machine learning algorithms were then used to automatically identify and label 28 key anatomical landmarks, including the head vertex, neck center, acromion, elbow, wrist, hip, knee, and ankle.

[0099] Furthermore, skin layer compressive modulus parameters are mapped to the human surface model from a biomechanical property library. The biomechanical property library contains skin layer compressive modulus data for different ages, genders, and body types. Based on the user's personal information, the best matching parameter set is selected and mapped to the corresponding area of ​​the human surface model.

[0100] Then, we set rotational constraints for the main joints based on key anatomical landmarks. For example, the neck joint is set to three degrees of freedom for rotation, the shoulder joint is set to three degrees of freedom for rotation, the elbow joint is set to one degree of freedom for rotation, the hip joint is set to three degrees of freedom for rotation, and the knee joint is set to one degree of freedom for rotation. These constraints ensure that the model's motion conforms to human physiology.

[0101] Finally, a pose interpolation algorithm verifies the displacement range of the main joints and generates model deformation tolerance data. Joint motion is interpolated using the spherical linear interpolation algorithm (SLERP) to simulate joint displacements under different poses. Through repeated iterations, the maximum rotation angle and displacement range of each joint are determined, and a deformation tolerance data table is generated to provide a reference for subsequent dynamic simulations.

[0102] This application further proposes to identify high-friction areas of a color three-dimensional human body model through a finite element stress analysis tool, where the high-friction areas include the armpit curved surface and the knee flexion surface; set a density gradient distribution scheme for pressure sensors based on the contours of the high-friction areas; verify whether the pressure sensors cover the seam areas of the garments, and complete the sensor distribution in uncovered areas; use a triangular meshing algorithm to discretize the pressure-sensitive areas into a set of triangular facets; dynamically adjust the sensor density based on the curvature changes of the triangular facet set to generate a dynamic pressure map.

[0103] Among them, the finite element stress analysis tool can calculate the stress distribution in motion based on the skin layer compression modulus parameters and bone binding parameters, for example, using second-order tetrahedral units to divide the grid and set contact boundary conditions. In the density gradient distribution scheme, the sensor spacing in high curvature areas can be set to 3-5 mm, and the spacing in low curvature areas is extended to 8-12 mm. When verifying the coverage of clothing seams, the Euclidean distance between the seam trajectory and the sensor coordinates is calculated, and the completion algorithm is triggered when the distance exceeds the preset threshold. The triangular meshing algorithm can use Delaunay triangulation to divide the surface into triangular units with side lengths not exceeding a preset value. When dynamically adjusting the sensor density, the density update mechanism is triggered when the curvature change rate exceeds 0.15 radians per second.

[0104] Specifically, the finite element stress analysis tool simulates the stress distribution at the interface between soft tissue and clothing during human motion by loading skeletal binding parameters and motion posture data, accurately locating the actual boundaries of the underarm and knee flexion zones. When generating a sensor placement plan based on the contours of high-friction areas, denser monitoring points are placed at the intersection of seams, such as a circular sensor array around the intersection of the back crotch line of pants. After discretizing the surface using a triangular meshing algorithm, the curvature change of each triangle is monitored in real time. When the local curvature increases due to limb bending, the number of sensor sampling points in that area is automatically increased. During dynamic motion, changes in the curvature of the triangles in the knee flexion zone trigger adjustments to the sensor density, ensuring that the pressure map reflects the changes in clothing-skin contact pressure in real time. A completion algorithm also ensures the continuity of pressure data in the seam area. This solution combines mechanical analysis with geometric calculations to achieve adaptive optimization of the sensor placement plan, effectively improving the spatial resolution and temporal response accuracy of the dynamic pressure map.

[0105] As a preferred embodiment, the solution of this application is specifically implemented as follows:

[0106] First, a finite element stress analysis tool was used to analyze the color 3D human body model to identify high-friction areas. These areas included the axillary curved surface and the knee flexion surface. The finite element analysis used a tetrahedral meshing method, with a cell size of 0.5 mm to ensure accuracy.

[0107] Next, the density gradient distribution of pressure sensors is set based on the contours of the high-friction areas. In the armpit area, the sensor density can be set to 4 per square centimeter; on the knee flexion surface, the sensor density can be set to 3 per square centimeter; and in other areas, the sensor density can be set to 1 per square centimeter.

[0108] Furthermore, verify that the pressure sensors cover the garment seams. For any uncovered seam areas, complete the sensor placement. To complete the gap, add a row of sensors on each side of the seam, with a spacing of 5 mm.

[0109] Subsequently, a triangular meshing algorithm is used to discretize the pressure-sensitive area into a set of triangular facets. The side length of the triangular facets can be set to 10 mm to balance computational accuracy and efficiency.

[0110] Finally, the sensor density is dynamically adjusted based on the curvature changes of the triangular facets to generate a dynamic pressure map. When the local curvature change exceeds a preset threshold, the sensor density in that area can be doubled. For example, when the knee flexion angle exceeds 60 degrees, the sensor density on the flexion surface of the knee can be increased to 6 sensors per square centimeter.

[0111] This application further proposes loading a clothing pattern vector diagram into a physical engine and inputting the warp yarn elastic modulus and weft yarn bending stiffness parameters, simulating the draping state of clothing on the surface of a colored three-dimensional human body model based on the physical engine and calculating the displacement vectors of the mesh vertices, generating a vertex normal vector offset map, mapping the stress scalar value to the HSV hue ring, and projecting the three-dimensional stress distribution into a two-dimensional heat map through UV unfolding technology.

[0112] The loading of the garment pattern vector image must be aligned with the surface topology of the 3D human body model. The warp yarn elastic modulus can be set to a range of 100-500 MPa, and the weft yarn bending stiffness parameter can be set to 0.1-5.0 N·mm² / mm. Both are independently input to control the deformation characteristics in the warp and weft directions respectively. The displacement vector is calculated based on the cloth dynamics equation and implemented using a mass-spring model or finite element method. The displacement of each mesh vertex is recorded as 3D vector data. The normal vector offset map is generated using a triangular mesh in-plane stress distribution algorithm. The stress scalar value of each triangle is calculated using thin plate theory and converted into a normal vector direction change. In the HSV hue ring mapping rule, blue corresponds to the low pressure area of ​​0-0.3 MPa, and red corresponds to the high pressure area of ​​0.7-1.0 MPa. The intermediate pressure values ​​are assigned hues using linear interpolation. The UV unfolding process needs to maintain the topological correspondence between the three-dimensional model and the original pattern, convert the surface stress distribution into a planar thermal map through a parametric unfolding algorithm, and mark the overlapping area of ​​the suture line and the high-pressure area in the two-dimensional map.

[0113] Specifically, the garment pattern vector is imported into the physics engine via DXF format and spatially aligned with the surface curvature of the 3D human model, ensuring that the pattern boundaries align with the model's anatomical landmarks. During the draping simulation phase, the physics engine calculates the fabric's tensile stiffness based on the input warp elastic modulus. The weft bending stiffness parameter controls the degree of fabric wrinkling. These two parameters work together to ensure that the simulation results align with the mechanical behavior of the real fabric. After normalization, the displacement vector data is converted into normal vector offsets via a vector cross product, forming a scalar field that represents the local stress intensity. During HSV hue mapping, the pressure scalar values ​​are quantized into 256 color levels. The blue-to-red gradient covers the entire stress distribution range. The underarm area appears as a dark red patch due to concentrated contact pressure, while the hem area appears as a light blue distribution due to gravity. UV unfolding uses a conformal parameterization algorithm to unfold the 3D model surface into a 2D pattern consistent with the original garment pattern. The correspondence between high-pressure areas and pattern edges in the heat map is achieved through coordinate mapping. For example, the high-pressure band at the knee precisely corresponds to the curvature correction area on the back of the trousers in the 2D projection. Therefore, the three-dimensional stress data is converted into a plane map that can directly guide pattern optimization. The adjustment amount of the cutting contour line can be determined by measuring the offset distance between the color block coverage area in the heat map and the pattern boundary.

[0114] As a preferred embodiment, the solution of this application is specifically implemented as follows:

[0115] Load the garment pattern vector image into a physics engine and input the warp yarn elastic modulus and weft yarn bending stiffness parameters. For example, you can use the open-source physics engine Bullet Physics and use its cloth simulation module to load vector images exported from garment CAD software. Set the warp yarn elastic modulus to 200 GPa and the weft yarn bending stiffness to 0.5 N·m².

[0116] The physics engine simulates the draping of clothing on a colored 3D human body model and calculates the displacement vectors of mesh vertices. Specifically, the clothing model is placed above the human body model, and gravity simulates the natural drooping of the clothing. During the simulation, the initial and final positions of each mesh vertex are recorded, and the displacement vector is calculated.

[0117] Based on the displacement vectors, a vertex normal vector offset map is generated. Furthermore, the displacement vectors can be used to calculate the change in normal vector at each vertex, generating a normal vector offset map. This map reflects the local deformation of the garment surface.

[0118] The stress scalar values ​​in the vertex normal offset map are mapped to the HSV color wheel, where blue represents low-pressure areas and red represents high-pressure areas. This allows the calculated stress range to be divided into multiple intervals, each corresponding to a color in the HSV color wheel. For example, a hue angle range of 0-30° could be used to represent low-pressure areas, and a hue angle range of 300-360° could be used to represent high-pressure areas.

[0119] UV unwrapping technology is used to project the 3D stress distribution into a 2D heat map. Specifically, an isometric parameterization method is used to unwrap the 3D model into a 2D plane, preserving the stress distribution information. The resulting 2D heat map intuitively displays the stress distribution in various parts of the garment.

[0120] This application further proposes a technical solution including the following steps: extracting the clothing mesh deformation rate of the key frames in the physical simulation; performing matrix difference operation on the real-time joint rotation angle and the simulated joint rotation angle to generate a rotation deviation matrix; calculating the deviation gradient of the deformation rate and the pressure threshold based on the rotation deviation matrix; setting multi-level deformation rate thresholds, including a visual distortion warning threshold and a physical shape deformation exceeding limit threshold; and generating a fit deviation report containing a timestamp mark.

[0121] Keyframes are extracted based on the moment when the acceleration rate of the garment's deformation exceeds a preset threshold. For example, a keyframe capture is triggered when the acceleration rate exceeds 5% per second. A matrix difference operation uses Euclidean distance to calculate the three-dimensional spatial difference between the real-time and simulated rotation angles, generating a rotation deviation matrix containing the X / Y / Z axis deviations. The deviation gradient is calculated by iteratively solving the nonlinear mapping between deformation rate and pressure threshold using a gradient descent algorithm. The pressure threshold is dynamically adjusted based on the elastic modulus of the garment material. Multiple deformation rate thresholds are defined, including visual distortion warning thresholds and physical deformation thresholds. For example, the visual distortion warning threshold is set at a deformation rate of 5%-15%, and the physical deformation threshold is set at a deformation rate of 20%-30%. Timestamps are synchronized at the millisecond level to ensure spatiotemporal alignment of the motion trajectory with the deformation data.

[0122] Specifically, during the physical simulation process, keyframes are extracted by monitoring the second-order derivatives of the displacement vectors of the garment mesh vertices. Keyframe capture is triggered when a sudden change in displacement acceleration is detected. For example, the corresponding frame data is automatically captured when the knee flexion angle reaches 45 degrees. The matrix difference between the real-time and simulated joint rotation angles uses a quaternion difference calculation method to generate a rotation deviation matrix containing the rotation axis deviation angle and direction vector. Based on this matrix, a backpropagation algorithm is used to calculate the deviation gradient of the deformation rate and pressure threshold. The pressure threshold is dynamically adjusted based on the warp elastic modulus of the garment fabric. For example, when the elastic modulus falls below 2 GPa, the pressure threshold is automatically reduced by 10%. The visual distortion warning threshold and the physical deformation over-limit threshold are graded based on material mechanical property test data. For example, the visual distortion threshold for cotton fabric is set at 15%, while that for polyester fabric is set at 10%. In the generated fit deviation report, each abnormal deformation event is associated with a timestamp accurate to 10 milliseconds and synchronized with the skeletal animation timeline of the motion capture system, so that the deformation limit event in the armpit area at the 3.2th second of the movement can be accurately located in the middle stage of the arm lifting action. This technical solution combines keyframe focusing with matrix operations to concentrate computing resources on the movement stage with significant deformation, reducing the amount of calculation by more than 60% compared to the full-time data processing method; through quantitative modeling of deviation gradients, the nonlinear correlation analysis of joint motion deviation and clothing deformation is realized, and the pressure distribution prediction accuracy is improved to 92%; through the dual-threshold grading mechanism, the different risk levels of surface wrinkles and material failure are effectively distinguished, making the pattern optimization solution more targeted by 40%.

[0123] As a preferred embodiment, the solution of this application is specifically implemented as follows:

[0124] Extract the clothing mesh deformation rate for keyframes in the physics simulation. Specifically, set the keyframe sampling interval in the physics engine, for example, extracting clothing mesh deformation data every 10 frames. For each keyframe, calculate the displacement vectors of the clothing mesh vertices and, based on these displacement vectors, calculate the deformation rate of each mesh cell.

[0125] Perform a matrix difference operation on the real-time joint rotation angles and the simulated joint rotation angles to generate a rotation error matrix. For example, the joint rotation angles can be described using quaternions. The real-time joint rotation quaternions are then differentiated with the joint rotation quaternions of the corresponding frame in the simulation to obtain a matrix representing the rotation error.

[0126] The deviation gradient between the deformation rate and the pressure threshold is calculated based on the rotational deviation matrix. Specifically, a tensor product operation is performed on the rotational deviation matrix and the clothing mesh deformation rate to obtain a gradient matrix that reflects the impact of motion deviation on clothing deformation. Furthermore, the gradient matrix is ​​compared with a preset pressure threshold to quantify the degree of deviation between the deformation rate and the pressure threshold.

[0127] Set multiple deformation rate thresholds, including a visual distortion warning threshold and a physical deformation threshold. For example, a deformation rate of 0.05 can be set as the visual distortion warning threshold, and a deformation rate of 0.1 as the physical deformation threshold. When the detected deformation rate exceeds the visual distortion warning threshold, the system issues a visual anomaly warning; when the deformation rate exceeds the physical deformation threshold, the system determines that the material's mechanical properties have failed.

[0128] Generate a timestamp-labeled fit deviation report. Specifically, the deformation rate deviation data for each keyframe can be associated with the corresponding timestamp to form a time-series fit deviation dataset. Furthermore, the dataset can be visualized as a curve graph, with time on the horizontal axis and deformation rate deviation on the vertical axis, to intuitively show the change trend of fit over time.

[0129] The present application further proposes a technical solution comprising the following steps: analyzing the curvature deviation of the underarm triangle piece of the garment; adjusting the fabric Poisson's ratio parameter based on the fabric stretch in the knee flexion area; optimizing the piece contour line through a B-spline curve reconstruction algorithm; injecting curvature redundancy into the knee bend area, and superimposing warp elastic compensation allowance on the side seams; converting the optimized pattern parameters into JSON structured data and outputting it.

[0130] The analysis of the curvature deviation of the panel is performed by fusing 3D scanning data with motion capture data. A curvature deviation threshold can be set between 0.05 and 0.15 radians. Reconstruction is triggered when the deviation exceeds the threshold. The fabric Poisson's ratio parameter is adjusted based on the relationship between knee flexion angle and fabric stretch. For example, when the knee flexes to 90 degrees, the Poisson's ratio parameter can be adjusted to a range of 0.35-0.45. The B-spline curve reconstruction algorithm uses cubic spline interpolation with 8-12 control points to ensure curve continuity while allowing for local curvature adjustment. The amount of curvature margin injected into the knee bend is determined based on the range of human motion, for example, a 3-5 mm deformation margin is reserved at maximum knee flexion. Elastic compensation for the side seam warp yarns is calculated using a correlation model between the yarn elastic modulus and motion trajectory, with compensation controlled within a 2%-5% stretch range. JSON data conversion uses a key-value pair structure to store pattern parameters, including fields such as curvature parameters, elastic compensation, and Poisson's ratio, to ensure data compatibility with CAD systems.

[0131] Specifically, in dynamic motion scenarios, the system first accurately calculates the curvature deviation of the underarm triangle during shoulder abduction by fusing 3D scanning with inertial sensor data. When the curvature deviation exceeds a preset threshold, a B-spline curve reconstruction algorithm is triggered to refit the panel contour by adjusting the control point positions. For example, the density of control points in the axillary transition area is increased to improve curvature adaptation accuracy. Simultaneously, the fabric Poisson's ratio parameters are dynamically adjusted based on the knee joint motion trajectory data. For example, the Poisson's ratio is reduced as the flexion angle increases, reducing weft shrinkage during warp stretching. During the panel reconstruction process, the curvature redundancy injected into the knee bend is calculated using an inverse kinematics model to ensure sufficient deformation space. The warp elastic compensation allowance for the side seams is determined by a function correlating the yarn elastic modulus with motion speed, for example, increasing the compensation coefficient during rapid motion. The resulting JSON data contains the reconstructed panel parameters, material property corrections, and compensation data, and can be directly imported into garment CAD systems for automatic pattern correction. This technical solution effectively eliminates local stress concentration during movement through the synergistic effect of dynamic parameter adjustment and geometric reconstruction, allowing the garment to maintain a uniform fit during dynamic deformation.

[0132] As a preferred embodiment, the solution of this application is specifically implemented as follows:

[0133] First, the curvature deviation of the garment's underarm triangle is analyzed. Dynamic deformation data of the garment as it is worn is acquired through 3D scanning, and the surface equation of the underarm region is extracted. This surface equation is then compared with the underarm surface of a standard human model to calculate the curvature deviation. For example, Gaussian curvature and mean curvature can be used as evaluation metrics, respectively, to calculate the curvature difference between the two surfaces at corresponding points.

[0134] Secondly, the fabric Poisson's ratio parameter is adjusted based on the fabric stretch in the knee flexion area. As the user flexes the knee, strain data is collected from the fabric in the knee area. The fabric stretch ratios in the warp and weft directions are calculated from this strain data, and the dynamic Poisson's ratio is derived. This dynamic Poisson's ratio is compared with the fabric's original Poisson's ratio to determine the adjustment amount for the Poisson's ratio parameter.

[0135] Next, the contour of the cut piece is optimized using a B-spline curve reconstruction algorithm. The cut piece's contour is discretized into a series of control points, which are then fitted using a B-spline curve interpolation algorithm. By adjusting the positions and weights of the control points, the curve's shape is optimized to better fit the body's dynamic contours. An iterative optimization method can be used to continuously adjust the control points until the curve shape meets the preset fit criteria.

[0136] Next, curvature redundancy is added to the knee area, and warp elastic compensation is added to the side seams. Based on the knee flexion data obtained in the previous steps, the required curvature redundancy is calculated. This redundancy is added to the knee area of ​​the panel by increasing the fabric area. Simultaneously, the stress distribution of the side seams during human movement is analyzed to calculate the required warp elastic compensation. Appropriate fabric allowance is added to the side seam panel design to provide additional elastic support.

[0137] Finally, the optimized pattern parameters are converted into JSON structured data and output. The optimized parameters obtained in the above steps, including the panel curvature, fabric Poisson's ratio, panel outline control point coordinates, curvature redundancy, and elastic compensation allowance, are integrated into a JSON data structure. This JSON data can contain multiple nested structures to clearly represent the relationships and hierarchies between the various parameters.

[0138] This application further proposes to analyze the curvature deviation of the underarm triangle panel of the garment; adjust the fabric Poisson's ratio parameter based on the fabric stretch in the knee joint flexion area; optimize the contour line of the panel through the B-spline curve reconstruction algorithm; inject curvature redundancy into the knee bend area, and superimpose warp yarn elastic compensation allowance on the side seams; convert the optimized pattern parameters into JSON structured data and output it.

[0139] Among them, the analysis of the curvature deviation of the cutting piece can obtain the point cloud data of the armpit area through the 3D scanning equipment, and calculate the surface curvature change by combining the finite element stress analysis tool. For example, the curvature radius deviation threshold range is ±3mm to ±5mm; the adjustment of the fabric Poisson's ratio parameter is based on the linear relationship between the knee flexion angle and the fabric stretching amount. When the knee flexion angle reaches 90 degrees, the corresponding stretching compensation coefficient can be set to 1.2 to 1.5; the B-spline curve reconstruction algorithm uses cubic uniform B-spline, and the number of control points is dynamically adjusted according to the complexity of the cutting piece contour. For example, the number of control points in the knee bend area is set to The curvature redundancy is injected by reverse engineering to calculate the fabric deformation at a preset knee flexion angle. For example, 2%-5% arc length redundancy is added to the back piece of the knee bend area. The superposition of warp elastic compensation allowance is based on the matching relationship between the elastic modulus of the warp yarn and the tensile stress of the side seam. The compensation allowance can be set to 1.1 to 1.3 times the original warp length. The conversion of JSON structured data uses key-value pairs to store pattern parameters. For example, the curvature parameter uses "curvature_compensation" as the key, and the numerical precision is retained to two decimal places.

[0140] Specifically, after the dynamic pressure map is generated, the fabric stretching data in the knee flexion area is extracted and combined with the finite element analysis results to calculate the deviation between the current fabric Poisson's ratio and the ideal state. When the knee joint bending angle is detected to be more than 60 degrees, the Poisson's ratio parameter adjustment module is automatically triggered to increase the lateral shrinkage rate of the fabric to the preset threshold value. The curvature deviation of the underarm triangle piece is obtained by comparing the curvature matching degree between the armpit surface of the three-dimensional human body model and the clothing piece, and the least squares method is used to fit the optimal curvature compensation curve. During the B-spline curve reconstruction process, control points are inserted on the piece contour line according to the coordinates of the high-pressure area marked in the dynamic pressure map and the node vector is adjusted so that the reconstructed The curvature change rate of the curve is consistent with the human body movement trajectory; the injection of curvature redundancy in the knee bend area is achieved by inversely solving the maximum tensile deformation during knee flexion, and increasing the arc length of the back crotch line in the cutting piece CAD file. For example, a 3mm redundancy is added to the knee position of the back piece of pants; the superposition of the elastic compensation allowance of the side seam warp yarn is based on the correlation analysis of the elastic modulus of the warp yarn and the real-time motion data. When the tensile stress of the side seam line is detected to exceed 20N, the warp yarn length compensation coefficient is automatically increased; finally, the pattern data containing curvature parameters, Poisson's ratio adjustment value and elastic compensation amount is converted into JSON format and transmitted to the clothing production system through a standardized interface to realize the automatic correction of cutting piece parameters.

[0141] As a preferred embodiment, the solution of this application is specifically implemented as follows:

[0142] First, the curvature deviation of the garment's underarm triangle panel is analyzed. 3D scanning is used to obtain point cloud data of the underarm area of ​​the garment. A surface fitting algorithm is then used to calculate the actual curvature distribution of the panel. The actual curvature is compared with the ideal curvature to determine the curvature deviation.

[0143] Secondly, the fabric Poisson's ratio parameters are adjusted based on the fabric stretch in the knee flexion area. A high-speed camera system captures the fabric deformation process as the user flexes the knee. Image processing techniques are used to extract fabric stretch data and establish a relationship model between stretch and Poisson's ratio. Based on this model, the fabric Poisson's ratio parameters are dynamically adjusted.

[0144] Furthermore, the B-spline curve reconstruction algorithm is used to optimize the contour of the cutting piece. The original cutting piece contour is discretized into a sequence of control points, and the contour is reconstructed using the B-spline curve interpolation method. By adjusting the positions of the control points, the cutting piece shape is optimized to improve its fit with the human body surface.

[0145] Next, curvature margin is added to the knee area, and warp elastic compensation is added to the side seams. The required curvature margin is calculated based on the knee joint's range of motion. In the CAD file for the cut piece, the preset curvature value is added to the corresponding position. At the same time, warp elastic compensation is added to the side seams to improve the garment's longitudinal stretch.

[0146] Finally, the optimized pattern parameters are converted into JSON structured data and output. Modified piece outlines, fabric parameters, and other information are integrated into a standardized JSON format. A JSON file containing the optimized pattern data is generated, making it easier for subsequent production systems to read and process it.

[0147] The present application further proposes a density gradient distribution scheme, including: deploying a ring-shaped sensor array in the armpit area of ​​a color three-dimensional human body model, with each sensor distributed at a preset spacing; using a diamond-shaped sensor array in the chest and back area, with each sensor distributed at a first preset spacing; using a spiral sensor array in the limb area, with each sensor distributed at a second preset spacing; dynamically encrypting the spacing between adjacent sensors based on the density of the seam lines of the clothing; and adjusting the sensor sampling frequency based on real-time pressure feedback.

[0148] Among them, the preset spacing of the annular sensor array can be set to 3-5 mm, and its annular arrangement can fit the curvature change of the armpit surface and avoid uneven spacing caused by sudden changes in the surface. The first preset spacing of the diamond sensor array can be set to 5-8 mm, using a diamond symmetrical structure to cover the flat area of ​​the chest and back region, and the error in the spacing between the vertices of adjacent diamond units is controlled within ±0.2 mm. The second preset spacing of the spiral sensor array can be set to increase along the axial spacing of the limb, for example, gradually increasing from 2 mm at the proximal end to 6 mm at the distal end, to adapt to the circumferential contraction and deformation of the limb. During the dynamic encryption process, for every 10% increase in the suture line density, the spacing between adjacent sensors is compressed by a geometric coefficient of 0.8 times to form a locally encrypted grid. The sampling frequency adjustment can be set to increase the sampling frequency by 10 Hz for every 1 N / s increase in the pressure change rate, and the maximum does not exceed 200 Hz.

[0149] Specifically, when processing the axilla, the sensors in the circular array are arranged circumferentially along the underarm surface. The preset spacing is dynamically adjusted based on the radius of curvature. For example, a 3mm spacing is used when the radius is less than 50mm, and a 5mm spacing is used when the radius is greater than 50mm. The diamond array in the thoracic region uses the intersection of the diamond diagonals as the sensor positioning reference. The first preset spacing is scaled proportionally based on chest circumference. For example, a 6mm spacing is used for a 90cm chest circumference. The spiral array in the limb region is arranged along the limb axis at a spiral angle of 15-30 degrees. The second preset spacing is increased to 2mm at the elbow bend and increased to 6mm at the straight arm. When detecting garment seams, the spacing between sensors on either side of the seam is automatically reduced based on the seam density. For example, if the seam density is 5 stitches per centimeter, the spacing is reduced to 60% of the original value. During real-time pressure feedback, if a pressure gradient exceeding 5N / mm² is detected, the sampling frequency is increased from the default 50Hz to 150Hz and then restored to the base frequency after the pressure stabilizes. Therefore, through differentiated array layout and dynamic adjustment mechanism, the system resource allocation is optimized while ensuring data acquisition accuracy, and the coverage integrity problem of pressure detection on complex human body curves is solved.

[0150] As a preferred embodiment, the solution of the present application is specifically implemented as follows: a ring sensor array is arranged in the armpit area of ​​the color three-dimensional human body model, the sensors are distributed circumferentially along the armpit surface, the preset spacing is set to five millimeters, and the center point of the ring array coincides with the armpit vertex; a diamond sensor array is used in the chest and back area, the sensor nodes are distributed along the diagonal direction of the diamond, the first preset spacing is set to ten millimeters, and the diamond vertex is aligned with the center point of the shoulder blade; a spiral sensor array is arranged in the limb area, the sensors are distributed in a spiral along the longitudinal axis of the limb, the second preset spacing is set to eight millimeters, and the spiral spacing is adaptively adjusted with the change of the limb circumference; within five millimeters on both sides of the clothing seam line, the spacing between adjacent sensors is dynamically encrypted to three millimeters, and the encryption mode is activated when the direction of the seam line is orthogonal to the direction of the sensor array; when the real-time pressure value exceeds fifty kilopascals, the sampling frequency is increased to one hundred times per second, and when the pressure value is lower than ten kilopascals, the sampling frequency is reduced to ten times per second, and the pressure change rate threshold is set to five kilopascals per second to trigger frequency adjustment.

[0151] This application further proposes the generation of a vertex normal vector offset map, including: calculating the stress distribution within the triangular mesh surface through thin plate theory; mapping the stress scalar value to the continuous gradient interval of the HSV hue ring; using UV parameterization to unfold the three-dimensional stress distribution surface; marking the spatial correlation between the high-pressure area and the clothing seam line in the two-dimensional heat map; and generating a clothing pattern optimization weight coefficient matrix based on the two-dimensional heat map.

[0152] The calculation of in-plane stress distribution in a triangular mesh can be achieved by combining the bending stiffness matrix from thin plate theory with the in-plane strain tensor. For example, the triangular mesh elements can be treated as anisotropic thin plate elements, and the in-plane stress components can be calculated by multiplying the stiffness matrix with the displacement vector. The stress scalar value mapping can be performed using a continuous gradient from blue to red on the HSV hue circle, where the hue changes linearly with the stress value, the saturation is fixed at 100%, and the brightness is dynamically adjusted based on the mesh curvature. UV parametric unfolding can use a conformal mapping algorithm to parameterize 3D mesh vertices into 2D plane coordinates while maintaining the area ratio of adjacent triangles. Spatial correlation between high-pressure areas and seams can be marked by overlaying and comparing the thermal image layer with the CAD layer of the garment pattern. For example, areas with sudden changes in pressure gradient can be detected within 5 mm on either side of the seam. The weight coefficient matrix can be generated by multiplying the pressure value by the distance to the seam. The distance parameter is calculated as the Euclidean distance between the thermal image pixel coordinates and the pattern boundary.

[0153] Specifically, when simulating the draping state of a garment in a physics engine, the displacement vectors of the triangular mesh vertices are calculated using thin plate theory. Each triangular facet is modeled as a thin plate element with bending stiffness, and the in-plane stress distribution is determined by solving equilibrium equations. The calculated stress scalar values ​​are normalized to the range 0-1 and mapped to the blue-to-red gradient of the HSV color wheel, forming a three-dimensional stress distribution surface. This surface is then parametrically unfolded into a two-dimensional image using UV parameters, preserving the stress distribution characteristics of the original mesh vertices. The unfolded 2D image is spatially overlaid with a CAD drawing of the garment's seams. The relative positions of areas with pressure values ​​exceeding a set threshold and the seams are marked in a heat map. Based on the marking results, areas with pressure values ​​exceeding the threshold and a distance from the seams less than a preset value are assigned higher weights, forming a pattern optimization weight matrix. This matrix guides the prioritization of pattern contour adjustments during pattern optimization, such as increasing curvature compensation at the edges of the pattern corresponding to high-pressure areas and adjusting yarn elastic parameters at seam intersections. Through the above process, the calculation accuracy of pressure distribution is improved to the millimeter-level error range, the accuracy of identifying the spatial correlation between high-pressure areas and clothing structure lines is increased to more than 95%, and the reliability of the pattern optimization basis is effectively verified.

[0154] As a preferred embodiment, the solution of this application is specifically implemented as follows: Based on thin plate theory, the in-plane stress distribution of triangular facets at the vertices of a garment mesh is calculated. The surface deformation energy density is solved using a second-order partial differential equation, and the stress tensor components are inferred from the mesh vertex displacements. The calculated stress scalar values ​​are divided into 256 gradients according to the pressure intensity range, corresponding to the continuous hue of blue, green, yellow, orange, and red on the HSV color wheel. Zero stress is mapped to blue at 180 degrees on the color wheel, and maximum stress is mapped to red at 0 degrees on the color wheel. A conformal mapping algorithm is used to perform a UV parametric unfolding of the three-dimensional garment stress distribution surface. The surface is mapped to a two-dimensional plane using an angle-invariant conformal transformation to generate a UV unfolding image. A garment seam vector layer is overlaid on the unfolded two-dimensional heat map. A spatial registration algorithm is used to align the seam coordinates with the high-pressure area coordinates. A coordinate coincidence detection algorithm is used to identify the spatial overlap between the high-pressure area and the seam. An optimized weight coefficient matrix is ​​generated based on the area ratio and pressure intensity value of the overlapping area, where each matrix element corresponds to the pattern adjustment priority of the garment piece, and the weight coefficient is constrained to the range of 0 to 1 through matrix normalization.

[0155] This application further proposes a method for setting curvature redundancy, including measuring the tensile deformation of the back piece of the pants when the knee is flexed at a preset angle, inversely calculating the curvature compensation value of the back crotch line based on the tensile deformation, establishing a dynamic pattern parameterized template in the cutting piece CAD file, injecting curvature redundancy into the knee bend area, and superimposing warp elastic compensation allowance on the side seams.

[0156] The knee flexion angle is measured using a motion capture system to capture the human knee flexion trajectory. For example, a flexion range of 30 to 120 degrees is set, and the fabric stretch deformation at the corresponding angles is recorded using a 3D scanner. The back crotch curvature compensation value is determined using a reverse engineering algorithm. The measured fabric stretch is input into the finite element model, and the calculated curvature compensation value increment ranges from 5% to 15%. The dynamic pattern parameter template is created using a parameter-driven module in CAD software. For example, the curvature compensation value is linked to the coordinate offset of the panel control points to form an automatically adjustable template file. Curvature redundancy in the knee area is injected by adjusting the Bezier curve control points at the panel edge, specifically increasing the knee curvature radius by 8% to 12%. Elastic compensation for the warp yarns at the side seams is achieved by adding fabric allowance in the warp direction. For example, a 3-5 mm wide elastic allowance band with an elastic modulus set to 70% to 85% of the standard value is applied to the side seams.

[0157] Specifically, a motion capture system was used to capture the stretch deformation of the back panel at a 90-degree knee flexion angle, measuring a lateral stretch of 12%. Based on this data, a reverse engineering algorithm was used to calculate the curvature compensation value for the back crotch line, determining that the curvature radius needed to be increased by 10%. A parametric template was created in CAD software, linking the curvature compensation value to the coordinate parameters of the back crotch line control points, enabling dynamic adjustment. The Bezier curve control points of the knee-area panels were offset outward by 2.5 mm to create a curvature margin. A 4 mm wide elastic allowance band, made of stretch fabric with a warp elastic modulus of 80% of the standard value, was placed at the side seams. When the knee flexes, the curvature margin absorbs 12% of the fabric's stretch, while the elastic band provides an additional 3 mm of expansion through the warp elastic deformation. This structural combination increases the deformation tolerance of the knee-area panels to 15%, effectively eliminating localized tightness during dynamic movement.

[0158] As a preferred embodiment, the solution of the present application is specifically implemented as follows: During the production of the back piece of trousers, a three-dimensional optical scanner is first used to obtain the fabric tensile deformation data of the back crotch seam area when the knee is flexed forty-five degrees, wherein the flexion angle is precisely controlled by a knee joint simulation device driven by a stepper motor. Based on the obtained tensile deformation data, a curvature compensation value calculation model is established using a reverse engineering algorithm, and the increase in the curvature radius of the back crotch line is determined through iterative optimization. The piece vector diagram is imported into the CAD system, and a dynamic pattern template library containing curvature compensation parameters is established, wherein the dynamic pattern template supports parameterized drive adjustment. During the reconstruction of the contour line of the knee bend area piece, a B-spline curve algorithm is applied to convert the calculated curvature compensation value into a geometric correction value of the outer contour line of the piece, while retaining the structural constraints of the original pattern. In the side seam processing link, a composite fabric with a higher elastic modulus in the warp direction than in the weft direction is used, and an elastic allowance band with a width of three millimeters is set longitudinally along the side seam. The allowance band forms a gradient transition connection with the main piece through a hot pressing process.

[0159] This technical solution effectively addresses the tension caused by the mismatch between fabric deformation and pattern design during dynamic knee movement. By quantifying the mapping relationship between tensile deformation and curvature compensation, the panel design precisely adapts to the knee's range of motion, eliminating tensile stress concentration in the rear crotch area during movement. The inclusion of an elastic allowance significantly improves the side seam's ability to absorb complex deformation, increasing knee freedom of movement by over 40 percent while maintaining the overall silhouette stability and effectively preventing wrinkles.

[0160] The above description is merely an embodiment of the present application and is not intended to limit the scope of protection of the present application. For those skilled in the art, various modifications and variations of the present application are possible. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A virtual fitting method based on a color three-dimensional model, characterized in that: The following steps are involved: Step 1: Obtain user body point cloud data through a 3D scanning device, build a color 3D human body model based on the point cloud data, and set bone binding parameters; Step 2: Based on the surface curvature distribution of the color three-dimensional human body model, divide the pressure sensitive area and generate a dynamic pressure map; Step 3: Based on the garment draping dynamics parameters, perform cloth physics simulation on the colored three-dimensional human body model in a physics engine to extract deformation response data of garment mesh vertices; Step 4: collecting real-time motion posture data of the user through an inertial sensor, performing motion distortion correction on the motion posture data, and generating a joint motion trajectory; Step 5: Dynamically compare the joint motion trajectory with the deformation response data to generate a garment fit deviation report; Step 6: Optimize clothing pattern parameters and fabric physical properties based on the clothing fit deviation report, and output a clothing fit recommendation.

2. The virtual fitting method according to claim 1, wherein: The step 1 comprises: Step 1.1: Obtain the user's full body point cloud data through multi-angle depth scanning; Step 1.2: constructing a human body surface model based on the point cloud data and extracting key anatomical landmarks; Step 1.3, mapping the skin layer compression modulus parameter from the biomechanical property library to the human body surface model; Step 1.4, setting the rotational freedom constraint conditions of the main joint points based on the key anatomical landmarks; Step 1.5: Verify the displacement range of the main joint points through the posture interpolation algorithm to generate model deformation tolerance data.

3. The virtual fitting method according to claim 1, wherein: The step 2 includes: Step 2.1: using a finite element stress analysis tool to identify high-friction areas of the color 3D human body model, wherein the high-friction areas include the axillary curved surface and the knee flexion surface; Step 2.2, setting a density gradient distribution scheme of the pressure sensor based on the contour of the high friction area; Step 2.3: Verify whether the pressure sensor covers the seam area of ​​the garment, and complete the sensor deployment in the uncovered area; Step 2.4: discretize the pressure sensitive area into a set of triangular facets using a triangular mesh partitioning algorithm; Step 2.5: Dynamically adjust the sensor density based on the curvature change of the triangular facet set to generate a dynamic pressure map.

4. The virtual fitting method according to claim 1, wherein: The step 3 comprises: Step 3.

1. Load the garment pattern vector diagram into the physics engine and input the warp yarn elastic modulus and weft yarn bending stiffness parameters. Step 3.2, simulating the draping state of the clothing on the surface of the colored three-dimensional human body model based on the physical engine, and calculating the displacement vectors of the mesh vertices; Step 3.3, generating a vertex normal vector offset map based on the displacement vector; Step 3.4, mapping the stress scalar value in the vertex normal vector offset map to the HSV color wheel, where blue represents low pressure areas and red represents high pressure areas; Step 3.5: Project the three-dimensional stress distribution into a two-dimensional thermal map using UV unfolding technology.

5. The virtual fitting method according to claim 1, wherein: The step 4 comprises: Step 4.1, collect six-degree-of-freedom motion data of joint points through a multi-axis inertial sensor group; Step 4.2: Using a Kalman filter to eliminate drift errors of the six-degree-of-freedom motion data; Step 4.3: Calculate the joint rotation angle based on the kinematic chain model; Step 4.4: Construct a time series dataset of joint motion trajectories; Step 4.5: Fill in the missing frames in the time series dataset using an interpolation algorithm.

6. The virtual fitting method according to claim 1, wherein: The step 5 comprises: Step 5.1, extracting the clothing mesh deformation rate of the key frame in the physical simulation; Step 5.2: Perform matrix difference operation on the real-time joint rotation angle and the simulated joint rotation angle to generate a rotation deviation matrix; Step 5.3: Calculate the deviation gradient of the deformation rate and the pressure threshold based on the rotation deviation matrix; Step 5.4: Set multiple deformation rate thresholds, including a visual distortion warning threshold and a physical deformation exceeding threshold. Step 5.5: Generate a fit deviation report with a timestamp.

7. The virtual fitting method according to claim 1, wherein: The step 6 comprises: Step 6.1, analyzing the curvature deviation of the garment underarm triangle panel; Step 6.2: Adjust the fabric Poisson's ratio parameter based on the fabric stretch in the knee flexion area. Step 6.3, optimize the contour line of the cutting piece by using the B-spline curve reconstruction algorithm; Step 6.4: Inject curvature redundancy into the knee bend area and superimpose warp elastic compensation allowance on the side seams; Step 6.5: Convert the optimized template parameters into JSON structured data and output it.

8. The virtual fitting method according to claim 3, wherein: The density gradient distribution scheme includes: Step 8.1, deploying a ring sensor array in the axillary area of ​​the color 3D human body model, with each sensor being distributed at a preset interval; Step 8.2: Using a diamond-shaped sensor array on the dorsal region, with each sensor being distributed at a first preset spacing; Step 8.3: Using a spiral sensor array in the limb area, with each sensor distributed at a second preset spacing; Step 8.4: Dynamically increase the distance between adjacent sensors based on the density of clothing seams. Step 8.5: Adjust the sensor sampling frequency based on real-time pressure feedback.

9. The virtual fitting method according to claim 4, wherein: The generation of the vertex normal vector offset map includes: Step 9.

1. Calculate the stress distribution within the triangular mesh using thin plate theory. Step 9.2: Map the stress scalar value to the continuous gradient interval of the HSV color circle; Step 9.3, use UV parameterization to unfold the three-dimensional stress distribution surface; Step 9.4: Mark the spatial correlation between the high-pressure area and the seam line of the garment in the two-dimensional heat map; Step 9.5: Generate a clothing pattern optimization weight coefficient matrix based on the two-dimensional heat map.

10. The virtual fitting method according to claim 7, wherein: The setting of the curvature redundancy includes: Step 10.1, measuring the tensile deformation of the rear panel of the trousers when the knee is flexed at a preset angle; Step 10.2, inversely calculating the back crotch line curvature compensation value based on the stretching deformation; Step 10.3: Create a dynamic pattern parameter template in the cutting piece CAD file; Step 10.4, injecting curvature redundancy into the knee bend area; Step 10.5: Add warp elastic compensation allowance to the side seams.

Images