Clothing design system and method based on intelligent interaction

Abstract

The invention discloses a costume design system and method based on intelligent interaction, and belongs to the field of costume design, and the method specifically comprises the steps: collecting the human body form and motion data of a user through 3D scanning, a wearable sensor and a motion capture technology, and constructing a dynamic model; based on sports biomechanical analysis, four key parts of costume design are identified, and activity requirements and stress conditions of the four key parts in different sports states are analyzed; costume design is automatically optimized through a machine learning algorithm, and fabric recommendation is generated; according to the fabric recommendation, determining a final fabric and garment structure, and testing garment performance by using a virtual simulation technology; real-time fitting and dynamic adjustment are achieved through the augmented reality technology, and it is ensured that the design scheme meets the personalized requirements of the user and the exercise function requirements.

Description

Technical Field

[0001] The present invention belongs to the field of clothing design, and specifically relates to a clothing design system and method based on intelligent interaction. Background Art

[0002] Traditional clothing design often ignores the dynamic needs during human movement while meeting aesthetics, resulting in a sense of restraint or discomfort when wearing the clothing during activities. Moreover, it usually relies on designers' experience and manual operations, with a long design cycle, high costs, and difficulty in meeting the growing personalized needs of consumers. With the development of artificial intelligence and virtual reality technologies, there is an urgent need for an intelligent design method in the field of clothing design that can take into account human body shape and movement principles.

[0003] For example, Chinese Patent with publication number CN118260819A discloses a visual clothing design system and a clothing design method. The system includes: a body shape import module for receiving the body shape data of a target user and establishing a three-dimensional body shape model of the target user; a clothing piece import module for receiving clothing piece parameters and generating a flattened model of the clothing piece; an element configuration module for configuring clothing elements, where the clothing elements include element curling shapes, element alignment points, and element alignment point coordinates, and generating a three-dimensional model of the clothing piece based on the clothing elements and the flattened model of the clothing piece; a fusion module for receiving a movement operation instruction and fusing the three-dimensional model of the clothing piece with other clothing pieces and the three-dimensional body shape model based on the movement operation instruction to obtain a three-dimensional clothing model; an evaluation module for evaluating the color balance of the three-dimensional clothing model after receiving an evaluation instruction to obtain a reference evaluation result.

[0004] The above existing technologies have the following problems: lack of sports biomechanics analysis; more reliance on manual configuration and selection in clothing design and fabric selection, increasing the design time and costs, resulting in a low degree of automation; focusing on the display of the three-dimensional model of the clothing and the evaluation of color balance, and may lack sufficient simulation and testing of the actual performance of the clothing under different movement states; being relatively limited in terms of user feedback and dynamic adjustment, mainly relying on designers' subjective judgments and experience. Summary of the Invention

[0005] In view of the deficiencies of the prior art, the present invention proposes a clothing design system and method based on intelligent interaction, which uses 3D scanning, wearable sensors and motion capture technology to collect user's body shape and motion data, and constructs a dynamic model; based on sports biomechanics analysis, it identifies the four key parts of clothing design, and analyzes their activity requirements and force conditions under different motion states; automatically optimizes clothing design through machine learning algorithms, generates fabric recommendations; determines the final fabric and clothing structure according to the fabric recommendations, and uses virtual simulation technology to test the clothing performance; realizes real-time try-on and dynamic adjustment through augmented reality technology to ensure that the design scheme meets the user's personalized needs and sports functionality requirements.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A clothing design method based on intelligent interaction, including:

[0008] Step S1: Collect various types of data and construct a human body dynamic model of the user; the various types of data include body shape data, motion data and motion trajectory data;

[0009] Step S2: Based on the built-in sports biomechanics analysis module of the system, according to the human body dynamic model, analyze the body mechanics changes of the user under different motion states, and at the same time, identify the four parts in clothing design, and determine the activity requirements and force conditions of the four parts under different motion states;

[0010] Step S3: Through machine learning algorithms, combined with the activity requirements of the four parts, automatically optimize the clothing design, and generate an optimized clothing design scheme and a fabric recommendation list according to the results of sports biomechanics analysis;

[0011] Step S4: According to the fabric recommendation list, combined with the design requirements, determine the final fabric, and design the clothing structure according to the activity requirements of the four parts to obtain the final fabric selection scheme and clothing structure design drawing;

[0012] Step S5: Through virtual simulation technology, apply the final fabric selection scheme and clothing structure design drawing to the human body dynamic model, simulate the performance of the clothing under different motion states, the system simulates the performance of the clothing under different motion states, and conducts simulation tests;

[0013] Step S6: Through augmented reality method, try on the clothing design that passes the simulation test in real time, and simulate the motion state in the virtual environment. The system automatically adjusts the design scheme according to the user's feedback and dynamic performance.

[0014] Specifically, the specific steps of the step S1 include:

[0015] S1.1: Use 3D scanning technology to perform a full-body scan of the user to obtain the user's body shape data, and collect the user's motion data under specific actions through wearable sensors. At the same time, use an optical motion capture system to paste reflective marker points on the user's body key points to capture the motion trajectory data of the user under different motion states; the body shape data includes body type, contour, joint position, and muscle distribution; the motion data includes acceleration, angular velocity, and pressure distribution.

[0016] S1.2: Sort and match the body shape data, motion data, and motion trajectory data according to the time stamp; if the time stamps of multiple types of data are inconsistent, use the improved linear interpolation formula for alignment, where y(t) represents the interpolation result, y 1 and y 2 represent multiple types of data at adjacent time points, t 1 and t 2 represent the time stamps corresponding to y 1 and y 2 , t represents the current time, α represents the weight factor, and μ represents the smoothing parameter.

[0017] If there is a time offset between devices, calculate the time offset through the cross-correlation analysis method and perform correction; the cross-correlation analysis method is to calculate the cross-correlation function of two time series data and find the time delay τ max that maximizes the function value in the cross-correlation function. According to the found time delay τ max , adjust the time stamp of one type of data to match the time stamp of the other type of data.

[0018] Specifically, the specific steps of step S1 also include:

[0019] S1.3: According to the aligned body shape data and motion trajectory data, construct the user's bone model and use the inverse kinematics algorithm to calculate the joint angles and motion ranges.

[0020] S1.4: According to the aligned motion data and body shape data, construct the user's muscle model and skin model, simulate muscle contraction and skin extension, and use the finite element analysis method to simulate the deformation of the skin under different motion states.

[0021] S1.5: Take the bone model as the first layer model, the muscle model as the second layer model, and the skin model as the third layer model, and integrate them to form the user's human dynamic model, and use the physics engine to simulate the performance of the model under different motion states.

[0022] Specifically, the specific steps of S1.3 include:

[0023] S1.31: Define the bone structure according to human anatomy knowledge, and represent the bone hierarchy using a tree structure, including joint positions, bone lengths, and joint degrees of freedom;

[0024] S1.32: Initialize the bone model parameters based on the aligned human body shape data, and fit the bone model using the principal component analysis method;

[0025] S1.33: Define the target positions of the end effectors according to the aligned motion trajectory data; the end effectors include hands and feet;

[0026] S1.34: Use the inverse kinematics algorithm to calculate the joint angles θ = IK(p target , W, C, E, D), where θ represents the joint angles, p target represents the target position, W represents the weight vector, and W = [w 1 ,..., w n , w n represents the weight of the nth joint, n represents the number of joints, C represents the constraint condition matrix with n rows and 2 columns, and each row represents the minimum and maximum angles of a joint [θ min , θ max , E represents the error tolerance vector, and E = [e 1 , e 2 , e 3 , e 1 , e 2 , e 3 respectively represent the error tolerances in the horizontal, vertical, and vertical directions, D represents the damping coefficient vector, and IK(·) represents the inverse kinematics function;

[0027] S1.35: Define key frames based on the calculated joint angles, and generate a preliminary bone animation based on the key frames to simulate the user's performance in different motion states;

[0028] S1.36: Optimize the parameters of the bone model according to the aligned motion trajectory data and the preliminarily generated bone animation, and use the optimization algorithm to minimize the error between the bone model and the motion trajectory data;

[0029] If the error is greater than or equal to the preset error threshold, adjust the bone model parameters or the inverse kinematics algorithm;

[0030] S1.37: Generate the final bone animation based on the adjusted bone model.

[0031] Specifically, the specific steps of S1.35 include:

[0032] A1: Obtain the skeletal model from steps S1.31 and S1.32, including joint positions, bone lengths, and hierarchical relationships, and obtain the calculated joint angle data from step S1.34;

[0033] A2: Define key frames based on the joint angle data;

[0034] A3: Use the linear interpolation algorithm to generate intermediate frames between key frames, and use the formula to generate the interpolated joint angles, where σ represents the interpolation coefficient, q 1 and q 2 represent the quaternions of the key frames, and q(σ) represents the interpolated joint angles;

[0035] A4: Update the rotation matrix R(θ) of each joint in the skeletal model according to the interpolated joint angles, and update the global coordinates of the joints using the rotation matrix. The formula is: P global = M parent ×R(θ)×P local , where P global represents the global coordinates of the child node, M parent represents the transformation matrix of the parent node, and P local represents the local coordinates of the child node;

[0036] A5: Bind the skeletal model to the human body shape data, and use the formula to calculate the final position Y of each vertex final , where, represents the weight, n represents the number of vertices, M i represents the bone transformation matrix, and Y o represents the original position of the vertex;

[0037] A6: Use the graphics rendering engine to render the skeletal animation frame by frame, and render the skeletal pose of each frame onto the screen to generate an animation effect.

[0038] Specifically, the specific steps of S1.4 include:

[0039] S1.41: Obtain the aligned motion data and human body shape data, and extract key points and regions; the key points include joint positions and muscle attachment points; the regions refer to muscle groups and skin regions;

[0040] S1.42: Define the path of each muscle according to anatomical knowledge, and represent the muscle path using a B-spline curve;

[0041] S1.43: Initialize the muscle parameters according to the motion data and human body shape data, and the muscle parameters include length, cross-sectional area, and contraction force;

[0042] S1.44: Calculate the length change and contraction force of the muscle according to the joint angle and muscle path, and simulate the mechanical behavior of the muscle using a muscle model;

[0043] S1.45: Construct a surface mesh model of the skin based on the human body morphology data, and generate a skin mesh using a mesh generation algorithm;

[0044] S1.46: Initialize the skin parameters according to the biomechanical properties of the skin, and define the boundary conditions of the skin according to the muscle contraction force and joint movement;

[0045] S1.47: Discretize the skin mesh into finite elements, and based on the finite elements, establish a finite element equation according to the theory of elasticity, and solve the equation;

[0046] S1.48: Update the vertex positions of the skin mesh according to the solution results to generate a deformed skin model.

[0047] Specifically, the specific steps of step S2 include:

[0048] S2.1: Obtain multiple types of data and transmit the multiple types of data to the built-in motion biomechanics analysis module of the system;

[0049] S2.2: Adapt the multiple types of data to the human body dynamic model using a coordinate transformation algorithm;

[0050] S2.3: Analyze the human body dynamic model using an object detection algorithm based on deep learning, identify the four major parts in the clothing design, and extract the data of the four major parts during the movement process; the four major parts include the shoulder, elbow, knee, and waist; the movement process data includes position and morphological changes;

[0051] S2.4: Perform a mechanical analysis on the identified four major parts according to the principles of kinematics and dynamics. The mechanical analysis includes calculating the displacement, velocity, acceleration, force, and moment received by each part during the movement process;

[0052] S2.5: Determine the activity requirements and force conditions of the four major parts in different motion states according to the mechanical analysis results and combined with the characteristics of different sports events.

[0053] Specifically, the specific steps of step S3 include:

[0054] S3.1: Extract the activity requirements of the four major parts from the motion biomechanics analysis results, and quantify the activity requirements into feature vectors to obtain activity requirement feature vectors; the activity requirements include activity range, force distribution, and ductility requirements;

[0055] S3.2: Load the pre-trained machine learning model, which is configured based on the random forest algorithm;

[0056] S3.3: Input the activity requirement feature vector into the pre-trained machine learning model to generate a clothing design solution that meets the optimization goal, and output the clothing design parameters at the same time;

[0057] S3.4: Extract the fabric features from the fabric database and match the fabric features with the clothing design parameters;

[0058] S3.5: Generate a fabric recommendation list according to the matching result and sort it according to the priority.

[0059] The clothing design system based on intelligent interaction includes: a data acquisition module, a sports biomechanics analysis module, a clothing design optimization module, a fabric and structure design module, a virtual simulation test module, and an augmented reality module;

[0060] The data acquisition module is used to collect various types of data of the user;

[0061] The sports biomechanics analysis module is used to analyze the body mechanics changes of the user in different sports states according to the collected various types of data, and identify the key parts of clothing design and their activity requirements and force conditions;

[0062] The clothing design optimization module is used to automatically optimize the clothing design through machine learning algorithms, combine the activity requirements of the four major parts, and generate fabric recommendations;

[0063] The fabric and structure design module is used to determine the final fabric and design the clothing structure according to the fabric recommendation list and design requirements;

[0064] The virtual simulation test module is used to simulate the performance of the clothing in different sports states through virtual simulation technology and conduct tests;

[0065] The augmented reality module is used to try on the clothing design that passes the simulation test in real time through the augmented reality method, and automatically adjust the design scheme according to the user feedback and dynamic performance.

[0066] Compared with the prior art, the beneficial effects of the present invention are:

[0067] 1. The present invention proposes a clothing design method based on intelligent interaction. By comprehensively applying 3D scanning, motion capture and biomechanics analysis technologies, it accurately captures the human body shape and motion characteristics of the user, realizes highly personalized clothing design, not only improves the comfort and functionality of the clothing, but also ensures the fitting degree and expressiveness of the clothing in different sports states, meeting the personalized needs of users.

[0068] 2. The present invention proposes a clothing design method based on intelligent interaction, which uses machine learning algorithms to automatically optimize the design scheme, and combines virtual simulation and augmented reality technologies to achieve a comprehensive digital process from fabric selection to clothing structure design. This not only improves the design efficiency, reduces the trial-and-error cost, but also enables users to experience and feedback the design effect in real time in a virtual environment, thus promoting the continuous optimization and improvement of the design scheme. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 It is a schematic diagram of the clothing design method based on intelligent interaction of the present invention;

[0070] Figure 2 It is a principle flow chart of the clothing design method based on intelligent interaction of the present invention;

[0071] Figure 3 It is an architecture diagram of the clothing design system based on intelligent interaction of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0072] Embodiment 1

[0073] Please refer to Figure 1 and Figure 2 An embodiment provided by the present invention: A clothing design method based on intelligent interaction, including the following steps:

[0074] Step S1: Collect various types of data to construct a human body dynamic model of the user; the various types of data include human body shape data, motion data, and motion trajectory data;

[0075] Step S2: Based on the built-in motion biomechanics analysis module of the system, according to the human body dynamic model, analyze the body mechanics changes of the user in different motion states. At the same time, identify the four major parts in clothing design and determine the activity requirements and force conditions of the four major parts in different motion states;

[0076] Step S3: Through machine learning algorithms, combined with the activity requirements of the four major parts, automatically optimize the clothing design, and generate an optimized clothing design scheme and a fabric recommendation list according to the results of motion biomechanics analysis;

[0077] Step S4: According to the fabric recommendation list, combined with the design requirements, determine the final fabric, and design the clothing structure according to the activity requirements of the four major parts to obtain the final fabric selection scheme and clothing structure design drawing;

[0078] Furthermore, the specific steps of Step S4 include:

[0079] (1) Obtain fabric information from the fabric recommendation list generated by machine learning algorithms, including fabric type, elasticity, breathability, abrasion resistance and other characteristics. Among them, the fabric recommendation list is usually sorted by priority, and fabrics that meet the design requirements are recommended first;

[0080] (2) According to the clothing design requirements, such as comfort, freedom of movement, durability, screen the candidate fabrics in the fabric recommendation list;

[0081] (3) Match the fabric characteristics with the design requirements, and select the fabric that best meets the requirements. For example, select a high-elasticity fabric to meet the freedom of movement requirements, and select a breathable fabric to improve comfort;

[0082] Among them, the specific formula for matching fabric characteristics with design requirements is: Among them, Match j represents the matching degree of the j-th fabric, F jk represents the k-th characteristic of the j-th fabric, D k represents the k-th characteristic of the design requirement, sim(·) represents the similarity function, and m represents the design requirement characteristic category;

[0083] (4) Check the availability of the selected fabric, such as inventory, cost, production cycle. If the fabric is not available, select the second-best fabric and re-evaluate;

[0084] (5) Determine the final fabric according to the matching result and availability check, and output the final fabric selection plan;

[0085] (6) Integrate the final fabric selection plan with the clothing design plan to generate the final clothing structure design drawing, and generate production documents according to the clothing structure design drawing, such as cutting diagrams, sewing process descriptions.

[0086] Step S5: Through virtual simulation technology, apply the final fabric selection plan and clothing structure design drawing to the human body dynamic model to simulate the performance of the clothing under different movement states. The system simulates the performance of the clothing under different movement states and conducts simulation tests;

[0087] Furthermore, the specific steps of step S5 include:

[0088] (1) Obtain the human body dynamic model and motion trajectory data;

[0089] (2) Import the final fabric selection plan and clothing structure design drawing, and bind the clothing design data to the human body dynamic model;

[0090] (3) Define the movement states to be simulated, such as running, jumping, bending down;

[0091] (4) Define the simulation parameters, and set the physical parameters according to the fabric characteristics and clothing structure, such as the friction coefficient and elastic modulus;

[0092] (5) Generate the surface mesh model of the clothing according to the clothing structure design drawing, and use the mesh generation algorithm, such as Delaunay triangulation, to generate the clothing mesh. Among them, the Delaunay triangulation method is the prior art content in this field and is not the creative solution of this application, so it will not be elaborated here;

[0093] (6) Bind the clothing mesh to the human body dynamic model, use the physics engine to simulate the deformation of the clothing under different motion states, and calculate the deformation and force of the clothing according to the fabric characteristics and motion states;

[0094] (7) Run the simulation in the virtual simulation environment, record the performance of the clothing under different motion states, such as the amount of deformation and force distribution, and use the visualization tool Matplotlib to display the simulation results.

[0095] Step S6: Try on the clothing design that has passed the simulation test in real time through the augmented reality method, and simulate the motion state in the virtual environment. The system automatically adjusts the design scheme according to the user's feedback and dynamic performance.

[0096] Furthermore, the specific steps of step S6 include:

[0097] (1) Build the AR environment, and load the clothing design model and the human body dynamic model that have passed the simulation test;

[0098] (2) Obtain the human body shape data, motion data, and motion trajectory data;

[0099] (3) Bind the virtual clothing mesh to the user's human body dynamic model, update the deformation of the clothing mesh according to the user's real-time motion data, and at the same time, render it into the user's field of view through the AR device and collect the feedback data. Among them, the AR device rendering technology is the prior art content in this field and is not the creative solution of this application, so it will not be elaborated here;

[0100] (4) Quantify the feedback data, run the optimization algorithm to generate a new clothing design scheme, and update the clothing design model.

[0101] Exemplarily, a male runner with a height of 175 cm and a weight of 70 kg was selected as the test subject. Through 3D body scanning technology, his body shape data was collected, including key dimensions such as height, weight, shoulder width, chest circumference, waist circumference, and hip circumference. At the same time, a motion capture system was used to record his motion data and motion trajectory data when running at different speeds, such as step length, stride frequency, and body swing amplitude. Based on these data, an accurate human body dynamic model was constructed; the constructed human body dynamic model was imported into the built-in sports biomechanics analysis module of the system. This module can simulate the body mechanics changes of the user when running at different speeds, such as muscle contraction and joint angle changes. At the same time, the system identified four major parts in the clothing design: shoulders, chest, waist, and legs, and analyzed the activity requirements and force conditions of these parts when running at different speeds. For example, when running at high speed, the activity requirements of the legs and waist increase, requiring higher flexibility and support; while the shoulders and chest are relatively stable, but require good breathability and sweat-wicking functions;

[0102] Taking the activity requirements and force conditions of the four major parts as inputs, the clothing design was optimized through machine learning algorithms. Among them, the machine learning algorithms combined ergonomic principles, fabric performance data, and past design experience to automatically generate an optimized clothing design plan and a fabric recommendation list. For example, for the legs, the algorithm recommended fabrics with high elasticity and abrasion resistance, and for the waist, materials with good support and breathability were selected; according to the fabric recommendation list and design requirements, the final fabric was selected. Then, combined with the activity requirements of the four major parts, the clothing structure was designed. For example, tight-fitting pants were designed for the legs to improve muscle support, an elastic belt was used for the waist to ensure fit and comfort, and loose designs were adopted for the shoulders and chest to increase breathability and freedom. Finally, a fabric selection plan and a clothing structure design drawing were obtained;

[0103] The final fabric selection plan and clothing structure design drawing were imported into the virtual simulation system. The system simulated the clothing performance of the runner when running at different speeds, including the stretching, deformation, and breathability of the fabric, etc. Through the simulation test, the rationality and comfort of the clothing design were verified; finally, augmented reality technology was used to let the runner try on the clothing design that passed the simulation test in real time. In the virtual environment, the runner can simulate running states at different speeds and observe the clothing performance in real time. The system automatically adjusts the design plan according to the runner's feedback and dynamic performance. For example, if it is found that the waist design is too tight and causes discomfort, the system can automatically adjust the tightness of the belt or the fabric selection. After multiple adjustments and tests, an intelligent clothing design plan that meets the runner's needs was finally obtained.

[0104] The specific steps of step S1 include:

[0105] S1.1: Use 3D scanning technology to perform a full-body scan on the user to obtain the user's body shape data, including body type, contour, joint positions, and muscle distribution. At the same time, collect the user's motion data under specific actions, such as acceleration, angular velocity, and pressure distribution, through wearable sensors. Additionally, use the optical motion capture system OptiTrack to paste reflective marker points on the user's body key points to capture the user's motion trajectories in different motion states;

[0106] S1.2: Sort and match the body shape data, motion data, and motion trajectory data according to timestamps; if the timestamps of multiple types of data are inconsistent, use an improved linear interpolation formula for alignment, where y(t) represents the interpolation result, y 1 and y 2 represent multiple types of data at adjacent time points, t 1 and t 2 represent the timestamps corresponding to y 1 and y 2 , t represents the current time, α represents the weight factor used to adjust the weight of the interpolation result, and μ represents the smoothing parameter used to control the smoothness of the interpolation result as it changes with t;

[0107] It should be noted that, based on the original linear interpolation, the formula in the present invention hopes that the interpolation result is closer to y 1 or y 2 . Therefore, the weight factor α is introduced, and the weight of the interpolation result can be adjusted according to needs, so as to more flexibly adapt to different application scenarios. In order to maintain the smoothness of the interpolation result near the midpoint between t 1 and t 2 , the smoothing term is introduced to reduce the problem of unstable interpolation results caused by fewer data points or large data fluctuations. Therefore, the formula in the present invention not only retains the simplicity and intuitiveness of the original linear interpolation formula but also enhances its flexibility and smoothness by introducing additional parameters, making it applicable to more types of data and interpolation requirements.

[0108] If there is a time offset between devices, calculate the time offset through the cross-correlation analysis method and perform correction. The cross-correlation analysis method is to calculate the cross-correlation function of two time series data and find the time delay τ max that maximizes the function value in the cross-correlation function. According to the found time delay τ max , adjust the timestamp of one type of data to match the timestamp of the other type of data. Among them, the cross-correlation function is the prior art content in the field and is not the creative solution of this application, so it will not be elaborated here;

[0109] S1.3: Based on the aligned human body shape data and motion trajectory data, construct the user's skeletal model, including joint positions, bone lengths, and joint degrees of freedom, and use the inverse kinematics algorithm to calculate joint angles and motion ranges;

[0110] S1.4: Based on the aligned motion data and human body shape data, construct the user's muscle model and skin model, simulate muscle contraction and skin extension, and use the finite element analysis method to simulate the deformation of the skin under different motion states;

[0111] S1.5: Take the skeletal model as the first layer model, the muscle model as the second layer model, and the skin model as the third layer model, and integrate them to form the user's human body dynamic model, and use the physics engine to simulate the performance of the model under different motion states;

[0112] Among them, the physics engine simulation is to simulate the motion, deformation, and interaction of objects in the physical world through mathematical models and algorithms. In the human body dynamic model, the physics engine is used to simulate the performance of bones, muscles, and skin under different motion states. Specifically, the physics engine simulation method is the existing technical content in this field and is not the creative solution of this application, so it will not be elaborated here.

[0113] The specific steps of S1.3 include:

[0114] S1.31: According to human anatomy knowledge, define the bone structure, and use a tree structure to represent the bone hierarchy relationship, including joint positions, bone lengths, and joint degrees of freedom;

[0115] Furthermore, the specific steps of S1.31 include:

[0116] (1) According to human anatomy knowledge, determine the key joints, such as the shoulder joint, elbow joint, knee joint, and hip joint. Among them, each joint includes a position, a rotation axis, and a degree of freedom. For example, the shoulder joint has 3 degrees of freedom: flexion and extension, abduction and adduction, rotation;

[0117] (2) Use a tree structure to represent the bone hierarchy relationship, with the torso as the root node and the limbs as the child nodes;

[0118] Exemplarily, root node: torso

[0119] Child node 1: left shoulder → left upper arm → left elbow → left forearm → left hand

[0120] Child node 2: right shoulder → right upper arm → right elbow → right forearm → right hand

[0121] Child node 3: left hip → left thigh → left knee → left calf → left foot

[0122] Child node 4: right hip → right thigh → right knee → right calf → right foot.

[0123] S1.32: Initialize the parameters of the skeletal model according to the aligned human body shape data, and fit the skeletal model using the principal component analysis method. The principal component analysis method is the prior art in this field and not the creative solution of this application, so it will not be elaborated here.

[0124] Further, the specific steps of S1.32 include:

[0125] (1) Obtain the skeletal hierarchical relationship;

[0126] (2) Extract key points from the human body shape data, such as joint positions and bone endpoints, and initialize the position coordinates of each joint according to the extracted key points. For example, the position of the shoulder joint is the coordinate of the left shoulder key point.

[0127] (3) Calculate the length of each bone segment, such as the upper arm length and the thigh length, using the Euclidean distance formula based on the joint position coordinates. The Euclidean distance formula is the prior art in this field and not the creative solution of this application, so it will not be elaborated here.

[0128] (4) Construct the key point coordinates of the human body shape data into a data matrix, where each row represents the three-dimensional coordinates of a key point.

[0129] (5) Standardize the data matrix so that the mean is 0 and the variance is 1.

[0130] (6) Calculate the covariance matrix of the standardized data matrix. The covariance matrix calculation formula is the prior art in this field and not the creative solution of this application, so it will not be elaborated here.

[0131] (7) Perform eigenvalue decomposition on the covariance matrix to obtain eigenvalues and eigenvectors. The eigenvalue decomposition formula is the prior art in this field and not the creative solution of this application, so it will not be elaborated here.

[0132] (8) Select the first M principal components according to the eigenvalue magnitudes, where the principal component directions represent the main extension directions of the bones.

[0133] (9) Fit the skeletal model using the principal component directions to determine the orientation and length of the bones. For example, use the first principal component direction as the extension direction of the bones.

[0134] S1.33: Define the target positions of the end effectors according to the aligned motion trajectory data; the end effectors include hands and feet.

[0135] S1.34: Use the inverse kinematics algorithm θ = IK(p target,W,C,E,D) Calculate the joint angles required for the end effector to reach the target position, where θ represents the joint angles, p target represents the target position, W represents the weight vector, and W = [w 1 ,..., w n , w n represents the weight of the nth joint, n represents the number of joints, C represents an n×2 constraint matrix, and each row represents the minimum and maximum angles of a joint [θ min , θ max , E represents the error tolerance vector, and E = [e 1 , e 2 , e 3 , e 1 , e 2 , e 3 represent the error tolerances in the horizontal, vertical, and vertical directions respectively, D represents the damping coefficient vector, IK(·) represents the inverse kinematics function;

[0136] S1.35: Define key frames based on the calculated joint angles, and generate a preliminary skeletal animation based on the key frames to simulate the user's performance in different motion states;

[0137] S1.36: Optimize the parameters of the skeletal model according to the aligned motion trajectory data and the preliminarily generated skeletal animation, and use an optimization algorithm to minimize the error between the skeletal model and the motion trajectory data. Among them, the optimization algorithm uses the least squares method, and the least squares method is the prior art content in this field and is not the creative solution of this application, so it will not be elaborated here;

[0138] If the error is greater than or equal to the preset error threshold, adjust the skeletal model parameters or the inverse kinematics algorithm;

[0139] S1.37: Generate the final skeletal animation based on the adjusted skeletal model.

[0140] The specific steps of the said S1.35 include:

[0141] A1: Obtain the skeletal model from steps S1.31 and S1.32, including joint positions, bone lengths, and hierarchical relationships, and obtain the calculated joint angle data from step S1.34. Among them, the joint angle data is usually stored in the form of a time series, representing the user's motion state at different time points;

[0142] A2: Define key frames according to the joint angle data, and each key frame represents the skeletal pose at a certain time point;

[0143] A3: Use the linear interpolation algorithm to generate intermediate frames between key frames to make the animation transition smoothly. For the rotation angle, use quaternion interpolation Avoid the gimbal lock problem and generate interpolated joint angles. Here, σ represents the interpolation coefficient, q 1 and q 2 represent the quaternions of key frames, q(σ) represents the interpolated joint angles, and the linear interpolation algorithm is the prior art in this field and not the creative solution of this application, so it will not be elaborated here;

[0144] A4: According to the interpolated joint angles, update the rotation matrix R(θ) of each joint in the skeletal model, and use the rotation matrix to update the global coordinates of the joint. The formula is: P global = M parent × R(θ) × P local , where P global represents the global coordinates of the child node, M parent represents the transformation matrix of the parent node, and P local represents the local coordinates of the child node;

[0145] A5: Bind the skeletal model to the human body shape data and use the formula to calculate the final position Y of each vertex final , where represents the weight, n represents the number of vertices, M i represents the skeletal transformation matrix, and Y o represents the original position of the vertex;

[0146] A6: Use the graphics rendering engine to render the skeletal animation frame by frame, render the skeletal pose of each frame onto the screen, and generate an animation effect. Here, the graphics rendering engine is the prior art in this field and not the creative solution of this application, so it will not be elaborated here.

[0147] The specific steps of S1.4 include:

[0148] S1.41: Obtain the aligned motion data and human body shape data, and extract key points and regions; the key points include joint positions and muscle attachment points; the regions refer to muscle groups and skin regions;

[0149] Further, the specific steps of S1.41 include:

[0150] (1) Obtain the aligned motion data and human body shape data;

[0151] (2) Define key points according to human anatomy knowledge. The key points include shoulder joints, elbow joints, knee joints, hip joints, wrists, and ankles;

[0152] (3) Use wearable sensors to extract the positions and motion trajectories of the key points;

[0153] (4) Mark key points using a feature point detection algorithm. The feature point detection algorithm is prior art in this field and not a creative solution of this application, so it will not be elaborated here;

[0154] (5) Use a region segmentation algorithm to extract the regions corresponding to the four major parts from the human body shape data. For example, use the K-means clustering algorithm to segment the shoulder region from the human body shape data. The region segmentation algorithm is prior art in this field and not a creative solution of this application, so it will not be elaborated here;

[0155] (6) Extract the data of the regions corresponding to the four major parts according to the positions and movement trajectories of the key points. For example, extract the movement data of the shoulder region according to the movement trajectory of the shoulder joint;

[0156] (7) Integrate the movement data, the key points in the human body shape data, and the region data, and compare the accuracy of the key points and the region data to ensure data consistency.

[0157] S1.42: According to anatomical knowledge, define the path of each muscle and represent the muscle path using a B-spline curve;

[0158] Further, the specific steps of S1.42 include:

[0159] (1) According to anatomical knowledge, determine the origin and insertion of each muscle. For example, the origin of the biceps brachii is on the scapula and the insertion is on the radius;

[0160] (2) According to the anatomical structure of the muscle, determine the intermediate path points, such as the path points bypassing the joint. For example, the biceps brachii has an obvious bend at the elbow joint;

[0161] (3) Obtain the human body shape data to ensure that the data contains the position information of the muscle origin, insertion, and intermediate path points;

[0162] (4) Use a feature point detection algorithm to extract the three-dimensional coordinates of the muscle origin, insertion, and intermediate path points from the human body shape data;

[0163] (5) Use the muscle origin, insertion, and intermediate path points as the control points of the B-spline curve. The number of control points depends on the complexity of the muscle path;

[0164] (6) Use the B-spline curve formula to calculate the muscle path, where G(u) represents the B-spline curve, Boor l,v (u) represents the B-spline basis function, Q l represents the control point, q represents the number of control points, and v represents the curve order.

[0165] S1.43: Initialize muscle parameters according to the motion data and human body shape data, where the muscle parameters include length, cross-sectional area, and contractile force;

[0166] S1.44: Calculate the length change and contractile force of the muscle according to the joint angle and muscle path, and simulate the mechanical behavior of the muscle using a muscle model. Among them, the muscle model adopts the Hill-type model, and the Hill-type model is the prior art content in this field and not the creative solution of this application, so it will not be elaborated here;

[0167] Further, the specific steps for calculating the length change and contractile force of the muscle according to the joint angle and muscle path include:

[0168] (1) Obtain joint angle data and muscle path data;

[0169] (2) Calculate the length of the muscle in different motion states using the arc length formula of the B-spline curve according to the joint angle and muscle path where G′(u) represents the first derivative of the B-spline curve, and u 1 and u 2 represent control points;

[0170] (3) Calculate the arc length H using numerical integration method L , where the numerical integration method is the prior art content in this field and not the creative solution of this application, so it will not be elaborated here;

[0171] (4) Obtain the muscle length change ΔL according to the difference between the initial length and the current length;

[0172] (5) Calculate the contractile force of the muscle using the Hill-type muscle model according to the muscle length change and activation degree, where the Hill-type muscle model is the prior art content in this field and not the creative solution of this application, so it will not be elaborated here.

[0173] S1.45: Construct a surface mesh model of the skin according to the human body shape data, and generate a skin mesh using a mesh generation algorithm. Among them, the mesh generation algorithm is the prior art content in this field and not the creative solution of this application, so it will not be elaborated here;

[0174] Further, the specific steps for constructing the surface mesh model of the skin include:

[0175] (1) Obtain human body shape data, and extract point cloud data on the skin surface from the human body shape data using a surface extraction algorithm. Among them, the surface extraction algorithm is the prior art content in this field and not the creative solution of this application, so it will not be elaborated here;

[0176] (2) Based on the point cloud data of the skin surface, use the Delaunay triangulation algorithm to generate a triangular mesh of the skin surface. Among them, the Delaunay triangulation algorithm is the prior art in this field and is not the creative solution of this application, so it will not be elaborated here.

[0177] S1.46: Initialize the skin parameters according to the biomechanical properties of the skin, and define the boundary conditions of the skin according to the muscle contraction force and joint movement;

[0178] S1.47: Discretize the skin mesh into finite elements. Based on the finite elements, establish a finite element equation according to the theory of elasticity and solve the equation. Among them, the finite element analysis method and the finite element equation are the prior art in this field and are not the creative solution of this application, so it will not be elaborated here;

[0179] S1.48: Update the vertex positions of the skin mesh according to the solution results to generate a deformed skin model.

[0180] Further, the specific steps of S1.48 include:

[0181] (1) Obtain the displacement vector from the finite element analysis results;

[0182] (2) Obtain the surface mesh model of the skin, including vertex coordinates, edge and face information;

[0183] (3) Update the coordinates of each vertex according to the displacement vector;

[0184] (4) Use the topology check algorithm to check whether the updated mesh maintains the correct topology structure, such as no self-intersection and no degenerate faces. Among them, the topology check algorithm is the prior art in this field and is not the creative solution of this application, so it will not be elaborated here;

[0185] (5) Write the updated vertex coordinates into the mesh data file and use a visualization tool to display the deformed skin model. In the present invention, the visualization tool uses ParaView.

[0186] The specific steps of step S2 include:

[0187] S2.1: Obtain multiple types of data and transmit the multiple types of data to the built-in motion biomechanics analysis module of the system;

[0188] S2.2: Use the coordinate transformation algorithm to adapt the multiple types of data to the human body dynamic model. Among them, the coordinate transformation algorithm is the prior art in this field and is not the creative solution of this application, so it will not be elaborated here;

[0189] S2.3: Analyze the human body dynamic model using a deep learning-based object detection algorithm to identify the four major parts in clothing design and extract the data of the four major parts during the movement process; the four major parts include the shoulders, elbows, knees, and waist; the movement process data includes position and morphological changes. Among them, the deep learning-based object detection algorithm is the prior art content in this field and is not the creative solution of this application, so it will not be elaborated here.

[0190] S2.4: According to the principles of kinematics and dynamics, conduct a mechanical analysis on the identified four major parts. The mechanical analysis includes calculating the displacement, velocity, acceleration, force, and torque of each part during the movement process.

[0191] S2.5: According to the results of the mechanical analysis and combined with the characteristics of different sports events, determine the activity requirements and force conditions of the four major parts in different movement states.

[0192] The specific steps of step S3 include:

[0193] S3.1: Extract the activity requirements of the four major parts from the results of the sports biomechanics analysis and quantify the activity requirements into feature vectors to obtain the activity requirement feature vectors; the activity requirements include the activity range, force distribution, and ductility requirements.

[0194] S3.2: Load a pre-trained machine learning model, which is configured based on the random forest algorithm. Among them, the machine learning model is the prior art content in this field and is not the creative solution of this application, so it will not be elaborated here.

[0195] S3.3: Input the activity requirement feature vectors into the pre-trained machine learning model to generate a clothing design plan that meets the optimization goal and output the clothing design parameters at the same time.

[0196] S3.4: Extract the fabric features from the fabric database and match the fabric features with the clothing design parameters.

[0197] S3.5: According to the matching results, generate a fabric recommendation list and sort it according to the priority.

[0198] Embodiment 2

[0199] Please refer to Figure 3 , another embodiment provided by the present invention: A clothing design system based on intelligent interaction, including:

[0200] A data acquisition module, a sports biomechanics analysis module, a clothing design optimization module, a fabric and structure design module, a virtual simulation test module, and an augmented reality module;

[0201] The data acquisition module is used to collect various types of data of users, providing a basis for subsequent analysis and design;

[0202] The sports biomechanics analysis module is used to analyze the body mechanics changes of users in different exercise states according to the collected various types of data, and identify the key parts of clothing design, their activity requirements and force conditions;

[0203] The clothing design optimization module is used to automatically optimize clothing design and generate fabric recommendations through machine learning algorithms, combined with the activity requirements of the four major parts;

[0204] The fabric and structure design module is used to determine the final fabric and design the clothing structure according to the fabric recommendation list and design requirements;

[0205] The virtual simulation test module is used to simulate the performance of clothing in different exercise states and conduct tests through virtual simulation technology;

[0206] The augmented reality module is used to try on the clothing designs that pass the simulation test in real time through augmented reality methods, and automatically adjust the design scheme according to user feedback and dynamic performance.

[0207] The data acquisition module includes: a 3D scanning unit, a wearable sensor unit, and a motion capture unit;

[0208] The 3D scanning unit is used to obtain the body shape data of users using 3D scanning technology;

[0209] The wearable sensor unit collects the motion data of users through wearable devices;

[0210] The motion capture unit is used to accurately record the actions and motion trajectories of users using motion capture devices.

[0211] The sports biomechanics analysis module includes: a mechanics analysis unit, a part identification unit, and an activity requirement and force analysis unit;

[0212] The mechanics analysis unit is used to analyze the body mechanics changes of users during exercise;

[0213] The part identification unit is used to identify the four major parts of clothing design, such as shoulders, chests, waists, and legs;

[0214] The activity requirement and force analysis unit is used to determine the activity requirements and force conditions of the four major parts in different exercise states.

[0215] The clothing design optimization module includes: a machine learning algorithm unit and a fabric recommendation unit;

[0216] The machine learning algorithm unit is used to optimize clothing design using machine learning technology;

[0217] A fabric recommendation unit for generating a fabric recommendation list based on the analysis results.

[0218] The fabric and structure design module includes:

[0219] A fabric selection unit for selecting the final fabric by combining the recommendation list and design requirements;

[0220] A structure design unit for designing the clothing structure according to the movement requirements of the four major parts.

[0221] The virtual simulation test module includes: a simulation unit and a test evaluation unit;

[0222] The simulation unit for applying the fabric and structure design to a human body dynamic model for simulation;

[0223] The test evaluation unit for testing and evaluating the simulation results.

[0224] The augmented reality module includes: an augmented reality try-on unit and a design scheme adjustment unit;

[0225] The augmented reality try-on unit for simulating the movement state in a virtual environment and displaying the clothing design effect in real time;

[0226] The design scheme adjustment unit for automatically adjusting the design scheme according to user feedback and dynamic performance.

[0227] The embodiments of the present invention have been described above in conjunction with the accompanying drawings. However, the present invention is not limited to the above specific embodiments. The above specific embodiments are merely illustrative and not restrictive. Under the inspiration of the present invention, those of ordinary skill in the art can also make changes, modifications, substitutions, and variations to the above embodiments without departing from the purpose and scope of the present invention. These all fall within the protection scope of the present invention.

[0228] If the disclosed technical solution involves personal information, the product using the disclosed technical solution has clearly informed the personal information processing rules and obtained the individual's voluntary consent before processing the personal information. If the disclosed technical solution involves sensitive personal information, the product using the disclosed technical solution has obtained the individual's separate consent before processing the sensitive personal information, and at the same time meets the "explicit consent" requirement. For example, on personal information collection devices such as cameras, clear and prominent signs are set to inform that the personal information collection scope has been entered and personal information will be collected. If the individual voluntarily enters the collection scope, it is deemed that he or she agrees to the collection of his or her personal information; or on the device that processes personal information, when the personal information processing rules are notified by obvious signs / information, the individual's authorization is obtained through pop-up information or by asking the individual to upload his or her personal information; among them, the personal information processing rules may include information such as the personal information processor, the purpose of personal information processing, the processing method, and the type of personal information processed.

Claims

1. A clothing design method based on intelligent interaction, characterized in that: include: Step S1: Collect multiple types of data to build a human body dynamic model of the user; the multiple types of data include human body shape data, motion data and motion trajectory data; Step S2: Based on the built-in sports biomechanics analysis module of the system and according to the human body dynamic model, the changes in the user's body mechanics under different sports states are analyzed, and at the same time, the four major parts in the clothing design are identified, and the activity requirements and stress conditions of the four major parts under different sports states are determined; Step S3: Automatically optimize clothing design by combining the activity requirements of the four major parts through machine learning algorithms, and generate optimized clothing design solutions and fabric recommendation lists based on the results of sports biomechanics analysis; Step S4: Determine the final fabric according to the fabric recommendation list and the design requirements, and design the clothing structure according to the activity requirements of the four major parts to obtain the final fabric selection plan and clothing structure design drawing; Step S5: applying the final fabric selection scheme and clothing structure design drawing to the human body dynamic model through virtual simulation technology to simulate the performance of the clothing under different motion states. The system simulates the performance of the clothing under different motion states and performs simulation tests. Step S6: Through the augmented reality method, the clothing design scheme that has passed the simulation test is tried on in real time, and the movement state is simulated in the virtual environment. The system automatically adjusts the clothing design scheme according to the user's feedback and dynamic performance.

2. The clothing design method based on intelligent interaction according to claim 1, characterized in that: The specific steps of step S1 include: S1.1: Use 3D scanning technology to perform a full body scan on the user to obtain the user's human body shape data, and use wearable sensors to collect the user's motion data under specific movements. At the same time, use an optical motion capture system to attach reflective markers to key points on the user's body to capture the user's motion trajectory data under different motion states; the human body shape data includes body shape, contour, joint position, and muscle distribution; the motion data includes acceleration, angular velocity, and pressure distribution; S1.2: Sort and match the human body shape data, motion data and motion trajectory data according to timestamps; If the timestamps of multiple types of data are inconsistent, the improved linear interpolation formula is used Align, where y(t) represents the interpolation result, y1 and y2 represent multi-class data at adjacent time points, t1 and t2 represent the timestamps corresponding to y1 and y2, t represents the current time, α represents the weight factor, and μ represents the smoothing parameter; If there is a time offset between devices, the time offset is calculated and corrected by a cross-correlation analysis method; the cross-correlation analysis method is to calculate the cross-correlation function of two time series data and find the time delay τ that makes the function value reach the maximum in the cross-correlation function. max , according to the time delay τ found max , adjust the timestamp of one type of data to match the timestamp of the other type of data.

3. The clothing design method based on intelligent interaction as claimed in claim 2, characterized in that: The specific steps of step S1 also include: S1.3: Based on the aligned human morphology data and motion trajectory data, a skeleton model of the user is constructed, and the inverse kinematics algorithm is used to calculate the joint angle and range of motion; S1.4: Based on the aligned motion data and human morphology data, construct the user's muscle model and skin model, simulate muscle contraction and skin extension, and use the finite element analysis method to simulate the deformation of the skin under different motion states; S1.5: The skeleton model is used as the first layer model, the muscle model is used as the second layer model, and the skin model is used as the third layer model. Through integration, a dynamic human body model of the user is formed, and the physical engine is used to simulate the performance of the model in different motion states.

4. The clothing design method based on intelligent interaction as claimed in claim 3, characterized in that: The specific steps of S1.3 include: S1.31: Based on human anatomy, define the bone structure and use a tree structure to represent the bone hierarchy, including joint positions, bone lengths, and joint degrees of freedom; S1.32: Initialize the skeleton model parameters according to the aligned human morphology data, and fit the skeleton model using the principal component analysis method; S1.33: defining a target position of an end effector according to the aligned motion trajectory data; the end effector includes a hand and a foot; S1.34: Use the inverse kinematics algorithm to calculate the joint angle θ = IK (p target ,W,C,E,D), where θ represents the joint angle, p target represents the target position, W represents the weight vector, and W=[w1,...,w n ],w n represents the weight of the nth joint, n represents the number of joints, C represents the constraint matrix with n rows and 2 columns, and each row represents the minimum and maximum angles of a joint [θ min ,θ max ], E represents the error tolerance vector, and E=[e1,e2,e3], e1, e2, e3 represent the error tolerance in the horizontal, longitudinal and vertical directions respectively, D represents the damping coefficient vector, and IK(·) represents the inverse kinematics function; S1.35: Define keyframes based on the calculated joint angles, and generate preliminary skeletal animations based on the keyframes to simulate the user's performance in different motion states; S1.36: Optimize the parameters of the skeleton model according to the aligned motion trajectory data and the initially generated skeleton animation, and use an optimization algorithm to minimize the error between the skeleton model and the motion trajectory data; If the error is greater than or equal to a preset error threshold, the skeleton model parameters or the inverse kinematics algorithm are adjusted; S1.37: Generate the final skeletal animation based on the adjusted skeletal model.

5. The clothing design method based on intelligent interaction as claimed in claim 4, characterized in that: The specific steps of S1.35 include: A1: Obtain the skeleton model from steps S1.31 and S1.32, including joint positions, bone lengths and hierarchical relationships, and obtain the calculated joint angle data from step S1.34; A2: Define keyframes based on joint angle data; A3: Use linear interpolation to generate intermediate frames between key frames, and use the formula Generate interpolated joint angles, where σ represents the interpolation coefficient, q1 and q2 represent the quaternions of the key frames, and q(σ) represents the interpolated joint angles; A4: Update the rotation matrix R(θ) of each joint in the skeleton model according to the interpolated joint angle, and use the rotation matrix to update the global coordinates of the joint. The formula is: P global =M parent ×R(θ)×P local , where P global Represents the global coordinates of the child node, M parent Represents the transformation matrix of the parent node, P local Represents the local coordinates of the child node; A5: Bind the skeleton model to the human body shape data and use the formula Calculate the final position Y of each vertex final ,in, represents the weight, n represents the number of vertices, M i Represents the bone transformation matrix, Y o Indicates the original position of the vertex; A6: Use a graphics rendering engine to render skeletal animation frame by frame, render the skeletal posture of each frame onto the screen, and generate animation effects.

6. The clothing design method based on intelligent interaction as claimed in claim 5, characterized in that: The specific steps of S1.4 include: S1.41: Acquire the aligned motion data and human body shape data, and extract key points and regions; the key points include joint positions and muscle attachment points; the regions refer to muscle groups and skin regions; S1.42: Based on anatomical knowledge, define the path of each muscle and use B-spline curves to represent the muscle path; S1.43: Initializing muscle parameters according to the motion data and the human body morphology data, wherein the muscle parameters include length, cross-sectional area, and contraction force; S1.44: Calculate the length change and contraction force of the muscle based on the joint angle and muscle path, and use the muscle model to simulate the mechanical behavior of the muscle; S1.45: Construct a surface mesh model of the skin based on human body morphology data, and generate a skin mesh using a mesh generation algorithm; S1.46: Initialize skin parameters based on the biomechanical properties of the skin and define the boundary conditions of the skin based on muscle contraction and joint motion; S1.47: Discretize the skin mesh into finite elements, establish finite element equations based on the finite elements and solve the equations according to the theory of elasticity; S1.48: According to the solution results, the vertex positions of the skin mesh are updated to generate a deformed skin model.

7. The clothing design method based on intelligent interaction according to claim 6, characterized in that: The specific steps of step S2 include: S2.1: Acquire multiple types of data and transmit the multiple types of data to the sports biomechanics analysis module built into the system; S2.2: Use coordinate transformation algorithm to adapt multi-class data to human body dynamic model; S2.3: Use a deep learning-based target detection algorithm to analyze the human body dynamic model, identify four major parts in clothing design, and extract the motion process data of the four major parts; the four major parts include shoulders, elbows, knees, and waist; the motion process data includes position and shape changes; S2.4: Perform mechanical analysis on the four identified parts according to the principles of kinematics and dynamics, wherein the mechanical analysis includes calculating the displacement, velocity, acceleration, and force and torque of each part during the movement; S2.5: Based on the results of mechanical analysis and the characteristics of different sports, determine the activity requirements and stress conditions of the four major parts under different sports conditions.

8. The clothing design method based on intelligent interaction as claimed in claim 7, characterized in that: The specific steps of step S3 include: S3.1: Extract the activity requirements of the four major parts from the results of sports biomechanics analysis, and quantify the activity requirements into feature vectors to obtain activity requirement feature vectors; the activity requirements include activity range, force distribution and ductility requirements; S3.2: Load a pre-trained machine learning model, where the machine learning model is configured based on a random forest algorithm; S3.3: Input the activity demand feature vector into the pre-trained machine learning model to generate a clothing design solution that meets the optimization goal and output the clothing design parameters; S3.4: extract fabric features from the fabric database and match the fabric features with clothing design parameters; S3.5: Generate a fabric recommendation list based on the matching results and sort them by priority.

9. A clothing design system based on intelligent interaction, which is used to implement the clothing design method based on intelligent interaction according to any one of claims 1 to 8, characterized in that: include: Data acquisition module, sports biomechanics analysis module, clothing design optimization module, fabric and structure design module, virtual simulation test module, augmented reality module; The data collection module is used to collect multiple types of data of users; The sports biomechanics analysis module is used to analyze the changes in the user's body mechanics under different sports conditions based on the collected multiple types of data, and to identify the key parts of the clothing design and their activity requirements and stress conditions; The clothing design optimization module is used to automatically optimize clothing design and generate fabric recommendations by combining the activity requirements of the four major parts through a machine learning algorithm; The fabric and structure design module is used to determine the final fabric and design the garment structure according to the fabric recommendation list and design requirements; The virtual simulation test module is used to simulate the performance of the clothing in different motion states through virtual simulation technology and perform tests; The augmented reality module is used to try on the clothing design that has passed the simulation test in real time through the augmented reality method, and automatically adjust the design scheme according to user feedback and dynamic performance.

Images