A method for human body measurement and personalized clothing pattern generation based on multi-body adaptation

Abstract

The present invention belongs to the technical field of customized clothing pattern making and relates to a method for human body measurement and personalized clothing pattern generation that is adaptable to multiple body types. The method first obtains three-dimensional human body scanning data and fits the acquired data using a parametric human body model. Then, according to preset measurement item characteristics, key human body dimensions are automatically measured, and a basic clothing pattern prototype is generated based on the key human body dimension data. Then, the pattern style design parameters are adjusted according to design requirements to generate clothing patterns of corresponding styles. Finally, the generated clothing pattern effect is verified using three-dimensional virtual fitting technology. Based on the virtual fitting results, multiple rounds of parameter optimization and adjustment are performed until the clothing pattern achieves the desired effect. The generated clothing pattern and its related data files are stored, completing the personalized clothing pattern generation process. The present invention can achieve automated and high-precision measurement of human body dimensions of multiple body types and can quickly generate personalized clothing patterns in batches.

Description

Technical Field

[0001] The present invention belongs to the technical field of customized clothing pattern making, and relates to a method for generating human body measurements and personalized clothing patterns suitable for multiple body types. Background Art

[0002] With the growing demand for personalization and customization in the modern consumer market, the apparel industry is undergoing a profound transformation from large-scale standardized production to flexible manufacturing and personalized customization. When choosing clothing, consumers prioritize comfort, fit, and individual expression, in addition to fashion trends and style. Therefore, how to meet personalized demands while improving production efficiency and design precision has become a critical challenge for the apparel industry. Traditional clothing pattern making methods, such as the proportional method, the short-measurement method, and the prototype method, still have many limitations in practical application and are unable to meet the diverse and sophisticated demands of today's market.

[0003] First, the traditional pattern making process relies on manual operation, which is inefficient and has a long design cycle. This pattern making method not only fails to respond quickly to market demand, but also leads to increased production costs, weakening the market competitiveness of clothing products. At the same time, manual measurement and pattern making processes are easily affected by human errors, especially when dealing with dynamic human bodies or special body shapes. The accuracy of pattern making is difficult to guarantee, which in turn affects the fit and comfort of clothing. For example, the prototype method requires the operator to manually measure the human body size, which cannot achieve full process automation. In addition, the clothing prototypes mentioned in Literature 1 (Research on Eastern and Western Women's Clothing Prototypes [J]. Shandong Textile Science and Technology, 2018, 59(03): 6-9.), Literature 2 (Research Progress and Trends of Clothing Prototypes for Elderly Women [J]. Progress in Textile Science and Technology, 2022, (05): 44-49.) and CN117652742A are mostly based on statistical data of specific human bodies, and only use chest circumference, waist circumference, and hip circumference as the basis for universal estimation, which is difficult to meet the personalized customization needs of special body shapes.

[0004] In recent years, with the rapid development of 3D body scanning technology, clothing pattern-making methods based on human body data have garnered widespread attention. 3D scanning technology can quickly acquire precise geometric information about the human body surface, providing detailed dimensional data and body shape feature analysis support for clothing pattern-making. However, the application of 3D body scanning technology in the clothing industry is still in its exploratory stages and faces numerous technical bottlenecks. Existing 3D body scan data processing generally lacks efficient and accurate anthropometry algorithms, making it difficult to automatically extract key dimensions for multiple poses and specific body shapes. Consequently, the clothing pattern-making process has yet to be fully automated.

[0005] For example, the method proposed in Reference 3 (Research on Individualized Youth Women's Clothing Pattern Generation Rules Based on Local Feature Analysis [D]. Suzhou University, 2014) restricts the body scan posture and requires manual definition of measurement features, which introduces certain subjective biases. Furthermore, similar to the prototype method, some dimensions are still obtained through regression fitting of overall population measurement data, lacking personalized and automated pattern generation capabilities. The prototype patterns generated by this method have fixed styles and cannot automatically adjust, add, remove, or combine components based on body characteristics and style design requirements, resulting in relatively limited customization capabilities. Furthermore, the generated patterns cannot support automated virtual fitting, making it difficult to further optimize and adjust designs through rapid virtual fitting results.

[0006] Patent CN110163728B proposes a method for extracting human body features based on horizontal cross-sections. This method requires correction and regularization of the scanned human body data, which may interfere with the accurate acquisition of the real human body size to a certain extent. In addition, this method is difficult to accurately extract feature data when dealing with different postures or special body types (such as bending body types). For example, when a person bends, the real chest circumference line is not as large as the chest line. Figure 1 (a) shows that the chest circumference line extracted by horizontal cross-section is as follows Figure 1 As shown in (b), the two are quite different and cannot accurately reflect the actual size of the human body.

[0007] Reference 4 (Research and Application of Personalized Clothing Pattern Generation Method [J]. Shanghai Textile Science and Technology, 2020, 48(06): 5-7+22.) and patent CN108634459B proposed a clothing pattern generation method based on human body surface flattening. However, due to the complexity of the human body surface geometry, distortion and deformation will inevitably be introduced during the surface flattening process, resulting in uncontrollable deformation of the generated clothing pattern. The calculation results of different flattening algorithms may vary significantly, thus affecting the consistency of pattern generation. In addition, although this type of method generates clothing patterns based on the human body surface, in actual wearing, the clothing does not completely fit the human body surface, especially in the clothing shape of specific parts such as the female chest. Extracting a high-fitting pattern suitable for tight-fitting clothing from the flattening results usually requires a lot of manual post-processing operations, and this manual adjustment makes the generated tight-fitting clothing pattern difficult to adapt to further automated adjustment, limiting the diversity and flexibility of clothing style pattern design, thereby restricting the scope of application of this method.

[0008] Therefore, it is of great significance to study a method for human body measurement and personalized clothing sample generation that is suitable for multiple body types in order to solve the problems existing in the existing technology. Summary of the Invention

[0009] The purpose of the present invention is to solve the problems existing in the prior art and to provide a method for generating human body measurements and personalized clothing samples that are suitable for multiple body types.

[0010] In order to achieve the above object, the technical solution adopted by the present invention is as follows:

[0011] A method for human body measurement and personalized clothing pattern generation that adapts to multiple body types, first obtains 3D human body scan data and fits the obtained 3D human body scan data using a parametric human body model; then, based on preset measurement item features, automatically measures key human body dimensions and generates a basic clothing pattern prototype based on the key human body dimension data; then, adjusts pattern design parameters according to design requirements to generate clothing patterns of corresponding styles; finally, verifies the generated clothing pattern effect using 3D virtual fitting technology, and performs multiple rounds of parameter optimization and adjustment based on the virtual fitting results until the clothing pattern achieves the desired effect. The generated clothing pattern and its related data files are then stored, completing the personalized clothing pattern generation process;

[0012] Automatically measure key human dimensions based on preset measurement item features. This involves aligning the 3D human body with a parametric human body model, and combining it with preset joint points, surface partition dictionaries, and measurement feature point definitions to accurately measure the length, circumference, distance, and angle dimensions of the parametric human body model.

[0013] Generating basic clothing pattern prototypes based on key human body size data means designing pattern generation rules that can dynamically allocate darts based on the measured key human body size data, and based on the designed pattern generation rules, using the pattern generation system to generate basic clothing pattern prototypes and parameter configuration files to provide support for subsequent style adjustments.

[0014] Existing clothing pattern making technology usually only uses a small number of dimensions such as chest circumference and waist circumference as basic parameters, and the dimensions of other parts are calculated through empirical formulas or statistical data. Although this simplified design method based on limited human body dimensions has certain feasibility in standard body applications, it is difficult to accurately reflect individual body differences, especially for people with non-standard body shapes, their fit and comfort are often not effectively guaranteed. The present invention can automatically integrate multiple measurement data, including multiple data on length, distance, circumference and angle. Based on the professional design specifications of clothing industry samples, the size parameters and dart distribution of the clothing prototype are intelligently adjusted, and finally a clothing sample that is highly matched with the individual body characteristics is generated. Experiments show that the sample generated by the method of the present invention has a significant improvement in fit index compared with the existing technology. It can not only more accurately reflect personalized human characteristics, but also meet the dual requirements of precision and efficiency for modern clothing industrial production.

[0015] The pattern generation system includes pattern generation rules, domain-specific programming languages, and various other algorithms and interfaces. Existing technologies directly apply the pattern generation rules in the GarmentCode pattern generation system to generate clothing patterns and perform 3D virtual fitting. However, the existing pattern generation rules in the GarmentCode pattern generation system have obvious technical flaws, as analyzed below:

[0016] 1. Data dimension limitation problem:

[0017] Existing pattern generation rules rely on static, limited-dimensional human body measurement data, resulting in technical flaws in the generated clothing patterns, which are insufficiently adapted to three-dimensional human features. This is particularly true for dynamic human motion simulations and applications involving specific body types. The resulting patterns exhibit significant deviations in accuracy from the actual human body curve, failing to meet the production requirements of the modern apparel industry.

[0018] 2. Problems with intelligent allocation of provincial roads:

[0019] The current rules lack a dynamic dart allocation mechanism based on human morphology. Their fixed, parameterized dart generation model leads to inaccurate dart calculations and unbalanced spatial distribution. For example, the darts in the template may be too small. This technical shortcoming severely limits the application value of automated template systems in high-end customization and large-scale flexible production.

[0020] The present invention incorporates three-dimensional human body scanning technology to fit human body scan data into a parametric human model, from which multi-dimensional human body dimension data is extracted. This method not only accurately reflects static human features but also simulates dynamic motion states and special body shape scenarios, thereby achieving more comprehensive and detailed human body measurements. Furthermore, the present invention expands the data dimensions relied upon in the template generation rules, significantly improving the compatibility of the generated template with three-dimensional human features and resolving technical drawbacks caused by data dimensional limitations. The present invention also adaptively adjusts the existing template generation rules and introduces a dynamic dart allocation mechanism based on human morphological features. This mechanism intelligently calculates and rationally allocates dart positions and sizes based on multi-dimensional human key dimension data, ensuring accurate and balanced dart distribution. Furthermore, based on the improved generation system, the present invention first generates personalized prototype templates, and then further generates personalized templates that meet specific style requirements, significantly improving the flexibility and adaptability of template generation.

[0021] As the preferred technical solution:

[0022] The above-mentioned method for body measurement and personalized clothing sample generation for multiple body types specifically includes the following steps:

[0023] Step 1: Use 3D body scanning equipment to collect high-precision body data of the user, and use data processing software to pre-process the collected data. At the same time, record the thickness of the user's clothing during scanning, and export it to obtain a 3D scanned body mesh model;

[0024] Step 2: Select a parametric human model and register it with the 3D scanned human mesh model from Step 1 (using a prior art mesh registration algorithm, such as the one disclosed in 5 (Keep it SMPL: Automated Estimation of 3D Human Pose and Shape from a Single Image [C], ECCV 2016). The selected parametric human model has virtual joints for driving pose adjustment. Its shape and movement are controlled by a set of parameters, and pre-training learning establishes a correlation between human morphology and pose changes.

[0025] Step 3: Based on the clothing thickness information recorded in step 1 and the preset distance between the clothing and the human body, the parametric human body model that has been registered with the 3D scanned human body mesh model is scaled and adjusted as a whole, and the parametric human body model data before and after the adjustment is saved;

[0026] Step 4: Call the joint points, surface partition dictionary and measurement feature point definitions of the parametric human body model, apply various measurement algorithms, and automatically measure and record various key dimensions of the human body, including length, circumference, distance and angle;

[0027] Step 5: After verifying the integrity of the size information, the size data obtained in step 4 is used as the clothing prototype constraint. The pattern generation system automatically adjusts the size and dart position during the pattern generation process according to the clothing prototype constraint, generates and saves the personalized clothing prototype pattern and its parameter configuration file;

[0028] Step 6: According to the clothing style design requirements, the personalized clothing prototype sample parameters and configuration files obtained in step 5 are customized and adjusted, and the personalized style sample and parameter configuration files are generated using the sample generation system;

[0029] Step 7: Perform automated virtual fitting on the personalized style sample generated in step 6, repeatedly adjust the sample design based on the fitting effect until the expected effect is achieved, and finally save the personalized clothing sample and its related files.

[0030] In the above-described method for body measurement and personalized clothing pattern generation for multiple body types, the 3D body scanning device in step 1 needs to be able to obtain full-body data, with the defective area due to occlusion accounting for no more than 10%. The scanning accuracy error of the 3D body scanning device must be within ±5 mm, and must meet the standard GB / T 23698-2023 (General requirements for 3D scanning body measurement methods).

[0031] In the above-mentioned method for human body measurement and personalized clothing sample generation for multi-body adaptation, the raw data collected by the scanning device may have many defects and need to be preprocessed. The preprocessing in step 1 includes but is not limited to mesh denoising, defect repair and retopology processing.

[0032] In the above-described method for human body measurement and personalized clothing pattern generation for multiple body types, the parametric human body model selected in step 2 is the SMPL parametric human body model. The SMPL parametric human body model is derived from Reference 6 (SMPL:Askinned multi-person linear model[M] / / Seminal Graphics Papers:Pushing the Boundaries,Volume 2,2023:851-866.).

[0033] In the above-mentioned method for human body measurement and personalized clothing pattern generation for multi-body adaptation, after the registration is completed in step 2, the chamfer distance between the parameterized human body model and the 3D scanned human body mesh model is less than 5 mm, except for the head area. The calculation formula of the chamfer distance is:

[0034]

[0035] Among them, D chamfer (A, B) is the chamfer distance between A and B, A and B are the vertex sets of the parameterized human body model and the 3D scanned human body mesh model, respectively. a∈A and b∈B represent individual vertices of the parameterized human body model and the 3D scanned human body mesh model, respectively. ‖ab‖ is the Euclidean distance from point a to point b. |A| and |B| are the number of vertices in A and B, respectively.

[0036] In the above-mentioned method for human body measurement and personalized clothing pattern generation for multi-body adaptation, the overall scaling of the parameterized human body model registered with the 3D scanned human body mesh model in step 3 is controlled by the surface offset distance. The algebraic expression of the surface offset distance between the scaled model and the original model is:

[0037] D scale =D gap -D cloth ;

[0038] Among them, D scale D is the surface offset distance between the scaled model and the original model. gap is the preset distance between clothing and human body, D cloth is the clothing thickness recorded in step 1.

[0039] In the above-described method for human body measurement and personalized clothing pattern generation for multi-body adaptation, the surface partition dictionary in step 4 consists of the names of the parts of the parameterized human body model and the corresponding face indices; the measurement feature points are vertices on the surface of the parameterized human body model, or the average of the spatial coordinates of two surface vertices.

[0040] In the above-mentioned method for body measurement and personalized clothing pattern generation for multiple body types, verifying the integrity of the size information in step 5 refers to verifying the integrity of the body size measurement data to ensure that it meets the input conditions required for generating personalized clothing prototype patterns.

[0041] In the above-mentioned method for generating human body measurements and personalized clothing patterns for multiple body types, in step 7, a virtual sewing simulation operation is performed according to the spatial arrangement of the personalized style patterns and the definition of their connection relationships to achieve a virtual fitting effect.

[0042] The virtual fitting technology used is derived from literature 7 (Warp: A high-performance python framework for gpu simulation and graphics [C] / / NVIDIA GPU Technology Conference (GTC). 2022.).

[0043] Principle of the invention:

[0044] The present invention discloses a method for human body measurement and personalized clothing pattern generation for multi-body adaptation. Through innovative technical solutions, it effectively solves the problems of insufficient human body measurement accuracy and poor pattern adaptability in traditional clothing customization. In the human body measurement process, in response to the problems of low automation and large measurement errors caused by the existing technology's strict requirements on the posture of the measured person and reliance on manual annotation, this method adopts a technical solution that combines high-precision three-dimensional human body scanning with intelligent grid alignment. By accurately aligning the acquired three-dimensional scanning data with a pre-trained parametric human body model, automatic adaptation to different body shapes and multiple postures is achieved. In combination with a standardized intelligent measurement algorithm, the automation level and measurement accuracy of human body size extraction in multi-posture scenarios are significantly improved.

[0045] In terms of clothing pattern generation technology, the present invention overcomes the limitations of the existing technology that pattern design is based only on length and girth dimensions and the distribution of darts is unreasonable. By constructing a multi-dimensional measurement system containing four major parameters: length, girth, distance and angle, a comprehensive capture of human body morphological characteristics is achieved. Based on the geometric mapping relationship between human body morphology and clothing patterns, the system can automatically optimize the size parameters and dart distribution scheme of clothing prototypes to generate clothing prototype patterns that accurately match individual body shapes. On this basis, the style of the pattern is optimized through a parametric adjustment mechanism, which significantly improves the accuracy of the pattern design. In particular, the pattern generated by this method strictly follows the GarmentCode specification, provides a highly adjustable parametric interface, and supports intelligent adjustment through clothing style parameter configuration files, realizing an automatic virtual stitching function. This technical solution not only greatly improves the automation level and production efficiency of the plate making process, but also further ensures the accuracy and applicability of the pattern design through deep integration with virtual fitting technology.

[0046] The technical advantages of this invention are primarily reflected in the following aspects: first, by combining 3D scanning with an automated measurement algorithm, high-precision and automated body measurement is achieved; second, a pattern generation method based on a multi-dimensional parameter system significantly improves garment fit; and finally, a parametric design interface provides flexible technical support for personalized customization. This technical solution provides an efficient and accurate solution for personalized clothing customization and has significant application value.

[0047] Beneficial effects:

[0048] (1) The present invention provides a method for human body measurement and personalized clothing pattern generation that is adaptable to multiple body types. The method generates a parametric human body model through three-dimensional human body scanning technology, extracts multi-dimensional size data, accurately reflects static features, and simulates dynamic movements and special body types, thereby achieving more refined human body measurement. On this basis, the data dimension of the pattern generation rule is expanded, the adaptability of the pattern to the three-dimensional human body features is improved, and the problem of data limitations is solved. At the same time, a dynamic dart allocation mechanism is introduced to intelligently optimize the position and size of the darts to ensure a reasonable and balanced distribution. The improved system first generates personalized prototype patterns and then generates specific style patterns, which greatly improves the flexibility and adaptability of pattern generation.

[0049] (2) The present invention provides a method for human body measurement and personalized clothing pattern generation that is adaptable to various body shape differences and can realize the automated measurement of key human dimensions. Compared with the traditional method that relies on manual labeling and manual measurement, the present invention eliminates the influence of human errors and significantly improves the accuracy and consistency of measurement results. At the same time, it simplifies the operation process, reduces the threshold for technical use and time cost, ensures the efficiency and practicality of the system, and provides high-precision data support for subsequent clothing pattern generation.

[0050] (3) The present invention provides a method for generating human body measurements and personalized clothing patterns that are adaptable to multiple body types. The adaptive pattern generation system adopted by the present invention can quickly generate personalized prototype patterns that are highly matched with human body dimensions, and has high customization flexibility, supporting a variety of style changes and virtual fitting requirements. By combining clothing style configuration files with virtual fitting technology, the present invention realizes rapid optimization and adjustment of clothing patterns, provides highly personalized style design functions, and fully meets the requirements of modern clothing design for rapid response and high-quality output.

[0051] (4) The method of the present invention for generating human body measurements and personalized clothing patterns for multiple body types not only significantly improves the production efficiency and large-scale application capability of personalized clothing customization, but also breaks through the limitations of existing technologies in terms of human body size extraction accuracy, plate-making efficiency, and degree of automation, providing an efficient and accurate overall solution for modern personalized clothing customization, and providing strong technical support for the continuous innovation and high-quality development of the clothing industry in the fields of intelligent design, rapid production, and personalized customization. It has broad application prospects and significant industrial value. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 Comparison diagram of the actual chest circumference line and the chest circumference line of the horizontal cross-section; (a) is the actual chest circumference line of the human body when bending, and (b) is the chest circumference line extracted by horizontal cross-section when the human body is bending;

[0053] Figure 2 Schematic diagram of the joint points of the parameterized human body model;

[0054] Figure 3 Schematic diagram of the partition of the parameterized human body model; wherein, (a) is a schematic diagram of the partition of the front side of the parameterized human body model, and (b) is a schematic diagram of the partition of the back side of the parameterized human body model;

[0055] Figure 4 Schematic diagram of characteristic points measured on a parametric human body model; (a) is a schematic diagram of characteristic points measured on the front side of the parametric human body model, and (b) is a schematic diagram of characteristic points measured on the back side of the parametric human body model;

[0056] Figure 5 Schematic diagrams of measurement items in a specific embodiment; wherein (a) is a schematic diagram of measurement items on the front of a human body, and (b) is a schematic diagram of measurement items on the back of a human body;

[0057] Figure 6 Figure 1 is the prototype drawing process of the top. (a) is the prototype frame line drawing result of the back half of the top, and (b) is the prototype frame line drawing result of the front half of the top.

[0058] Figure 7 Figure II shows the process of drawing the prototype of the top. (a) shows the result of the process of drawing the prototype darts of the back half of the top, and (b) shows the result of the process of drawing the prototype darts of the front half of the top.

[0059] Figure 8 Figure III shows the process of drawing the prototype of the top. (a) shows the result of drawing the prototype arc of the back half of the top, and (b) shows the result of drawing the prototype arc of the front half of the top.

[0060] Figure 9 Create a schematic diagram for the sleeve prototype;

[0061] Figure 10 Comparison diagrams of prototype and personalized clothing samples; (a) is a comparison diagram of the back half of the top sample, (b) is a comparison diagram of the front half of the top sample, and (c) is a comparison diagram of the sleeve sample;

[0062] Figure 11 The virtual fitting results of the prototype and the personalized clothing sample; among them, (a) is the fitting results of the top prototype and the sleeve prototype, and (b) is the schematic diagram of the personalized clothing sample finally generated after adjustment.

[0063] Figure 12 The results of trying on clothing for the generated sample based on the new cultural prototype rule; (a) is the front view of the clothing deformation visualization, (b) is the side view of the clothing deformation visualization, and (c) is the back view of the clothing deformation visualization.

[0064] Figure 13 The following are the garment fitting results of the sample generated by the built-in rules of GarmentCode; (a) is the front view of the garment deformation visualization, (b) is the side view of the garment deformation visualization, and (c) is the back view of the garment deformation visualization.

[0065] Figure 14 This is the result of trying on a sample garment generated by the method of the present invention; wherein, (a) is a front view of garment deformation visualization, (b) is a side view of garment deformation visualization, and (c) is a rear view of garment deformation visualization.

[0066] Figure 15 Waist cross-section diagrams of different fitting results: (a) is the new cultural prototype rule, (b) is the built-in rule of GarmentCode, and (c) is the rule of the present invention.

[0067] Figure 16 This is an overall flow chart of a method for human body measurement and personalized clothing sample generation for multiple body types adapted to the present invention. DETAILED DESCRIPTION

[0068] Below in conjunction with specific embodiment, further set forth the present invention.Should be understood that these embodiments are only used to illustrate the present invention and are not used in limiting the scope of the present invention.In addition, should be understood that after reading the content taught by the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms fall equally within the scope limited by the appended claims of the application.

[0069] A method for human body measurement and personalized clothing pattern generation based on multiple body types, such as Figure 16 The specific steps are as follows:

[0070] Step 1: Use 3D body scanning equipment to collect high-precision body data of the user, and use data processing software to pre-process the collected data. At the same time, record the thickness of the user's clothing during scanning, and export it to obtain a 3D scanned body mesh model;

[0071] Among them, the scanning accuracy error of the three-dimensional human body scanning equipment is within ±5mm; the preprocessing includes mesh denoising, defect repair and retopology processing; the model after preprocessing must meet the following requirements: a triangular mesh structure is used, the vertex positions are evenly distributed, and the number of vertices is not less than 500; the mesh structure has no non-manifold boundaries and self-intersection phenomena.

[0072] Step 2: Select an SMPL parametric human body model and align it with the 3D scanned human mesh model from Step 1. The selected parametric human body model has virtual joints to drive posture adjustment. Its body shape and movement can be controlled by group parameters. After pre-training, the relationship between human morphology and posture changes is established.

[0073] After registration, except for the head area, the chamfer distance between the parametric human body model and the 3D scanned human body mesh model is less than 5 mm. The calculation formula of the chamfer distance is:

[0074]

[0075] Among them, D chamfer (A, B) is the chamfer distance between A and B, A and B are the vertex sets of the parameterized human body model and the 3D scanned human body mesh model, respectively. a∈A and b∈B represent individual vertices of the parameterized human body model and the 3D scanned human body mesh model, respectively. ‖ab‖ is the Euclidean distance from point a to point b. |A| and |B| are the number of vertices in A and B, respectively.

[0076] Step 3: Based on the clothing thickness information recorded in Step 1 and the preset distance between the clothing and the human body, the parametric human body model that has been aligned with the 3D scanned human body mesh model is scaled and adjusted as a whole using the surface offset distance, and the parametric human body model data before and after the adjustment is saved. The parametric human body model before the scaling adjustment is used for the subsequent virtual fitting, while the parametric human body model after the scaling adjustment is used for the key human body dimension measurement in Step 4. The saved parametric human body model data includes the definition files of the body shape and motion parameters and the corresponding mesh model files.

[0077] The algebraic expression for the surface offset distance between the scaled model and the original model is:

[0078] D scale =D gap -D cloth ;

[0079] Among them, D scale D is the surface offset distance between the scaled model and the original model. gap is the preset distance between clothing and human body, D cloth is the clothing thickness recorded in step 1;

[0080] The spatial coordinate system of the parametric human body model is defined as follows: the frontal plane (the plane parallel to the forehead and perpendicular to the sagittal plane) is set as the x-axis of the coordinate system, the center of the left and right bisection of the human body is set as the x-axis origin, the up-down direction is defined as the y-axis, the ground direction is the y-axis origin, the sagittal direction is defined as the z-axis, and the center of the front-back bisection of the human body is set as the z-axis origin;

[0081] Step 4: Call the joint points, surface partition dictionary and measurement feature point definitions of the parametric human body model, apply various measurement algorithms, and automatically measure and record various key dimensions of the human body, including length, circumference, distance and angle;

[0082] The surface partition dictionary consists of the names of each part of the parameterized human body model and the corresponding face index; the face index is used to uniquely identify each face in the mesh model, and its index is an increasing number starting from 0; each face consists of three vertex indexes;

[0083] The measured feature points are vertices on the surface of the parameterized human body model, or the average of the spatial coordinates of two surface vertices; the vertex index is used to uniquely identify each vertex in the mesh model, and its index is also a number that increases from 0 and corresponds one-to-one to the spatial coordinates of the vertex; the spatial coordinates of the vertex represent the position of the vertex in three-dimensional space, and each vertex is defined by (x, y, z) coordinate information.

[0084] The surface partition dictionary and measurement feature point data are stored in the template generation system and can be repeatedly called under the same parameterized human body model conditions. The automatic measurement methods for each category are as follows:

[0085] Length: Based on the preset measurement item definition, select two feature points and calculate the Euclidean distance between the two points in three-dimensional space as the length of the current measurement item. The calculation formula for Euclidean distance is:

[0086]

[0087] Among them, L eu is the length of the current measurement item, (x1, y1, z1) and (x2, y2, z2) are the three-dimensional coordinates of feature point 1 and feature point 2.

[0088] Circumference size: According to the preset measurement item definition, a feature point and two joint points are selected to determine a cutting plane. The origin of the cutting plane is set as the spatial coordinate of the feature point, and its normal direction is determined by the vector formed by the two joint points, wherein the first joint point is the starting point of the vector and the second joint point is the end point of the vector. The parameterized human body model is intercepted using the cutting plane. In view of the fact that changes in human body shape and movement may lead to the existence of multiple cross-sections in a local area, the obtained cross-sections are screened according to the human body partition dictionary to determine the only valid cross-section. A three-dimensional convex hull processing is performed on the boundary of the valid cross-section to generate a three-dimensional convex hull cross-section. The sum of the lengths of all sides of the three-dimensional convex hull cross-section is calculated and the length of the closed side is subtracted to obtain the sum of the lengths of all sides of the three-dimensional convex hull cross-section as the circumference size of the current measurement item.

[0089] Distance Dimension: Based on the preset measurement item definition, two feature points are selected and the geodesic distance between the two feature points on the surface of the parametric human body model is calculated as the distance dimension of the current measurement item.

[0090] Angle Dimension: Based on the preset measurement item definition, a target feature point is selected as the vertex of the measured angle and set as the starting point of a vector. Two other feature points are further selected as the end points of these vectors, and two measurement angle vectors are constructed. Based on these two measurement angle vectors, the included angle is calculated in 3D space and used as the angle dimension of the current measurement item.

[0091] Among them, the measurement accuracy error of length, circumference and distance categories is ±1mm, and the measurement accuracy error of angle category is ±0.1°.

[0092] Step 5: Verify the integrity of the size information: This is to verify the integrity of the human body size measurement data to ensure that it meets the input conditions required for generating the personalized clothing prototype. If the measurement data does not meet the preset requirements, the system will trigger an error prompt and interrupt subsequent operations. After verifying the integrity of the size information, the size data obtained in step 4 is used as the clothing prototype constraint. The pattern generation system automatically adjusts the size and dart position during the pattern generation process according to the clothing prototype constraint, and generates and saves the personalized clothing prototype pattern and its parameter configuration file.

[0093] The pattern generation system includes pre-configured parametric prototype generation models for various garment categories, including tops, sleeves, skirts, and pants. By default, each prototype pattern consists of the front and back pieces of the right side of the human body, with the left side pattern symmetrically derived from the right side. For situations where key dimensions differ significantly between the left and right sides of the human body, or for other special needs, a separate left side prototype pattern can be generated. The dimensions of the left side prototype pattern are generated based on left-side anthropometric data, and the surface partitioning dictionaries and feature points used for the left and right anthropometric items adhere to the principle of midsagittal symmetry.

[0094] Step 6: Based on the clothing style design requirements, the personalized clothing prototype template and its parameter configuration file obtained in step 5 are customized and adjusted, and a personalized style template and its parameter configuration file are generated using the template generation system;

[0095] The template generation system should support calling configurable parameter templates, allowing direct modification and writing of relevant configurations. It is not only necessary to have the function of parametric adjustment of the basic dimensions of the prototype, but also to provide a flexible personalized adjustment interface to support automatic adjustment based on human body size and style requirements. The adjustment content includes but is not limited to design elements such as the number of darts and the shape of the armhole line. According to the measurement data obtained in step 4 and the design requirements of the template structure, the value range of the template parameters needs to be reasonably set to avoid self-intersection of the template edges and ensure the feasibility and stability of the design. At the same time, the generated template should have an automated virtual fitting function, defining the connection relationship between the templates and their placement in three-dimensional space, so that users can intuitively evaluate the design effect and make further optimizations and modifications based on feedback.

[0096] The pattern generation system should include built-in component definitions compatible with the pattern generation system in step 4, including but not limited to various neckline shapes, armhole shapes, cuff shapes, slit shapes, etc. Users can modify, add, or combine component design features in the parameter configuration file within the set value range and regenerate the pattern using the pattern generation system, thereby changing the shape and style of the pattern.

[0097] Step 7: Based on the spatial arrangement and connection relationship definition of the personalized style sample generated in step 6, perform virtual sewing simulation operations to achieve an automated virtual fitting effect. Then, adjust the sample design repeatedly according to the fitting effect until the expected effect is achieved. Finally, save the personalized clothing sample and its related files. The files should include the sample vector format, parameter configuration file, and virtual fitting related data files.

[0098] The following uses a specific embodiment to illustrate a method for generating a human body measurement and personalized clothing pattern for multiple body types according to the present invention. The method takes the generation of a V-neck T-shirt pattern with a loose waist as an example. The details are as follows:

[0099] (1) The subject wears a 1mm thick body-fitting garment and stands in a fixed position on the ground in a specific posture. A handheld 3D infrared scanner is used to obtain complete human body data. Magic3D software is used to pre-process the basic mesh model and export the 3D scanned human body mesh model file in obj format.

[0100] (2) Select the SMPL parametric human body model. The SMPL parametric human body model is composed of 10-dimensional shape parameters, 72-dimensional posture parameters, and 3-dimensional translation parameters to determine the shape, posture, and position of the human body. Apply the grid registration algorithm to iteratively optimize the shape, posture, and translation parameters of the model, and match the scan to obtain a human body mesh model. The number of iterations of the grid registration algorithm is set to 500, the learning rate is set to 0.1, the starting number of learning rate decay is 450, and the normal vector threshold angle is 30°. After the human body posture and movement matching is completed, the chamfer distance between the scanned mesh and the parametric human body mesh is 0.32mm, which meets the set requirements. Save the parameter definition file of the parametric human body model in json format and the mesh model file in obj format. This human body mesh model is used when fitting clothes.

[0101] (3) Based on the clothing thickness information recorded in step (1) and the preset distance between clothing and the human body, the parameterized human body model aligned with the three-dimensional scanned human body mesh model is scaled and adjusted as a whole by the surface offset distance, and the parameterized human body model data before and after the adjustment is saved; in the present invention, the preset distance between clothing and the human body is set to 2.5 mm, and the clothing thickness during human body scanning is 1 mm. The optimized parameterized human body model is expanded outward by 1.5 mm as a whole, and the expanded human body model is used for measurement;

[0102] (4) Calling parameterized human joint points, surface partition dictionaries and measurement feature point definitions, and applying various measurement algorithms to automatically measure key dimensions of the human body, including length, circumference, distance and angle categories;

[0103] Parameterized human body model joints such as Figure 2As shown, it includes B01 head, B02 neck, B03 right shoulder, B04 right elbow, B05 right wrist, B06 left shoulder, B07 left elbow, B08 left wrist, B09 right clavicle, B10 left clavicle, B11 spine 3, B12 spine 2, B13 spine 1, B14 pelvis, B15 right hip, B16 right knee, B17 right ankle, B18 left hip, B19 left knee and B20 left ankle.

[0104] The human body model surface partitioning used is as follows Figure 3 As shown, the human body surface is divided into Z01 head and neck, Z02 right upper arm, Z03 right forearm, Z04 right hand, Z05 left upper arm, Z06 left forearm, Z07 left hand, Z08 chest, Z09 back, Z10 abdomen, Z11 waist and hip, Z12 right thigh, Z13 right calf, Z14 right foot, Z15 left thigh, Z16 left calf, and Z17 left foot. The names of the corresponding locations and the involved triangular faces constitute the human body surface partitions.

[0105] The definition of the measurement feature points used is as follows Figure 4 The names of the feature points with corresponding numbers and the settings of the human body vertex indexes in the specific implementation are shown in Table 1 below. Among them, some feature points can be obtained by indirect calculation of other feature points, such as M20 and M21, and the calculation method is the average value of the coordinates of the relevant feature points.

[0106] Table 1 Definition table of measurement feature points in specific implementation

[0107] serial number name Vertex Index serial number name Vertex Index M01 Front neck point 3171 M13 Left axillary point 1418 M02 Right neck point 4306 M14 Attacking midfielder 3501 M03 Left neck point 818 M15 Right waist point 4164 M04 Back of neck point 829 M16 Left waist point 676 M05 Right shoulder point 5325 M17 midpoint of the back 3027 M06 Right front axillary point 4092 M18 Right posterior axillary point 4202 M07 Left anterior axillary point 714 M19 Midfielder 3022 M08 Right chest point 6489 M20 mid-chest point M08,M09 M09 Left chest point 3043 M21 mid-waist M14, M19 M10 Chest point 4154 M22 Xiushan High Point 4917 M11 Chest point 4101 M23 ulnar protuberance 4669 M12 Right axillary point 4766

[0108] The definition of the preset length dimension measurement items is shown in Table 2. The Euclidean distance between two feature points in three-dimensional space is calculated and used as the length dimension of the current measurement item.

[0109] Table 2 Definition of length dimension measurement items in specific implementation manner

[0110] serial number name Feature point 1 Feature point 2 L01 front middle upper M01 M20 L02 Front, Middle, and Lower M20 M14 L03 Neck width M02 M03 L04 Side chest width M12 M13 L05 Side waist width M15 M16 L06 chest distance M20 M08

[0111] The preset girth measurement item definitions are shown in Table 3. First, a specific cutting plane is determined, and the origin of the cutting plane is set to the spatial coordinates of feature point 1. The normal direction of the cutting plane is determined by the vector formed by two joint points, where joint point 1 is the starting point of the vector and joint point 2 is the end point of the vector. The parameterized human body model is intercepted based on the above-mentioned cutting plane. In view of the fact that changes in human body shape and movement may lead to the existence of multiple cross-sections in a local area, the obtained cross-sections are screened according to the human body partition dictionary to determine the only valid cutting section. A three-dimensional convex hull processing is performed on the boundary of the valid cutting section to generate a three-dimensional convex hull cutting section, and the sum of all side lengths of the three-dimensional convex hull cutting section is further calculated. Since the convex hull edge intercepted by partitioning will automatically close, it is necessary to subtract the length of the closed edge from the sum of all side lengths to use this as the girth size of the current measurement item.

[0112] Table 3 Definition of girth measurement items in a specific embodiment

[0113] serial number name Feature point 1 Joint 1 Joint 2 Body Partitions Closed edge C01 Front chest circumference M20 B14 B11 Z08 L04 C02 Back bust M20 B14 B11 Z09 L04 C03 Front waist M21 B14 B11 Z10 L05 C04 Back waist M21 B14 B11 Z11 L05

[0114] The definition of the preset distance dimension measurement items is shown in Table 4. The geodesic distance between two feature points on the surface of the parametric human body model is calculated and used as the distance dimension of the current measurement item.

[0115] Table 4 Definition table of distance dimension measurement items in a specific embodiment

[0116] serial number name Feature point 1 Feature point 2 D01 Chest point-side neck point M08 M02 D02 Chest point-shoulder point M08 M05 D03 shoulder width M02 M05 D04 Back middle upper M04 M17 D05 Back middle and lower M17 M19 D06 Back width M17 M18 D07 Back axillary point-side neck point M18 M02 D08 Back armpit point-shoulder point M18 M05 D09 Xiushan Gao M05 M22 D10 Sleeve Length M05 M23 D11 chest width M06 M07

[0117] Table 5 shows the predefined angular dimension measurement items. Feature point 1 is selected as the vertex of the angle measurement and set as the starting point of the vector. Feature points 2 and 3 are further selected as the end points of the vectors, and two measurement angle vectors are constructed. Based on these two measurement angle vectors, the included angle is calculated in three-dimensional space and used as the angular dimension of the current measurement item.

[0118] Table 5 Definition of angle dimension measurement items in a specific embodiment

[0119] serial number name Feature point 1 Feature point 2 Feature point 3 A01 Upper chest corner M08 M10 M06 A02 thoracic angle M08 M06 M11

[0120] In the specific implementation, it involves the generation of coat prototype and sleeve prototype, and the measurement items involved are shown in Figure 5 The measurement results are shown in Table 6.

[0121] Table 6 Key human body dimension measurements for specific embodiments

[0122] name Measurements name Measurements name Measurements L01 16.44cm C03 41.99cm D07 22.52cm L02 20.84cm C04 43.71cm D08 19.72cm L03 15.22cm D01 25.48cm D09 12.44cm L04 32.88cm D02 24.56cm D10 54.61cm L05 30.61cm D03 8.35cm D11 31.22cm L06 9.34cm D04 18.26cm A01 33.17° C01 52.56cm D05 24.13cm A02 37.23° C02 46.61cm D06 16.70cm

[0123] (5) Use the automated pattern generation system according to the size and setting rules to automatically adjust the size of the clothing prototype and the position of the darts, generate and save personalized clothing prototype patterns and parameter setting files, including the prototype of the top and the prototype of the sleeves; the parameter configuration file includes the default values and value ranges of the component style and component size. Among them, the default value of the style is the selected style name, and the value range is the names of the alternative styles; the default value of the size is based on the measured size and is set to 1, and the value range is a specific multiple of the default size, such as 0.8~1.2, which is 0.8 to 1.2 times the default size. Users can select a specific component name or size multiple in the component style library and size value range to change the style design of the generated pattern according to style needs.

[0124] like Figures 6-9 As shown, the drawing steps for the prototype of the top and sleeves are as follows:

[0125] (5.1) Draw the frame line of the front half;

[0126] Width of the front half frame line Based on L02 and L06, draw the chest line and mark the chest point M08. The chest point is located on the chest line, and the horizontal distance from the front center line is L06. Determine the height of the front half frame. Based on D01 and D02, determine the shape of the shoulder slope line. Extend the shoulder slope line to the front side base line FSL to form the shoulder slope line of the front half.

[0127] (5.2) Draw the frame line of the second half of the film;

[0128] The width of the frame line of the second half is According to D05 and D06, determine the position of the rear axillary point M18 in the template. Determine the height of the back half frame. Based on D07 and D08, determine the shape of the back half shoulder slope line and extend it to the back side base line BSL to form the back half shoulder slope line.

[0129] (5.3) Front bust dart design;

[0130] The bust dart size is calculated based on the bust dart angle and bust dart width. Among them, the bust dart angle ACD=(90°-A01-A02), the bust dart width Due to the addition of the chest dart, the basic side seam is tilted towards the center, with an angle of The offset distance between the midpoint of the bust dart and the lower end point of the inclined side line is

[0131] (5.4) Waist dart design on the front half;

[0132] Adjust the waistline of the front piece according to the front bust-waist difference FD. When C01 > C03, distribute the waist darts according to Table 7. Waist dart 1 is located at the vertical projection position of the bust point, and waist dart 2 is located at the midpoint between the vertical projection of the bust point and the side line SL. The height of waist darts 1 and 2 is 0.8 * L02. The side dart is formed by offsetting the lower end point of the side line towards the lower middle front direction. When C01 < C03, no waist darts are added, but the side line and the lower middle front line are each extended outwards

[0133] (5.5) Design of the back piece waist darts;

[0134] Adjust the waistline of the back piece according to the back bust-waist difference BD. When C02 > C04, distribute the waist darts according to Table 7. Waist dart 1 is located at the projection position of the back underarm point, and waist dart 2 is located at the midpoint between the projection of the back underarm point and the lower middle back. The height of waist darts 1 and 2 is L02. The side dart is formed by offsetting the lower end point of the side line towards the lower middle back direction. When C02 < C04, no waist darts are added, but the side line and the lower middle front line are each extended outwards

[0135] Table 7 Waist Dart Allocation Rules for Upper Body Prototype

[0136] Chest-waist difference (cm) Waist dart 1 Waist dart 2 Side darts Greater than 0 and less than 1 / / 100％ Greater than 1 and less than 5 / 80％ 20％ Greater than 5 40％ 40％ 20％

[0137] (5.6) Add the armhole and neckline arcs;

[0138] Add the armhole and neckline arcs in the armhole and neck areas. Generate the armhole and neckline arcs using an optimization method based on the end points and their tangent directions. The specific generation constraint rules are shown in Table 8. The schematic diagram of the tangent directions of each end point is shown in Figure 8 the direction of the black arrow in

[0139] Table 8 Optimization Generation Constraint Rules for Armhole and Neckline Arcs

[0140] name Upper endpoint Upper endpoint tangent direction Lower endpoint Lower endpoint tangent direction Front neckline arc Side neck point (0,1) Front neck point (-1,0) Front armhole curve shoulder point Shoulder point → Chest point Upper end point of bust dart Upper end point of bust dart → bust point Back neckline arc Side neck point (0,1) Back of neck point (1,0) Back armhole curve shoulder point Shoulder point → back armpit point Upper end point of side seam (-1,0)

[0141] (5.7) The drafting steps of the sleeve prototype are as follows:

[0142] Draw the sleeve frame line with a height of D10. Generate the sleeve cap arc and the sleeve width end points using an optimization method based on D09, the length of the armhole arc, and the tangent direction of the sleeve cap arc. The specific generation constraint rules are shown in Table 9. The schematic diagram of the tangent directions of each end point is shown in Figure 9 the direction of the black arrow in

[0143] Table 9 Optimization Generation Constraint Rules for Sleeve Cap Arc

[0144]

[0145] (6) According to the clothing style design requirements, the user selects a loose V-neck T-shirt, then the top prototype dart definition is deleted in the parameter configuration file, the collar type default value is selected as a V-neck, and the sleeve length multiple default value is adjusted to 0.3. The template generation system is used to generate a personalized style template and its parameter configuration file;

[0146] (7) Perform virtual fitting on the sample generated in step (6); the user can adjust the sample parameter setting file multiple times according to the fitting effect and perform virtual fitting until the desired effect is achieved. Generate and save the sample file in dxf format that meets the AAMA specification, the clothing sample image file in png format, the virtual fitting 3D model file in obj format, the virtual fitting image file in png format, and the sample parameter configuration file in json format.

[0147] The comparison between the prototype and the final personalized clothing sample is shown in the figure below. Figure 10 As shown in the figure, the dotted lines are the prototypes of the top and sleeves generated in step (4), and the solid lines are the samples generated after the style is adjusted. Figure 11 As shown, Figure 11 (a) is the fitting result of the top prototype and sleeve prototype. Figure 11 (b) is a schematic diagram of the personalized clothing sample finally generated after adjustment.

[0148] In order to verify the accuracy and effectiveness of the method provided by the present invention, the method of the present invention is compared and verified with the existing technology from two aspects: the accuracy of human body measurement and the effect of clothing pattern generation.

[0149] (1) Comparison of anthropometric accuracy;

[0150] The definition of accuracy in the national standard GB / T 23698-2023 refers to the degree to which a measurement result deviates from the true value. In this comparative experiment, the true value refers to the average of multiple measurements taken by anthropometric experts using traditional measuring instruments (such as tape measures and calipers). The automated anthropometric measurements obtained using the present invention and existing technologies were compared with the true value, and their relative errors were calculated. Considering that existing automated measurement technologies typically support fewer measurement items than the present invention, this experiment used the production of a prototype top as an example. With the subjects wearing close-fitting clothing, seven key measurement items that can be measured by both technologies (chest circumference, waist circumference, back length, front chest width, chest point distance, shoulder length, and chest point-side neck point distance) were compared. The subjects' anthropometric dimensions were measured in a standard standing posture and in a running position (referred to as standing and running positions), and the relative errors were calculated and recorded. The specific data are shown in Table 10.

[0151] Table 10 Comparison of relative measurement errors between the prior art and the method of the present invention under different human postures

[0152]

[0153] The data in Table 10 show that under standard standing posture, the relative measurement error of the method of the present invention is generally lower than that of the existing technology. The errors of both methods under these static conditions are within the range permitted by national standards. While the relative measurement errors of both methods increase slightly compared to those under static conditions during running, the relative error of the method of the present invention is significantly lower than that of the existing technology. This is particularly true for key garment structural dimensions such as chest circumference, waist circumference, and back length, demonstrating superior measurement stability and accuracy.

[0154] In summary, the experimental results show that the anthropometry method provided by the present invention, especially in dynamic anthropometry, has higher accuracy than the existing technology.

[0155] (2) Comparison of clothing sample generation effects;

[0156] To evaluate the fit of samples generated by different sample generation rules, this section uses a virtual try-on approach for qualitative comparison. The comparison targets include: GarmentCode's built-in sample generation rules, the existing commonly used chest measurement sample generation rules (New Culture Prototype Rules), and the sample generation rules of this invention.

[0157] The experimental steps are as follows: for a specific 3D virtual human model, the three rules mentioned above are applied to generate a basic top sample. Then, in the professional clothing simulation software CLO3D, the same fabric properties, stitching relationships and simulation parameters are set, and the generated sample is virtually tried on. By visualizing the clothing deformation map and observing the static shape of the clothing on the virtual human model (such as Figures 12-14 As shown in the figure, the red area usually indicates stress concentration or large deformation, indicating that there may be problems with the fit of the area), and combined with the cross-sectional view of the waist (such as Figure 15 As shown) to comprehensively evaluate the fit and balance of the sample.

[0158] The pattern generated based on the new cultural prototype rule mainly relies on a few parameters such as chest circumference and back length to push the pattern. For the specific human body model used in the specific implementation, the generated clothing appears too loose. Although the deformation of the clothing surface is not large, the overall fit is insufficient. Figure 15 As can be seen from the waist cross-section shown in (a), the gap between the garment and the body is unevenly distributed in the front and back, indicating that there is a problem with the balance of the garment.

[0159] The pattern generated using GarmentCode's built-in rules showed improved fit compared to the New Culture prototype. However, the virtual try-on still revealed a noticeable rise in the back waistline compared to the front waistline, and unwanted lift at the top of the sleeves, suggesting a lack of structural balance. Furthermore, the garment deformation visualization showed significant deformation in areas such as the neckline, shoulders, and armholes.

[0160] Compared to the two existing technologies mentioned above, the patterns generated using the present invention exhibit fewer surface wrinkles after virtual try-on, and the garment shape better fits the three-dimensional contours of the human body. The waistline remains essentially level front and back, and the overall garment body is better balanced.

[0161] Comprehensive comparative experimental results show that the pattern generation method provided by the present invention can generate clothing patterns with a higher degree of matching with the target human body and a more reasonable distribution of darts, thereby achieving better clothing fit, appearance and wearing comfort.

Claims

1. A method for human body measurement and personalized clothing pattern generation for multiple body types, characterized by: The specific steps include: Step 1: Use 3D body scanning equipment to collect high-precision body data of the user, and use data processing software to pre-process the collected data. At the same time, record the thickness of the user's clothing during scanning, and export it to obtain a 3D scanned body mesh model; Step 2: Select a parametric human body model and align it with the 3D scanned human mesh model from Step 1. The selected parametric human body model has virtual joints, and its shape and movement are controlled by a set of parameters. After pre-training, the relationship between human morphology and posture changes is established. Step 3: Based on the clothing thickness information recorded in step 1 and the preset distance between the clothing and the human body, the parametric human body model that has been registered with the 3D scanned human body mesh model is scaled and adjusted as a whole, and the parametric human body model data before and after the adjustment is saved; Step 4: Call the joint points, surface partition dictionary, and measurement feature point definitions of the parametric human body model scaled and adjusted in step 3, apply various measurement algorithms, and automatically measure and record various key dimensions of the human body, including length, circumference, distance, and angle; Step 5: After verifying the integrity of the size information, the size data obtained in step 4 is used as the clothing prototype constraint. The pattern generation system automatically adjusts the size and dart position during the pattern generation process according to the clothing prototype constraint, generates and saves the personalized clothing prototype pattern and its parameter configuration file; Step 6: Based on the clothing style design requirements, the personalized clothing prototype template and its parameter configuration file obtained in step 5 are customized and adjusted, and a personalized style template and its parameter configuration file are generated using the template generation system; Step 7: Perform a 3D virtual fitting on the personalized style sample generated in step 6, repeatedly adjust the sample design based on the fitting effect until the desired effect is achieved, and finally save the personalized clothing sample and its related data files; In step 4, the surface partition dictionary consists of the names of the parts of the scaled parameterized human body model and the corresponding face indices; the measured feature points are the vertices of the scaled parameterized human body model surface, or the average spatial coordinates of two surface vertices.

2. The method for generating human body measurements and personalized clothing patterns for multiple body types according to claim 1, characterized in that: The scanning accuracy error of the 3D human body scanning device in step 1 is within ±5mm.

3. The method for generating human body measurements and personalized clothing patterns for multiple body types according to claim 1, characterized in that: The preprocessing in step 1 includes mesh denoising, defect repair and retopology processing.

4. The method for generating human body measurements and personalized clothing patterns for multiple body types according to claim 1, characterized in that: The parametric human body model selected in step 2 is the SMPL parametric human body model.

5. The method for generating human body measurements and personalized clothing patterns for multiple body types according to claim 1, characterized in that: After the registration is completed in step 2, the chamfer distance between the parametric human body model and the 3D scanned human body mesh model is less than 5 mm, except for the head area. The calculation formula of the chamfer distance is: Among them, D chamfer (A, B) is the chamfer distance between A and B, A and B are the vertex sets of the parameterized human body model and the 3D scanned human body mesh model, respectively. a∈A and b∈B represent individual vertices of the parameterized human body model and the 3D scanned human body mesh model, respectively. ‖ab‖ is the Euclidean distance from point a to point b. |A| and |B| are the number of vertices in A and B, respectively.

6. The method for generating human body measurements and personalized clothing patterns for multiple body types according to claim 1, characterized in that: The overall scaling of the parametric human body model registered with the 3D scanned human body mesh model in step 3 is controlled by the surface offset distance. The algebraic expression of the surface offset distance between the scaled model and the original model is: D scale =D gap -D cloth ; Among them, D scale D is the surface offset distance between the scaled model and the original model. gap is the preset distance between clothing and human body, D cloth is the clothing thickness recorded in step 1.

7. The method for generating human body measurements and personalized clothing patterns for multiple body types according to claim 1, characterized in that: Verifying the integrity of the size information in step 5 refers to checking the integrity of the human body size measurement data to ensure that it meets the input conditions required for generating personalized clothing prototype samples.

8. The method for generating human body measurements and personalized clothing patterns for various body types according to claim 1, characterized in that: In step 7, a virtual sewing simulation operation is performed according to the spatial arrangement of the personalized style pattern and the definition of its connection relationship to achieve a virtual fitting effect.

Images