Garment size conversion method and device based on 3D scanning, medium and product

Abstract

The invention provides a ready-made garment size conversion method and device based on 3D scanning, a medium and a product, and the method comprises the following steps: obtaining three-dimensional scanning data of a target object in a current dressing state, and carrying out the feature recognition to determine garment category information; calling a benchmark reference sample clothes model corresponding to the category information, wherein the model comprises a benchmark structure relaxation amount, a fabric thickness and a fabric elasticity coefficient; human body feature points are positioned based on the three-dimensional scanning data, geometric measurement is calculated, and garment contour size data are extracted; performing reverse reduction on the overall size of the ready-made garment by combining the sample garment model parameters to obtain the net body size of the human body; and obtaining a forward garment model of the target customized category, determining a target relaxation amount according to a model addition rule, and superposing the target relaxation amount on the net body size to generate a target garment production size. By implementing the technical scheme provided by the invention, intelligent and automatic calculation of the customized garment size is realized, and the size conversion precision and the production efficiency are improved.

Description

Technical Field

[0001] This application relates to the fields of clothing customization and data processing technology, and in particular to a method, device, medium and product for converting garment sizes based on 3D scanning. Background Technology

[0002] With the rapid development of new retail models and artificial intelligence technologies, offline unmanned clothing stores and personalized clothing customization services are gradually becoming industry trends, and consumers' demands for convenience and customization in clothing purchases continue to rise. How to quickly and contactlessly collect consumers' body size data in unmanned self-service scenarios, and achieve efficient and accurate conversion from body scan data to target garment sizes, has become a core requirement for improving the unmanned retail experience and realizing intelligent clothing customization.

[0003] In existing technologies, human body size data is typically acquired through manual measurement or visual estimation using 3D scanning technology. Manual measurement is heavily influenced by the individual experience and operating habits of the measurement personnel, making data standardization difficult and resulting in low efficiency. Existing automated technologies often employ ergonomic structures or general spatial algorithms to directly calculate garment sizes from a scanned model of the human body wearing clothing. However, this method of direct estimation based on external contours often fails to adequately consider the specific impacts of fabric thickness, wrinkle accumulation, and the ease allowance of different garment patterns on the scanned data. Because it fails to effectively distinguish the physical boundary between the garment's outer contour and the human body, the calculated body size often deviates from the true value when faced with clothing of varying thickness and elasticity. This poses a risk of insufficient accuracy in garment size customization due to the inability to accurately eliminate interference from clothing structure and physical properties. Summary of the Invention

[0004] In view of this, this application provides a method, device, medium, and product for garment size conversion based on 3D scanning to solve the above problems.

[0005] Firstly, a method for converting garment sizes based on 3D scanning is provided, the method comprising:

[0006] Acquire 3D scan data of the target object in its current clothing state, perform feature recognition on the 3D scan data, and determine the clothing category information of the target object's current clothing.

[0007] The reference garment model corresponding to the garment category information is retrieved from the preset reference garment database. The reference garment model stores the reference structure ease value, fabric thickness value and fabric elasticity coefficient.

[0008] Based on 3D scanning data, the human body feature points of the target object are located, and geometric measurement calculations are performed based on the human body feature points to extract the garment outline size data.

[0009] Based on the baseline structural relaxation value, fabric thickness value and fabric elasticity coefficient of the reference sample garment model, the garment outline size data is reverse-calculated to obtain the net body size data of the target object.

[0010] Obtain the forward ready-to-wear model corresponding to the target customized product category. The forward ready-to-wear model is configured with the pattern addition rules for the target customized product category.

[0011] Input the human body net size data into the forward garment model, use the pattern addition rules to determine the target ease amount, and superimpose the target ease amount on the human body net size data to generate the target garment production size for the target customized product category.

[0012] The above technical solution, by acquiring 3D scan data of the target object in its dressed state and combining it with the ease allowance, fabric thickness, and elasticity parameters of the benchmark garment model for reverse reconstruction, can accurately calculate the true net body size from the garment size in its dressed state, reducing measurement errors caused by differences in clothing thickness and elasticity. Then, by utilizing the forward garment model and pattern allowance rules to generate the target garment production size, automatic conversion from scanning to customized production size can be achieved, thereby improving the accuracy and efficiency of customized clothing size generation to a certain extent.

[0013] Optionally, feature recognition can be performed on the 3D scan data to determine the clothing category information of the target object's current attire, specifically including:

[0014] Based on the preset grid resolution, the spatial coordinates of the 3D scan data are discretized and quantized to construct a 3D voxel index space.

[0015] Traverse the non-empty voxels in the 3D voxel index space, and construct a local covariance matrix based on the spatial coordinate distribution of the non-empty voxels and the point cloud data in their neighborhoods.

[0016] Perform eigenvalue decomposition on the local covariance matrix to obtain the set of eigenvalues ​​and the corresponding set of eigenvectors;

[0017] The minimum eigenvalue is determined from the set of eigenvalues, and the eigenvector corresponding to the minimum eigenvalue is determined as the surface normal vector representing the local plane direction;

[0018] The surface normal vector is mapped to the mesh position corresponding to the non-empty voxel to generate a three-dimensional surface normal feature tensor.

[0019] A deep residual network is used to perform 3D convolution operations on the 3D surface normal feature tensor to extract morphological feature vectors.

[0020] The morphological feature vector is matched with the reference feature vector in the preset category feature library using Euclidean distance, and the category label corresponding to the reference feature vector with the smallest distance is determined as the clothing category information.

[0021] The above technical solution extracts 3D surface normal features by voxelizing and decomposing local covariance features from 3D scan data and inputting them into a deep residual network for convolutional feature recognition. This enables the identification of the type of clothing worn by a target object in complex scanning scenarios, thereby reducing errors caused by manual judgment. This feature recognition process maintains high accuracy even under conditions of different fabric wrinkles, lighting conditions, or pose variations, providing a reliable basis for subsequently calling the correct benchmark garment model.

[0022] Optionally, based on 3D scan data, locate the human feature points of the target object, specifically including:

[0023] Perform surface differential geometry analysis on the 3D scan data to calculate the local geometric curvature values ​​of each data point in the 3D scan data;

[0024] Data points whose local geometric curvature values ​​exceed a preset gradient threshold are selected from the 3D scan data, and the set of selected data points is determined as a subset of the candidate feature point cloud;

[0025] Based on prior knowledge of human topology, spatial clustering and filtering of candidate feature point cloud subsets are performed to identify rigid anatomical feature regions that respectively represent the acromion position and the cervical spine position.

[0026] Call the preset parametric human skeleton template, which has pre-calibrated bone anchor points corresponding to rigid anatomical feature areas;

[0027] Construct an objective function that minimizes distance error, calculate the optimal rigid transformation matrix that maps skeletal anchor points to rigid anatomical feature regions, and use the optimal rigid transformation matrix to lock the spatial pose of the parameterized human skeletal template.

[0028] Based on spatial posture, the relative positional mapping relationship between bones and soft tissues is used to deduce and locate human feature points, including the chest height point and waist node.

[0029] The above technical solution, through curvature filtering combined with prior knowledge of human topology, automatically identifies rigid anatomical feature areas such as the acromion and cervical spine from 3D scanned point clouds, and uses skeletal template matching to achieve spatial posture calibration. This allows for accurate positioning of key human feature points even without markers. This process reduces the need for manual annotation or multiple scans, helps improve the spatial positioning stability and consistency of measurement benchmarks such as the chest height point and lumbar node, thereby improving the accuracy of human geometric measurements.

[0030] Optionally, geometric measurements can be performed based on human feature points to extract garment outline size data, specifically including:

[0031] Based on a parametric human skeleton template, the central axis of the human torso is determined, and multiple measurement sections perpendicular to the central axis of the human torso are constructed.

[0032] Acquire the cross-sectional point cloud data of the 3D scan data on the measurement section, and perform convex hull fitting on the cross-sectional point cloud data to generate a fitted closed curve that removes noise from clothing wrinkles.

[0033] Calculate the perimeter of the fitted closed curve and use the perimeter as the circumference data in the garment outline size data.

[0034] The above technical solution, by determining the central axis of the torso based on a human skeletal template and establishing multiple measurement cross-sections in its vertical direction, effectively filters out abnormal point clouds caused by clothing wrinkles by performing cross-sectional convex hull fitting on the scanned data. The calculated cross-sectional perimeter can accurately reflect the clothing outline dimensions, thereby improving the stability and noise resistance of the circumference data while ensuring automated measurement.

[0035] Optionally, based on the baseline structural relaxation value, fabric thickness value, and fabric elastic coefficient of the reference garment model, the garment outline size data is reverse-calculated to obtain the net body size data of the target object, specifically including:

[0036] Based on the vertical height distribution of garment outline dimensions on the human body, a dynamic gap distribution coefficient that varies with height is generated using a preset gravity drape model.

[0037] Multiply the relaxation amount of the reference structure by the dynamic gap distribution coefficient to obtain the actual gap amount after gravity correction.

[0038] The girth increment caused by the physical thickness of the fabric is calculated based on the fabric thickness value, and the lateral deformation correction coefficient is determined based on the fabric elasticity coefficient.

[0039] The garment outline size data is differentially subtracted by using the girth increment value and the actual gap amount, and the subtracted garment outline size data is numerically scaled by using the lateral deformation correction coefficient to obtain the net body size data.

[0040] The above technical solution establishes a dynamic gap distribution coefficient that varies with human body height by introducing a gravity-based sag model. Combined with fabric thickness and elasticity corrections, it performs differential subtraction on the garment's silhouette dimensions. This effectively eliminates the effects of sag and deformation caused by gravity, ensuring that the reconstructed net body dimensions match the actual human body surface contour height. This method improves the accuracy of human body size inversion and guarantees consistency in calculation results across different fabrics and garment structures.

[0041] Optionally, the target relaxation amount can be determined using pattern-based ease rules, specifically including:

[0042] Obtain the stiffness and drape parameters of the target fabric for the target customized product category;

[0043] Query the preset basic ease table in the pattern addition rules to obtain the basic ease value corresponding to the level of the human body net size data;

[0044] An ease correction factor is constructed using the target fabric stiffness parameter and the target fabric drape parameter. The ease correction factor is then used to weight and adjust the basic ease value to determine the target ease amount.

[0045] The above technical solution, by modifying the basic ease allowance scale using fabric stiffness and drape parameters, can dynamically adjust the ease allowance according to different fabric characteristics, ensuring that the garment is neither too tight nor too loose after the ease allowance is added. This method directly links the ease allowance calculation to the physical properties of the fabric, thereby improving the precision control of the comfort and fit of customized clothing.

[0046] Optionally, the target ease allowance is overlaid on the human body net size data to generate the target garment production size for the target customized product category, specifically including:

[0047] The target ease amount is allocated to each circumference corresponding to the human body net size data to obtain the target garment specification value;

[0048] Using a forward garment model, the target garment specification values ​​are mapped to the coordinate parameters of two-dimensional pattern control points;

[0049] Based on coordinate parameters, industrial production pattern data containing seam allowance and garment shrinkage rate is generated, and the industrial production pattern data is used as the target garment production size.

[0050] The above technical solution distributes the target relaxation amount to various circumference areas, automatically generates two-dimensional pattern control points using a forward garment model, and then incorporates seam allowance and garment shrinkage to directly generate industrial-grade pattern data. This process reduces manual pattern-making errors and repetitive adjustments, helping to improve the consistency of pattern generation and production efficiency.

[0051] In a second aspect, an electronic device is provided, including a processor, a memory, a user interface, and a network interface, wherein the memory is used to store instructions, the user interface and the network interface are both used to communicate with other devices, and the processor is used to execute the instructions stored in the memory to cause the electronic device to perform the method as described in any of the above.

[0052] Thirdly, a computer-readable storage medium is provided that stores instructions which, when executed, perform the method as described in any of the preceding descriptions.

[0053] Fourthly, a computer program product containing instructions is provided, which, when run on a server, causes the server to perform the method described in the first aspect and any possible implementation thereof.

[0054] Understandably, the electronic device provided in the second aspect, the computer-readable storage medium provided in the third aspect, and the computer program product provided in the fourth aspect are all used to execute the method provided in this application. Therefore, the beneficial effects they can achieve can be referred to the beneficial effects in the corresponding methods, and will not be repeated here.

[0055] In summary, implementing one or more technical solutions provided in this application has at least the following technical effects or advantages:

[0056] By combining 3D human body scanning technology with parametric modeling of garment structure, a fully automated conversion process is constructed, from "scanning data in clothing state" to "net body size inversion" and then to "garment production size generation," achieving integrated processing of human body measurement, size calculation, and pattern generation. This process eliminates the need for undressing for scanning or manual measurement, simplifying the measurement process and improving the convenience of measurement operations. By introducing a benchmark sample garment model and fabric physical property parameters, the size calculation results can be adaptively corrected under different garment types and fabric conditions, making size conversion universal and stable across garment categories. Simultaneously, this technical solution achieves a unified digital expression of human body data, fabric parameters, and pattern-making rules, enabling direct integration with intelligent pattern-making systems and digital production lines. It provides data interface support for personalized customization, virtual fitting, and online garment retail, thereby reducing the workload of manual pattern making and sample garment trial production to a certain extent. Therefore, the technical solution of this application has a high degree of automation and can be applied to garment customization scenarios. Attached Figure Description

[0057] Figure 1 This is an exemplary system architecture diagram of a garment size conversion method based on 3D scanning disclosed in this application;

[0058] Figure 2 This is a flowchart illustrating a garment size conversion method based on 3D scanning disclosed in this application;

[0059] Figure 3 This is a schematic diagram of the structure of an electronic device disclosed in this application.

[0060] Explanation of reference numerals in the attached figures: 100, System architecture; 101, First terminal device; 102, Second terminal device; 103, Third terminal device; 104, Network; 105, Server; 301, Processor; 302, Communication bus; 303, User interface; 304, Network interface; 305, Memory. Detailed Implementation

[0061] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0062] In the description of the embodiments of this application, the words "for example" or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design that is described as "for example" or "for instance" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design options. Rather, the use of the words "for example" or "for instance" is intended to present the relevant concepts in a specific manner.

[0063] In the description of the embodiments of this application, the term "multiple" means two or more. For example, multiple systems means two or more systems, and multiple screen terminals means two or more screen terminals. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the indicated technical features. Thus, a feature defined with "first" or "second" may explicitly or implicitly include one or more of that feature. The terms "comprising," "including," "having," and variations thereof all mean "including but not limited to," unless otherwise specifically emphasized.

[0064] Figure 1 An exemplary system architecture diagram is shown, illustrating an embodiment of a 3D scanning-based garment size conversion method applicable to this application.

[0065] like Figure 1 As shown, the system architecture 100 may include a first terminal device 101, a second terminal device 102, a third terminal device 103, a network 104, and a server 105. The network 104 is used as a medium to provide communication links between the terminal devices 101, 102, 103, and the server 105. The network 104 may include various connection types, such as wired or wireless communication links or fiber optic cables, etc.

[0066] Users can use terminal devices 101, 102, and 103 to interact with server 105 via network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, and 103, such as model training applications, video recognition applications, web browser applications, social platform software, etc.

[0067] Terminal devices 101, 102, and 103 can be either hardware or software. When terminal devices 101, 102, and 103 are hardware, they can be various electronic devices with displays, including but not limited to smartphones, tablets, e-book readers, MP3 (Moving Picture Experts Group Audio Layer III) players, MP4 (Moving Picture Experts Group Audio Layer IV) players, laptops, and desktop computers, etc. When terminal devices 101, 102, and 103 are software, they can be installed in the aforementioned electronic devices. They can be implemented as multiple software programs or software modules (e.g., multiple software programs or software modules used to provide distributed services) or as a single software program or software module. No specific limitations are imposed here.

[0068] When terminals 101, 102, and 103 are hardware devices, video capture devices can also be installed on them. These video capture devices can be various devices capable of capturing video, such as cameras, sensors, etc. Users can use the video capture devices on terminals 101, 102, and 103 to capture video.

[0069] Server 105 can be a server that provides various services, such as a backend server for processing data displayed on terminal devices 101, 102, and 103. The backend server can analyze and process the received data and can feed back the processing results (such as recognition results) to the terminal devices.

[0070] It should be noted that a server can be either hardware or software. When the server is hardware, it can be implemented as a distributed server cluster consisting of multiple servers, or as a single server. When the server is software, it can be implemented as multiple software programs or software modules (e.g., multiple software programs or software modules used to provide distributed services), or as a single software program or software module. No specific limitations are made here.

[0071] It should be understood that Figure 1The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included. In particular, if the target data does not need to be obtained remotely, the above system architecture may exclude the network and include only terminal devices or servers.

[0072] Figure 2 This is a flowchart illustrating a garment size conversion method based on 3D scanning, as described in an embodiment of this application. This method can be implemented using a computer program or a microcontroller. The computer program can be integrated into an application or run as a standalone utility application. The specific steps of a garment size conversion method based on 3D scanning are described in detail below.

[0073] S201: Obtain 3D scan data of the target object in its current clothing state, perform feature recognition on the 3D scan data, and determine the clothing category information of the target object's current clothing.

[0074] For example, the system acquires real-time point cloud data of the target object's appearance using a connected 3D scanning device (such as an infrared structured light scanner or LiDAR). This process not only obtains the spatial coordinates of the human body surface but also captures the overall contour and local wrinkle details of the clothing. The system performs feature extraction and topological analysis on this unstructured geometric data to identify the style characteristics of the current clothing from a macroscopic geometric perspective and automatically categorizes it into a specific garment category (e.g., "hooded sweatshirt" or "business shirt"), thus providing an accurate index for subsequent calls to the corresponding benchmark garment model.

[0075] In one possible implementation, feature recognition is performed on the 3D scan data to determine the clothing category information of the target object's current attire. Specifically, this includes: discretizing and quantizing the spatial coordinates of the 3D scan data based on a preset grid resolution to construct a 3D voxel index space; traversing the non-empty voxels in the 3D voxel index space, and constructing a local covariance matrix based on the spatial coordinate distribution of the non-empty voxels and their neighborhood point cloud data; performing eigenvalue decomposition on the local covariance matrix to obtain a set of eigenvalues ​​and a corresponding set of eigenvectors; determining the minimum eigenvalue from the eigenvalue set, and identifying the eigenvector corresponding to the minimum eigenvalue as the surface normal vector representing the local planar direction; mapping the surface normal vector to the grid position corresponding to the non-empty voxel to generate a 3D surface normal feature tensor; performing a 3D convolution operation on the 3D surface normal feature tensor using a deep residual network to extract morphological feature vectors; matching the morphological feature vectors with reference feature vectors in a preset category feature library using Euclidean distance, and determining the category label corresponding to the reference feature vector with the smallest distance as the clothing category information.

[0076] In this embodiment, the three-dimensional voxel index space refers to a gridded data structure formed by dividing a continuous three-dimensional space into a series of regularly arranged, discrete small cubic units (i.e., voxels) according to a preset resolution. This structure represents the spatial distribution and topological relationships of three-dimensional scan data. Similar to the pixel matrix in a two-dimensional image, the three-dimensional voxel index space locates each voxel unit using three-dimensional coordinate indices (e.g., x, y, z), transforming unstructured point cloud data into structured spatial data, facilitating subsequent deep learning operations such as three-dimensional convolution.

[0077] Specifically, the system acquires raw 3D scan data of the dressed state, which typically consists of high-density discrete point clouds. To process this unstructured data, based on a pre-defined grid resolution (e.g., setting the side length of each voxel to a specific value), the spatial coordinates of each point in the point cloud are discretized and quantized, mapping continuous coordinate values ​​to integer indices, thus constructing a 3D voxel index space. Based on this, the index space is traversed, processing only non-empty voxels containing data points. For each non-empty voxel, all point cloud data within its interior and adjacent voxel ranges are collected. The geometric centroid of all point cloud data in this neighborhood is calculated, and the centroid coordinates are subtracted from the coordinates of each point to obtain a decentralized vector. The cross product of the decentralized vector and its transpose is calculated, and the sum of the cross products of all points is divided by the number of points in the neighborhood, thus constructing a 3×3 symmetric local covariance matrix. The eigenvalues ​​of this matrix directly reflect the degree of dispersion of the point cloud in the three orthogonal principal directions. Eigenvalue decomposition is performed on this matrix to obtain a set of eigenvalues ​​and their corresponding eigenvectors. According to the principle of Principal Component Analysis (PCA), the eigenvector corresponding to the smallest eigenvalue represents the direction of the least data change, that is, the direction perpendicular to the local fitting plane. Therefore, it is determined as the surface normal vector characterizing the local plane direction at that location.

[0078] Furthermore, the calculated surface normal vectors at each location are remapped back to their corresponding coordinates on the 3D voxel mesh, thereby generating a 3D surface normal feature tensor with geometric orientation information. To identify the clothing type, this tensor is input into a pre-trained deep residual network (ResNet) for 3D convolution operations. The network extracts and aggregates edge, texture, and overall shape features from the image layer by layer, ultimately outputting a high-dimensional morphological feature vector. This morphological feature vector is then compared one by one with the reference feature vectors of each standard clothing category stored in a pre-defined category feature library. The Euclidean distance between them is calculated, and the label corresponding to the reference feature vector with the smallest distance (i.e., the highest similarity) is selected to determine the clothing category information of the target object's current attire.

[0079] S202: Retrieve the benchmark reference garment model corresponding to the garment category information from the preset benchmark sample garment database. The benchmark reference garment model stores the benchmark structure ease value, fabric thickness value, and fabric elastic coefficient.

[0080] In the embodiments of this application, the benchmark reference garment model refers to a pre-constructed digital three-dimensional pattern model that conforms to industrial pattern making standards and represents a specific garment category under standard wearing conditions. It is used as a geometric reference for separating the thickness and ease of clothing from three-dimensional scanning data. For example, for the "business shirt" category, the benchmark reference garment model is a three-dimensional data model of a shirt that conforms to national sizing standards and includes standard ease.

[0081] Specifically, after determining the current clothing category information of the target object, the system retrieves a pre-set benchmark sample garment database. This database employs a hierarchical classification index structure, storing standard model data covering various common clothing categories (such as suits, sweatshirts, dresses, etc.). By matching category tags with the database index, the system quickly locates and retrieves the corresponding benchmark reference garment model. This model not only contains three-dimensional geometric mesh data but also serves as a metadata carrier, storing three key physical parameters: the benchmark structural relaxation value, which is the standard amount of clearance reserved on the basis of the human body's net circumference to ensure wearing comfort and aesthetic appearance; the fabric thickness value, which is the physical thickness of the fabric in the direction perpendicular to the human body surface; and the fabric elastic coefficient, reflecting the lateral elongation rate of the fabric under stress. These parameters, along with the model, are loaded into memory to provide the necessary physical boundary conditions and correction parameters for subsequent mathematical inverse calculations to eliminate the influence of clothing occlusion and accurately restore the human body's net dimensions.

[0082] S203: Based on 3D scanning data, locate the human feature points of the target object, and perform geometric measurement calculations based on the human feature points to extract the garment outline size data.

[0083] For example, addressing the technical challenge of key anatomical points (such as the waist and chest) being completely obscured by clothing and resulting in blurred surface features when the user is dressed, this step employs a "skeleton-driven" positioning and measurement strategy. Based on human biomechanics principles, the system reverse-engineers the internal skeletal structure from the scanning data of the clothing surface, constructing a virtual skeleton precisely aligned with the current scanning posture. This virtual skeleton serves as a navigation reference to lock the coordinates of hidden human features. When extracting contour dimensions, the system simulates the effect of a tightly bound tape measure in the physical world, geometrically shaping the cut-off garment section and automatically ignoring noise caused by natural fabric accumulation or wrinkles, thereby extracting effective outer contour data that objectively reflects the actual shape of the garment.

[0084] In one possible implementation, the human feature points of the target object are located based on 3D scanning data, specifically including: performing surface differential geometry analysis on the 3D scanning data to calculate the local geometric curvature value of each data point in the 3D scanning data; selecting data points whose local geometric curvature values ​​exceed a preset gradient threshold from the 3D scanning data, and determining the set of selected data points as a subset of candidate feature point clouds; performing spatial clustering and filtering on the subset of candidate feature point clouds based on prior knowledge of human topology to identify rigid anatomical feature areas representing the acromion position and the cervical spine position respectively; calling a preset parameterized human skeleton template, which is pre-calibrated with skeletal anchor points corresponding to the rigid anatomical feature areas; constructing an objective function to minimize distance error, calculating the optimal rigid transformation matrix to map the skeletal anchor points to the rigid anatomical feature areas, and using the optimal rigid transformation matrix to lock the spatial pose of the parameterized human skeleton template; and based on the spatial pose, using the relative positional mapping relationship between bones and soft tissues to deduce and locate human feature points including the chest height point and the lumbar node.

[0085] In the embodiments of this application, the parameterized human skeleton template refers to a pre-constructed digital three-dimensional skeleton model that can change its geometric shape and spatial posture by adjusting parameters. It is used to simulate the rigid support structure inside the human body and serve as an anatomical reference benchmark for inferring the position of external soft tissues. For example, the template can be a standard human skeleton data structure that includes key bone nodes such as the cervical spine, shoulder joint, and spine, and pre-marks virtual anchor points at the corresponding anatomical positions.

[0086] Specifically, to accurately locate key points of the human body that are obscured while the person is dressed, the system performs differential geometric analysis on the surface of the acquired 3D scan data. It uses a discrete curvature estimation algorithm to traverse every vertex of the model surface, calculating local geometric curvature values ​​that reflect the degree of surface unevenness. A preset gradient threshold is set to distinguish between flat areas (such as areas where fabric naturally drapes) and areas with abrupt changes (such as areas supported by bones or hard creases). High-frequency detail data points with curvature values ​​exceeding this threshold are filtered out, and smooth background data is removed, thus forming a subset of candidate feature point clouds. Since this subset may contain pseudo-feature points generated by clothing wrinkles, it is necessary to further incorporate prior knowledge of human topology (i.e., a preset relative coordinate constraint range based on anthropometrics). Specifically, the system normalizes the scan data along the vertical axis (Z-axis) and sets vertical search intervals for each anatomical region based on standard human proportions (such as head-to-body ratio). (For example, the cervical spine feature area is limited to 80%-90% of the total human height, and the acromion feature area is limited to a slightly lower local bounding box on both sides of the cervical spine.) Using these quantified spatial constraints, spatial density clustering and logical filtering are performed on the selected point clouds to eliminate outliers that do not conform to human body structure. This allows for the accurate identification of the regions least affected by clothing thickness and looseness that best reflect the true skeletal position of the human body, namely the rigid anatomical feature areas that respectively characterize the acromion and cervical spine positions.

[0087] Furthermore, the system invokes a preset parameterized human skeleton template, on which bone anchor points corresponding to the anatomical significance of the aforementioned identified areas are pre-marked with precision. To spatially align the virtual skeleton model with the real scan data, an objective function is constructed to minimize the Euclidean distance between the virtual anchor points and the actual rigid anatomical feature areas. Rotation and translation parameters are solved using an iterative optimization algorithm (such as the Iterative Closest Point Algorithm, ICP), and the optimal rigid transformation matrix that achieves the best point-to-point fit is calculated. Specifically, when constructing the objective function to minimize distance error, to prevent misjudgment of skeleton posture due to clothing occlusion (such as the loss of waist features due to a heavy down jacket), this embodiment employs a weighted optimization function with biomechanical constraints, the expression of which can be constructed as: Where E is the total value of the objective function that minimizes the distance error. (Fitting term) is used to calculate the sum of squared Euclidean distances between the projected bone anchor points and the point cloud of the identified rigid anatomical feature areas to ensure positional alignment; (Biomechanical Constraints) Introducing human joint degree-of-freedom constraints, the system pre-determines the legal range of motion for each joint (such as spinal curvature and upper arm opening / closing angle). When the transformation matrix causes the skeletal posture to exceed these physiological limits, This will result in a huge penalty value, thus forcing the skeleton to maintain a natural posture; To balance the weighting coefficients between fitting error and biomechanical constraints, a layered locking strategy is employed when solving for the optimal rigidity transformation matrix: In the first stage, only high-confidence rigid points in the shoulders and neck are used to lock the trunk's main axis through rigidity transformation; in the second stage, while keeping the main axis fixed, the angular parameters of the limb bones are fine-tuned within biomechanical constraints using the limb extremity direction vectors. This matrix is ​​used to spatially transform the template, thereby locking the spatial pose of the parameterized human skeleton template to match the human pose in the scanned data. Using this aligned skeleton as a reference, the relatively fixed relative positional mapping between anatomical skeletal nodes (such as the sternum and lumbar vertebrae) and key points of soft tissue on the body surface is utilized. Through geometric deduction, the coordinates of parts completely obscured by clothing and unable to be directly measured are calculated, thus accurately locating human feature points, including the chest height point and waist node.

[0088] In one possible implementation, geometric measurement calculations are performed based on human feature points to extract garment outline size data. Specifically, this includes: determining the central axis of the human torso based on a parameterized human skeleton template, and constructing multiple measurement sections perpendicular to the central axis of the human torso; acquiring point cloud data of the three-dimensional scan data on the measurement sections, and performing convex hull fitting processing on the point cloud data of the sections to generate a fitted closed curve that removes noise from clothing wrinkles; calculating the perimeter of the fitted closed curve, and determining the perimeter as the circumference data in the garment outline size data.

[0089] In this embodiment, convex hull fitting refers to an algorithm based on computational geometry principles, used to construct a minimum convex polygon or convex polyhedron containing a given discrete set of points. In this solution, it specifically refers to simulating the physical behavior of a measuring tape tightly attached to the outermost edge of clothing on a two-dimensional cross-section, aiming to filter out concave areas formed by fabric draping or piling. For example, when processing the point cloud of the waist section, which presents a wavy concave-convex shape due to clothing folds, this processing can generate a smooth fitting contour that closely fits the outermost point cloud data, thereby shielding the interference of internal fold noise.

[0090] Specifically, to extract reliable geometric dimensions from scanned data with complex folds, the system uses a parameterized human skeleton template, precisely positioned and positioned in the previous steps, as an anatomical reference. It connects key skeletal nodes such as the cervical vertebrae and the pelvic center point to fit a central axis running vertically through the human torso. Based on standard anthropometry definitions, normal planes are established along this axis at key feature locations such as the chest height and waist nodes, thus constructing multiple measurement cross-sections strictly perpendicular to this axis. By calculating the intersection of these mathematical cross-sections with the surface of the 3D scanned model, discrete point cloud data distributed on these planes is obtained. Since the scanned clothing surface point cloud often exhibits local undulations, directly measuring the point cloud path would lead to inflated values. Therefore, computational geometry algorithms are used to perform convex hull fitting on these cross-sectional point cloud data. This process automatically ignores concave noise points within the point cloud caused by clothing folds, connecting only the outermost data points in each direction to generate a tight, smooth, and self-intersecting closed fitting curve. By calculating the geometric perimeter of the closed curve, it is determined as the circumference data that can represent the actual shape of the outer layer of the garment, and stored as a key component in the garment outline size data, providing a high-precision input basis for subsequent calculation of human net size.

[0091] S204: Based on the baseline structural relaxation value, fabric thickness value, and fabric elastic coefficient of the reference garment model, reverse calculation is performed on the garment outline size data to obtain the net body size data of the target object.

[0092] For example, suppose the target object is scanned while wearing a jacket with a thick fabric and a loose fit. The obtained garment outline dimensions are actually a superposition of data consisting of the human body, the fabric layer, and the air gap layer between the garment and the body. This step aims to use a digital benchmark garment model as a geometric reference to peel away the non-biological features covering the human body layer by layer through a physical inversion algorithm. The system will deduct the circumference increment caused by the fabric thickness and correct for deformation errors caused by tightness or constriction based on the fabric's elasticity coefficient. At the same time, combined with the natural drape characteristics of the garment structure under gravity, it will distinguish the differences in looseness of different body parts (such as the close-fitting shoulders and the unsupported waist), thereby inferring the actual human body shape inside based on the outer garment outline and restoring high-confidence body data.

[0093] In one possible implementation, based on the baseline structural relaxation value, fabric thickness value, and fabric elastic coefficient of the reference garment model, the garment outline size data is reverse-calculated to obtain the net body size data of the target object. Specifically, this includes: generating a dynamic gap distribution coefficient that varies with height using a preset gravity drape model based on the vertical height distribution of the garment outline size data on the human body; multiplying the baseline structural relaxation value by the dynamic gap distribution coefficient to obtain the actual gap amount after gravity correction; calculating the girth increment value caused by the physical thickness of the fabric based on the fabric thickness value, and determining the lateral deformation correction coefficient based on the fabric elastic coefficient; performing differential subtraction processing on the garment outline size data using the girth increment value and the actual gap amount, and numerically scaling the subtracted garment outline size data using the lateral deformation correction coefficient to obtain the net body size data.

[0094] In this application embodiment, the gravity drape model refers to a mathematical distribution model based on the physical and mechanical properties of fabrics, used to simulate the drape state and spatial distribution law of clothing fabric relative to the human body surface under the action of gravity. For example, the model can quantitatively describe the physical phenomenon that clothing is close to the human body at the shoulder support point, while the waist and abdomen are suspended in the air and produce a large gap due to gravity traction, thereby providing an algorithmic basis for converting static structural looseness into dynamic spatial distance.

[0095] Specifically, to accurately reconstruct the true net body dimensions from the external contours including clothing thickness and drape gaps, the system first analyzes the height position of the garment contour dimensions (i.e., the perimeter measured in the previous steps) on the vertical axis of the human body. Given that clothing does not uniformly conform to the body, a preset gravity-based drape model is used to calculate the degree of influence of gravity at this height position, generating a dynamic gap distribution coefficient that changes dynamically with height. More specifically, when constructing the gravity-based drape model, the system first identifies rigid support areas (such as the acromion) and suspended areas (such as the waistline) in the scanned data. The direction of gravity is set vertically downwards, and for any height on the human body surface... The cross-section at that location, the dynamic gap distribution coefficient The calculation logic is as follows: Calculate the current height. relative to the height of the most recent upper support point vertical distance Constructing association functions ,in The maximum porosity factor related to fabric drape. Let be the gravitational decay constant. According to this function, when When the coefficient approaches 0 (i.e., at the shoulder support), it indicates no gap; as the distance increases (moving down to the waist suspension area), the coefficient increases. Furthermore, the model can introduce a normal correction factor, fine-tuning the coefficient by calculating the angle between the human body surface normal vector and the gravity vector. Based on these calculations, a dynamic gap distribution coefficient that dynamically changes with height is generated (this coefficient may approach 0 in support areas such as the shoulders, but significantly increases in suspension areas such as the waist). The theoretical baseline structural relaxation value retrieved from the database is multiplied by this dynamic coefficient to calculate the actual gap between the clothing and the human body under the current gravitational field. Simultaneously, the interference of the fabric's own physical properties is considered: on the one hand, based on the fabric thickness value, the increase in perimeter due to fabric stacking, i.e., the girth increment (usually about 6.28 times the fabric thickness), is calculated using circumferential geometry formulas; on the other hand, based on the fabric elasticity coefficient, the lateral deformation correction coefficient of the fabric under lateral stretching or natural shrinkage is determined. Finally, the comprehensive back-calculation logic is executed: First, the actual gap amount and girth increment value are subtracted from the measured garment outline data through differential subtraction processing. The influence of air layer and fabric thickness is subtracted. Then, the intermediate data after subtraction is numerically scaled (e.g., divided by stretch ratio) using the lateral deformation correction coefficient, thereby eliminating the influence of fabric elastic deformation and finally obtaining high-precision human body net size data.

[0096] S205: Obtain the forward ready-to-wear model corresponding to the target customized product category. The forward ready-to-wear model is configured with the pattern addition rules for the target customized product category.

[0097] In the embodiments of this application, the forward garment model refers to a digital pattern generation module or parameterized template with parameterized driving capabilities built for the garment production and manufacturing process. It is used to map discrete human body size data into garment specification data that meets specific style requirements. Internally, it encapsulates the logical mapping relationship that transforms ergonomic data into industrial pattern making data. For example, for the category of "slim-fit suit", the forward garment model is a parameterized template that includes the formula for calculating the ease allowance of each part of the suit (such as chest circumference, waist circumference, and armhole) and the pattern adjustment logic.

[0098] Specifically, when a customization request for a specific garment type is received (i.e., a generation request for a target customized category), the system indexes and retrieves data from a pre-defined forward design database based on the category identifier (e.g., "retro jacket" or "sports sweatshirt") carried in the request. Unlike the previous reverse engineering process, the model invoked here is no longer used for measurement but for generation. Therefore, what is obtained is a forward garment model that precisely corresponds to the target customized category. This model is not a static geometric mesh but is pre-configured and bound to a complete set of pattern allowance rules. These rules define the conversion logic from net body measurements to finished garment measurements. They not only cover the minimum functional ease required to maintain normal limb movement but also include decorative ease required to reflect specific style (e.g., oversized or fitted style) and pre-defined compensation parameters for different fabric characteristics. This ensures that subsequent input net body data can be transformed into production data that fits well and meets design expectations.

[0099] S206: Input the human body net size data into the forward garment model, use the pattern addition rules to determine the target ease amount, and add the target ease amount to the human body net size data to generate the target garment production size for the target customized product category.

[0100] For example, assuming the target customized product category is a "slim-fit business suit," the system uses the highly accurate reconstructed human body dimensions data (such as net chest circumference and net shoulder width) from the previous steps as source data input. At this point, the forward garment model executes a parametric mapping function. It reads the "slim-fit style" rules defined for this suit category and automatically calculates the functional space required for basic human activities (such as raising arms and breathing) while also providing the decorative space (i.e., target ease allowance) that reflects the suit's crisp shape. The system superimposes these ease allowance values ​​onto the human body dimensions to calculate the final finished garment specifications and encapsulates them into digital instructions that can be recognized by the production terminal. This process achieves automated mapping from human biometric data to garment manufacturing data, enabling the generated production dimensions to take into account both user body characteristics and pattern design rules.

[0101] In one possible implementation, the target ease amount is determined using pattern easing rules, specifically including: obtaining the target fabric stiffness parameters and target fabric drape parameters for the target customized product category; querying the preset basic ease amount table in the pattern easing rules to obtain the basic ease amount value corresponding to the level of the human body net size data; constructing an ease amount correction factor using the target fabric stiffness parameters and target fabric drape parameters; and using the ease amount correction factor to perform weighted adjustment on the basic ease amount value to determine the target ease amount.

[0102] In the embodiments of this application, the ease correction factor refers to a dynamic weighted coefficient constructed based on the physical and mechanical properties of the fabric. It is used to quantify the degree of influence of the stiffness and drape of different fabric materials on the internal space requirements of the garment, thereby making personalized corrections to the standardized basic ease. For example, for fabrics with high stiffness and poor drape (such as heavy denim), the factor is usually set to a value greater than 1 to increase the internal space required for shape support, while for fabrics with excellent drape (such as modal), the factor is less than 1 to reduce the accumulation of excess gaps.

[0103] Specifically, to achieve precise conversion from human body dimensions to clothing dimensions, the system first acquires the physical boundary conditions that determine the final shape of the garment. This involves reading the physical test data of the fabric selected for the target customized product category from a digital fabric database and extracting the target fabric stiffness parameter (denoted as...) to quantitatively characterize its resistance to bending deformation. ) and the drape parameters of the target fabric that quantify its natural draping shape under gravity (denoted as ) It should be noted that the target fabric drape parameter in this application is used to quantify the fabric's drape ability. The larger the value of this parameter, the softer and more fitted the fabric (i.e., the better the drape).

[0104] Furthermore, a size benchmark is established. Using the previously calculated net body size data as the index key, a lookup operation is performed in the preset basic ease table in the pattern allowance rules to match the specification level to which the size belongs, thereby obtaining the general basic ease value set based on the standard intermediate state fabric for that level (denoted as ). ).

[0105] Furthermore, physical attributes are transformed into mathematical weights to construct an ease correction factor (denoted as ) that reflects the current fabric style characteristics. In a preferred implementation, the construction logic of this factor follows a nonlinear weighting law based on physical properties: the ease correction factor is positively correlated with the fabric stiffness parameter and negatively correlated with the fabric drape parameter. Specifically, the system calculates the ease correction factor using the following mathematical model: Where 1 represents the baseline correction factor, that is, when the fabric properties are consistent with the standard fabric, the correction factor is 1; This represents the stiffness weighting coefficient, used to adjust the sensitivity of stiffness changes to the effect of looseness. This represents the drape weighting coefficient, used to adjust the sensitivity of drape changes to the effect of looseness. This represents the relative deviation of stiffness, and its calculation method is as follows: ,in These are the preset standard fabric stiffness parameters; when the target fabric is stiffer than the standard fabric... If it is a positive value, then Enlarge it to increase internal support space and prevent dead wrinkles; The relative deviation of sag is expressed as follows: ,in These are the preset standard fabric drape parameters; when the target fabric has better drape (is softer) than the standard fabric, If it is a positive value, then The size is reduced, thereby reducing excess gaps and improving fit.

[0106] Furthermore, the system utilizes the calculated loosening correction factor. For general basic loose value Perform weighted adjustment (i.e.) This allows for the calculation of the target relaxation amount that satisfies both the needs of human movement and the specific fabric's design style. .

[0107] In one possible implementation, a target ease allowance is superimposed on the human body net size data to generate the target garment production size for the target customized product category. Specifically, this includes: allocating the target ease allowance to each circumference corresponding to the human body net size data to obtain the target garment specification value; using a forward garment model, mapping the target garment specification value to the coordinate parameters of two-dimensional pattern control points; generating industrial production pattern data containing seam allowance and garment shrinkage based on the coordinate parameters, and using the industrial production pattern data as the target garment production size.

[0108] In the embodiments of this application, two-dimensional pattern control points refer to geometric anchor points used to define the key positions of the pattern outline shape and structure in digital garment pattern design. Changes in their coordinate values ​​directly determine the direction of the pattern's edge lines and its size. For example, the movement of the control point coordinates located at the bottom of the armhole will directly change the size of the bust and sleeve width, while the control point located at the neckline determines the size and depth of the neckline.

[0109] Specifically, to convert the target garment specification values ​​into digital pattern data that conforms to industrial production standards, the system first executes data fusion logic. It maps the target ease amount, determined based on the fabric and style, to the corresponding subdivisions of the human body's net dimensions using a preset part allocation coefficient matrix. For example, the total ease amount for "chest circumference" is not evenly distributed, but rather based on the difference between the scapular protrusion on the back and the movement of the chest, setting a distribution ratio coefficient for the front chest and back (e.g., setting the front chest allocation coefficient to 0.45 and the back allocation coefficient to 0.55). The system multiplies the total ease amount by the corresponding coefficient and then superimposes it onto the net body data for the front chest width and back width. After superposition calculation, the target garment specification value, including human body dimensions and comfort ease, is obtained. The automated design process is then initiated, utilizing a forward garment model. This model has a preset function mapping relationship between pattern geometry and specification values, enabling real-time parsing and mapping of the input target garment specification value into the coordinate parameters of two-dimensional pattern control points, thereby driving the virtual pattern to deform to match the target size. Based on this, and combined with the actual production process requirements, the basic pattern outline is processed a second time to automatically generate industrial production pattern data (usually a vector file in DXF or PLT format) that includes seam allowance (i.e. the edge allowance required for sewing and splicing) and garment shrinkage rate (i.e. the compensation ratio for estimated fabric shrinkage during washing). Finally, this set of industrial production pattern data, which can be directly used for automatic cutting, is output as the target garment production size.

[0110] This embodiment also discloses an electronic device, as shown in the reference. Figure 3 The electronic device may include: at least one processor 301, at least one communication bus 302, user interface 303, network interface 304, and at least one memory 305.

[0111] The communication bus 302 is used to enable communication between these components.

[0112] The user interface 303 may include a display screen and a camera. Optionally, the user interface 303 may also include a standard wired interface and a wireless interface.

[0113] The network interface 304 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface).

[0114] The processor 301 may include one or more processing cores. The processor 301 connects to various parts of the server using various interfaces and lines, and performs various server functions and processes data by running or executing instructions, programs, code sets, or instruction sets stored in memory 305, and by calling data stored in memory 305. Optionally, the processor 301 may be implemented using at least one hardware form of Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), or Programmable Logic Array (PLA). The processor 301 may integrate one or a combination of several of the following: Central Processing Unit (CPU), Graphics Processing Unit (GPU), and modem. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content required for display; and the modem handles wireless communication. It is understood that the modem may also not be integrated into the processor 301 and may be implemented as a separate chip.

[0115] The memory 305 may include random access memory (RAM) or read-only memory. Optionally, the memory 305 may include a non-transitory computer-readable storage medium. The memory 305 can be used to store instructions, programs, code, code sets, or instruction sets. The memory 305 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the above-described method embodiments, etc.; the data storage area may store data involved in the above-described method embodiments, etc. Optionally, the memory 305 may also be at least one storage device located remotely from the aforementioned processor 301. Figure 3 As shown, the memory 305, which serves as a computer storage medium, may include an operating system, a network communication module, a user interface module, and an application program for garment size conversion based on 3D scanning.

[0116] exist Figure 3In the electronic device shown, the user interface 303 is mainly used to provide an input interface for the user and to obtain the user input data; while the processor 301 can be used to call an application stored in the memory 305 for a garment size conversion method based on 3D scanning. When executed by one or more processors 301, the electronic device performs one or more methods as described in the above embodiments.

[0117] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0118] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0119] In some embodiments of this application, a computer-readable storage medium is provided, including instructions that, when executed on the electronic device, cause the electronic device to perform a 3D scanning-based garment size conversion method according to an embodiment of this application.

[0120] In some embodiments of this application, a computer program product is also provided, which, when run on an electronic device, causes the electronic device to execute a garment size conversion method based on 3D scanning according to an embodiment of this application.

[0121] In the several embodiments provided in this application, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some service interface; the indirect coupling or communication connection between apparatuses or units may be electrical or other forms.

[0122] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0123] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0124] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory 305 and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned memory 305 includes various media capable of storing program code, such as a USB flash drive, external hard drive, magnetic disk, or optical disk.

[0125] The foregoing description is merely an exemplary embodiment of this disclosure and should not be construed as limiting the scope of this disclosure. Any equivalent changes and modifications made in accordance with the teachings of this disclosure shall still fall within the scope of this disclosure. Those skilled in the art will readily conceive of other embodiments of this disclosure upon considering the disclosure in this specification. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not described in this disclosure. The specification and embodiments are considered exemplary only, and the scope of this application is defined by the claims.

Claims

1. A method for converting garment sizes based on 3D scanning, characterized in that, The method includes: Acquire 3D scan data of the target object in its current clothing state, perform feature recognition on the 3D scan data, and determine the clothing category information of the target object's current clothing. The reference sample garment model corresponding to the garment category information is retrieved from the preset reference sample garment database. The reference sample garment model is associated with and stores the reference structure relaxation value, fabric thickness value and fabric elasticity coefficient. Based on the three-dimensional scan data, the human body feature points of the target object are located, and geometric measurement calculations are performed based on the human body feature points to extract the garment outline size data. Based on the reference structure relaxation value, fabric thickness value, and fabric elasticity coefficient of the reference garment model, the garment outline size data is reverse-calculated to obtain the net body size data of the target object. Obtain a forward ready-to-wear model corresponding to the target customized product category, wherein the forward ready-to-wear model is configured with the pattern addition rules of the target customized product category; The human body net size data is input into the forward garment model, the target ease amount is determined using the pattern ease rules, and the target ease amount is superimposed on the human body net size data to generate the target garment production size for the target customized product category.

2. The method according to claim 1, characterized in that, The step of performing feature recognition on the 3D scan data to determine the clothing category information of the target object's current attire specifically includes: The spatial coordinates of the three-dimensional scan data are discretized and quantized based on a preset grid resolution to construct a three-dimensional voxel index space. Traverse the non-empty voxels in the three-dimensional voxel index space, and construct a local covariance matrix based on the spatial coordinate distribution of the non-empty voxels and the point cloud data in their neighborhoods. Perform eigenvalue decomposition on the local covariance matrix to obtain a set of eigenvalues ​​and a corresponding set of eigenvectors; The minimum eigenvalue is determined from the set of eigenvalues, and the eigenvector corresponding to the minimum eigenvalue is determined as the surface normal vector characterizing the local plane direction; The surface normal vector is mapped to the mesh position corresponding to the non-empty voxel to generate a three-dimensional surface normal feature tensor. A deep residual network is used to perform a three-dimensional convolution operation on the three-dimensional surface normal feature tensor to extract morphological feature vectors. The morphological feature vector is matched with the reference feature vector in the preset category feature library using Euclidean distance, and the category label corresponding to the reference feature vector with the smallest distance is determined as the clothing category information.

3. The method according to claim 1, characterized in that, The process of locating human feature points of the target object based on the three-dimensional scan data specifically includes: Perform surface differential geometry analysis on the three-dimensional scan data to calculate the local geometric curvature value of each data point in the three-dimensional scan data; Data points whose local geometric curvature values ​​exceed a preset gradient threshold are selected from the three-dimensional scan data, and the set of selected data points is determined as a subset of the candidate feature point cloud; Based on prior knowledge of human topology, spatial clustering and filtering are performed on the subset of candidate feature point clouds to identify rigid anatomical feature regions that respectively represent the acromion position and the cervical spine position. A preset parameterized human skeleton template is invoked, wherein the parameterized human skeleton template is pre-calibrated with bone anchor points corresponding to the rigid anatomical feature area; Construct an objective function that minimizes distance error, calculate the optimal rigid transformation matrix that maps the skeletal anchor points to the rigid anatomical feature area, and use the optimal rigid transformation matrix to lock the spatial pose of the parameterized human skeleton template. Based on the spatial posture, the relative positional mapping relationship between bones and soft tissues is used to deduce and locate the human feature points, including the chest height point and the waist point.

4. The method according to claim 3, characterized in that, The step of performing geometric measurement calculations based on the human body feature points to extract garment outline size data specifically includes: Based on the parameterized human skeleton template, the central axis of the human torso is determined, and multiple measurement sections perpendicular to the central axis of the human torso are constructed. The cross-sectional point cloud data of the three-dimensional scan data on the measurement section is obtained, and the cross-sectional point cloud data is subjected to convex hull fitting processing to generate a fitted closed curve with the removal of clothing wrinkle noise. Calculate the perimeter of the fitted closed curve and determine the perimeter as the circumference data in the garment outline size data.

5. The method according to claim 1, characterized in that, The method involves reversing the calculation of the garment outline dimensions based on the baseline structural relaxation value, fabric thickness value, and fabric elasticity coefficient of the baseline reference garment model to obtain the net body dimensions of the target object. Specifically, this includes: Based on the vertical height distribution of the garment outline dimensions on the human body, a dynamic gap distribution coefficient that varies with height is generated using a preset gravity drape model. Multiply the relaxation amount of the reference structure by the dynamic gap distribution coefficient to obtain the actual gap amount after gravity correction. The girth increment caused by the physical thickness of the fabric is calculated based on the fabric thickness value, and the lateral deformation correction coefficient is determined based on the fabric elasticity coefficient. The garment outline size data is differentially subtracted using the girth increment value and the actual gap amount, and the subtracted garment outline size data is numerically scaled using the lateral deformation correction coefficient to obtain the human body net size data.

6. The method according to claim 1, characterized in that, The determination of the target relaxation amount using the pattern addition rules specifically includes: Obtain the target fabric stiffness parameters and target fabric drape parameters for the target customized product category; Query the preset basic ease table in the pattern addition rules to obtain the basic ease value corresponding to the grade of the human body net size data; An ease correction factor is constructed using the target fabric stiffness parameter and the target fabric drape parameter. The ease correction factor is then used to weight and adjust the base ease value to determine the target ease amount.

7. The method according to claim 6, characterized in that, The process of superimposing the target ease allowance onto the human body net size data to generate the target garment production size for the target customized product category specifically includes: The target relaxation amount is allocated to each circumference part corresponding to the human body net size data to obtain the target garment specification value; Using the forward garment model, the target garment specification value is mapped to the coordinate parameters of two-dimensional pattern control points; Based on the coordinate parameters, industrial production pattern data containing seam allowance and garment shrinkage rate is generated, and the industrial production pattern data is used as the target garment production size.

8. An electronic device, characterized in that, Including processor and memory; The memory is used to store computer program code, the computer program code including computer instructions, and the processor invokes the computer instructions to cause the electronic device to perform the method as described in any one of claims 1-7.

9. A computer-readable storage medium storing computer instructions, characterized in that, When the computer instructions are executed on the electronic device, the electronic device causes the electronic device to perform the method as described in any one of claims 1-7.

10. A computer program product, characterized in that, When the computer program product is run on an electronic device, it causes the electronic device to perform the method as described in any one of claims 1-7.

Images