A personalized baby clothes customization system based on traditional clothing elements

Abstract

The application discloses a kind of individualized baby clothes customization system based on traditional costume element, it is related to clothing customization technical field. Including: traditional costume element library module, for hierarchical construction database according to shape class, pattern class, color class, store traditional costume element, each traditional costume element is attached with the core characteristic parameter of AI annotation, for traditional costume element fast retrieval and call.The application accurately collects and labels the characteristic parameters of traditional costume elements through the traditional costume element library module, extracts the traditional costume element features and customization keywords in user demand by combining the multi-modal input analysis module, and then uses the proportion adjustment algorithm and the sparse adjustment algorithm through the AI adaptation optimization module to accurately adapt the traditional costume element to the size of baby clothes, while maintaining the core details such as pattern texture, shape structure, etc., effectively solving the problem of traditional costume element and baby clothes small size version adaptation distortion.

Description

Technical Field

[0001] This invention relates to the field of clothing customization technology, specifically a personalized doll clothing customization system based on traditional clothing elements. Background Technology

[0002] Traditional clothing elements carry rich cultural connotations, including patterns from Dong ethnic brocade and stylistic features such as the cross-collar, right-fastening, and wide sleeves of Hanfu. These elements have gradually been incorporated into BJD doll clothing design in recent years, creating a demand for customization that combines artistry and cultural significance. With the development of personalized consumption trends, users not only want doll clothing to reflect unique aesthetics but also seek accurate reproduction and adaptation of traditional elements. The maturity of AI technology, 3D modeling, and virtual try-on technology provides technical support for personalized customization of doll clothing featuring traditional elements, enabling a fully digital process from element extraction and design generation to effect preview.

[0003] In the prior art, patent CN104391594A discloses a touch-interactive clothing DIY ordering system. While this invention enables the selection and design of clothing styles and patterns, as well as online virtual try-on, it has significant shortcomings when it comes to custom-made doll clothes. The structural details of traditional clothing elements, such as the warp and weft weave of Dong brocade and the spacing ratio of Hanfu buttons, are difficult to accurately match with the small sizes of doll clothes, often resulting in distortion after scaling or layout inconsistencies. Furthermore, this invention can only display static try-on effects and cannot reproduce the dynamic characteristics of traditional elements such as wrinkles and drape caused by joint movement when the doll clothes are worn. This makes it difficult for users to predict the actual customized effect, reducing customization satisfaction. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a personalized doll clothing customization system based on traditional clothing elements, which solves the problems of mismatch between traditional clothing elements and small-sized doll clothing, and difficulty in predicting dynamic effects during static try-on.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a personalized doll clothing customization system based on traditional clothing elements, comprising:

[0006] The Traditional Clothing Element Library module is used to build a database in layers according to shape, pattern and color to store traditional clothing elements. Each traditional clothing element comes with AI-annotated core feature parameters for quick retrieval and retrieval of traditional clothing elements.

[0007] The multimodal input parsing module supports text descriptions, reference image uploads, and hand-drawn sketches as input methods. It extracts traditional clothing element features from user input using contrast enhancement algorithms and GrabCut image segmentation algorithms, and combines natural language processing technology to parse customized demand keywords.

[0008] The AI ​​adaptation and optimization module is used to automatically adapt the size, layout, and density of traditional clothing elements based on the doll's clothing size parameters using a proportional adjustment algorithm, in order to maintain the original characteristics of the traditional clothing elements; the calculation formula for the proportional adjustment algorithm is:

[0009] ;

[0010] in, Sizes adapted to fit traditional clothing elements. The original dimensions of traditional clothing elements. The ratio of doll clothing size parameters to the corresponding sizes of traditional clothing elements;

[0011] The dynamic virtual try-on module uses diffusion model technology to construct 3D models of doll clothes, and combines a skeletal animation system to simulate the joint movement of BJD dolls, rendering the fold changes, draping effects and light and shadow reflections of traditional clothing elements in a dynamic state in real time.

[0012] The order generation and feedback module is used to integrate customized solution parameters, generate standardized production orders, and support users to modify and provide feedback on the fitting effect. The system automatically iterates and optimizes the design solution.

[0013] Furthermore, the traditional clothing element library module is used to acquire physical images and craft materials of traditional clothing through image acquisition equipment, optimize image quality using a contrast enhancement algorithm, segment and extract traditional clothing elements using the GrabCut algorithm, and have technical personnel annotate their core feature parameters in detail. The core feature parameters include pattern size ratio, shape and structural details, and color value range. The annotated data is entered into a hierarchical database for regular updates and maintenance.

[0014] Furthermore, the calculation formula for the contrast enhancement algorithm is as follows:

[0015] ;

[0016] in, For enhanced pixel grayscale values, The original pixel grayscale value. The minimum grayscale value of a pixel in the image. The maximum grayscale value of a pixel in the image. The maximum value within the target grayscale range. This is the minimum value within the target grayscale range.

[0017] Furthermore, the AI ​​adaptation and optimization module is used to call the standard size parameters of the target doll clothing, including the length, shoulder width, and sleeve circumference. Through a proportional adjustment algorithm, the extracted pattern is reduced in size according to the standard size parameters, maintaining the warp and weft weave texture and details of the pattern and preventing pattern distortion. At the same time, the shape and structure of traditional clothing are adapted to the neckline size of the doll clothing, and the width and angle of the overlapping collar are adjusted.

[0018] Furthermore, the AI ​​adaptation and optimization module is also used to adjust the density of the pattern when it is scaled down proportionally; the density adjustment is achieved through the following formula:

[0019] ;

[0020] in, This indicates the adjusted pattern density. Indicates the density of the original pattern. Indicates the area of ​​the target region. This indicates the area of ​​the original pattern region.

[0021] Furthermore, the dynamic virtual try-on module is based on the Unity engine to build a 3D model of the BJD doll. It generates a high-precision 3D clothing model from the adapted doll clothing design scheme, associates each joint of the BJD doll with skeletal binding technology, and uses a diffusion model to render in real time the changes in the folds of the doll clothing, the draping effect of the fabric, and the reflection of ambient light and shadow under the three actions of the BJD doll walking, raising its hand, and sitting.

[0022] Furthermore, the calculation formula for the skeletal animation system is as follows:

[0023] ;

[0024] in, This indicates the vertex position of the doll's clothing in the skeletal animation. This indicates the vertex position of the doll's clothing in the bound pose. The vertex weight matrix represents the degree to which a vertex is affected by the skeleton. This represents the global transformation matrix of the skeleton, which describes the three types of transformations performed on the skeleton in three-dimensional space: translation, rotation, and scaling.

[0025] Furthermore, the dynamic virtual try-on module is also used to allow users to switch the BJD doll's pose and viewing angle through system interface controls, so as to preview the dynamic features of the doll wearing clothes from multiple angles and all perspectives. The dynamic features include the real-time changes of folds and textures, the natural drape of the fabric, and the subtle performance of light and shadow reflection.

[0026] Furthermore, the order generation and feedback module is used to automatically integrate all parameters of the customized solution after the user confirms the virtual try-on effect. The parameters include element combinations, size data, fabric selection and receiving information, generate a standardized production order, and send the order information to the backend production system.

[0027] Furthermore, the order generation and feedback module is also used to save the detailed parameters of the customized solution to the user's personal account while generating the order, so that the user can review or modify the customized solution later. The system can also automatically iterate and optimize the current design solution based on the user's modification feedback.

[0028] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0029] This invention accurately collects and labels characteristic parameters of traditional clothing elements through a traditional clothing element library module. Combined with a multimodal input parsing module, it extracts traditional clothing element features and customization keywords from user needs. Then, an AI adaptation and optimization module uses proportion and density adjustment algorithms to ensure that traditional clothing elements are precisely adapted to the size of the doll clothes while maintaining core details such as patterns, textures, and structural forms. This effectively solves the problem of mismatch between traditional clothing elements and small-sized doll clothes. The dynamic virtual try-on module builds a 3D model of the BJD doll based on the Unity engine. Combined with a diffusion model and skeletal animation system, it renders the wrinkles, drape, and lighting effects of the doll clothes in a dynamic state in real time, supporting multi-view viewing and solving the problem that static try-on cannot predict the actual dynamic effect of wearing the doll. The order generation and feedback module integrates parameters to generate standardized orders and supports iterative optimization, comprehensively improving the accuracy and user satisfaction of doll clothing customization and realizing the efficient application of traditional clothing elements in doll clothing customization. Attached Figure Description

[0030] Figure 1 This is a system structure diagram of the present invention;

[0031] Figure 2 This is a flowchart illustrating the construction process of the traditional clothing element library of the present invention.

[0032] Figure 3 This is a schematic diagram illustrating the dynamic virtual try-on and skeletal animation principle of the present invention. Detailed Implementation

[0033] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0034] Please see Figure 1-3 This invention provides a personalized doll clothing customization system based on traditional clothing elements, comprising:

[0035] The Traditional Clothing Element Library module is used to build a database in layers according to shape, pattern and color to store traditional clothing elements. Each traditional clothing element comes with AI-annotated core feature parameters for quick retrieval and retrieval of traditional clothing elements.

[0036] The multimodal input parsing module supports text descriptions, reference image uploads, and hand-drawn sketches as input methods. It extracts traditional clothing element features from user input using contrast enhancement algorithms and GrabCut image segmentation algorithms, and combines natural language processing technology to parse customized demand keywords.

[0037] The AI ​​adaptation and optimization module is used to automatically adapt the size, layout, and density of traditional clothing elements based on the doll's clothing size parameters using a proportional adjustment algorithm, in order to maintain the original characteristics of the traditional clothing elements; the calculation formula for the proportional adjustment algorithm is:

[0038] ;

[0039] in, Sizes adapted to fit traditional clothing elements. The original dimensions of traditional clothing elements. The ratio of doll clothing size parameters to the corresponding sizes of traditional clothing elements;

[0040] The dynamic virtual try-on module uses diffusion model technology to construct 3D models of doll clothes, and combines a skeletal animation system to simulate the joint movement of BJD dolls, rendering the fold changes, draping effects and light and shadow reflections of traditional clothing elements in a dynamic state in real time.

[0041] The order generation and feedback module is used to integrate customized solution parameters, generate standardized production orders, and support users to modify and provide feedback on the fitting effect. The system automatically iterates and optimizes the design solution.

[0042] Specifically, the traditional clothing element library module first collects images of real objects such as Hanfu cross-collar and Dong ethnic brocade patterns using high-definition image acquisition equipment. These images are then combined with craft information recorded in ancient books. After the image quality is optimized using a contrast enhancement algorithm, the GrabCut algorithm is used to accurately segment the elements into three categories: shape, pattern, and color. Technicians then label the core feature parameters such as the size ratio of the pattern, the details of the shape and structure, and the color value range. The elements are then stored in a hierarchical database according to their categories for easy retrieval and access later.

[0043] The multimodal input parsing module allows users to describe "horse-face skirt pattern doll clothes" via text, upload reference images of traditional clothing, or hand-drawn sketches. Figure 3The requirements can be submitted in several ways. For hand-drawn sketches, a line optimization algorithm is used to enhance the clarity of the outline and reduce the impact of blurry hand-drawn strokes on element extraction, ensuring that the extraction accuracy is consistent with the reference image. A contrast enhancement algorithm is used to improve the clarity of the input image. Then, the GrabCut algorithm is used to extract the traditional clothing element features. Natural language processing technology is combined to analyze keywords such as "horse-face skirt" and "pattern" to clarify the core customization requirements.

[0044] The AI ​​adaptation and optimization module calls upon a pre-set standard size library for different BJD doll clothing models to obtain parameters such as the length, shoulder width, and sleeve circumference of the target doll clothing. Simultaneously, considering the elasticity characteristics of different fabrics, it assigns corresponding adaptation coefficients to elastic fabrics (such as spandex blends) and non-elastic fabrics (such as cotton, linen, and silk). Elastic fabrics receive an additional 0.05-0.1 stretch allowance compensation. The adaptation size is calculated using a proportional adjustment algorithm, with the formula as follows:

[0045] ;

[0046] in Sizes adapted to fit traditional clothing elements. The original dimensions of traditional clothing elements. The ratio of doll clothing size parameters to the corresponding sizes of traditional clothing elements is used to proportionally reduce the pattern while maintaining the warp and weft weave texture. The width and angle of the overlapping collar are adjusted to make the shape fit the neckline size.

[0047] The dynamic virtual try-on module builds a 3D model of a BJD doll based on the Unity engine. It generates a high-precision 3D clothing model from the adapted doll clothing design scheme. It connects each joint of the doll through skeletal binding technology and uses diffusion model to render the wrinkle changes, draping effects and light and shadow reflections under three actions: walking, raising an arm and sitting. Users can switch the action posture and viewing angle through interface controls to preview the dynamic effects from all angles.

[0048] After the user confirms the fitting effect, the order generation and feedback module integrates parameters such as element combination, size data, fabric selection, and shipping information to generate a standardized production order, which is simultaneously sent to the backend production system. At the same time, detailed parameters of the customized design are saved to the user's personal account, allowing the user to view or submit modifications at any time. The system automatically iterates and optimizes the design based on feedback. After implementation, traditional clothing elements are accurately adapted, and the dynamic fitting effect closely resembles the actual wearing condition, improving customization satisfaction.

[0049] In this embodiment, the traditional clothing element library module is used to acquire physical images and craft materials of traditional clothing through image acquisition equipment. After optimizing the image quality using a contrast enhancement algorithm, the traditional clothing elements are segmented and extracted using the GrabCut algorithm. Technicians annotate the core feature parameters in detail, including pattern size ratio, shape and structural details, and color value range. The annotated data is entered into a hierarchical database for regular updates and maintenance.

[0050] Specifically, in the implementation of the traditional clothing element library module, high-definition industrial cameras were used as image acquisition devices to photograph the front and back of traditional clothing such as Hanfu, Tang suits, and cheongsams, as well as details, ensuring that the image resolution met the requirements for element extraction. Simultaneously, craft materials such as the craft notes of intangible cultural heritage inheritors and publicly available clothing artifacts from museums were collected. The acquired images were first optimized using contrast enhancement algorithms to address issues of uneven brightness and blurred details. Then, the GrabCut algorithm was used to manually mark the foreground and background areas, segmenting independent shape, pattern, and color elements. Technicians used professional image analysis tools to annotate the core feature parameters of each element. For pattern elements, the size ratio and latitude / longitude direction were highlighted; for shape elements, structural details such as the angle of the collar and the curvature of the wide sleeves were recorded in detail; and for color elements, the color value range was clearly defined. After labeling, the data is categorized into form, pattern, and color and entered into the database. The database is set up with a regular update mechanism. Every quarter, new traditional clothing resources and craft information are collected through three methods: cooperation with intangible cultural heritage workshops to collect new products, digital authorization of museum cultural relics, and collection and review of user-created traditional clothing elements. This supplements and improves the element library, ensuring the comprehensiveness and timeliness of traditional clothing elements, and providing rich and accurate material support for subsequent customization.

[0051] In this embodiment, the calculation formula for the contrast enhancement algorithm is as follows:

[0052] ;

[0053] in, For enhanced pixel grayscale values, The original pixel grayscale value. The minimum grayscale value of a pixel in the image. The maximum grayscale value of a pixel in the image. The maximum value within the target grayscale range. This is the minimum value within the target grayscale range.

[0054] Specifically, contrast enhancement algorithms are mainly applied in the preprocessing stage after traditional clothing image acquisition to improve image feature clarity, facilitating subsequent element segmentation and extraction. In implementation, the original grayscale value of each pixel in the acquired image is first obtained. The minimum value among all pixel grayscale values ​​is obtained by traversing the image. and maximum value Set the maximum value of the target grayscale range. and minimum value The target grayscale value range is typically set to 0-255 to fit standard image display. Substituting the above parameters into the formula... The numerator and denominator are both grayscale value differences, which have the same dimensions and do not require additional normalization processing. The enhanced pixel grayscale value is then calculated. This algorithm enhances the texture of images that were originally unclear in dark areas, while also preserving details in overly bright areas. This makes the subsequent GrabCut algorithm more accurate in segmenting traditional clothing elements, effectively improving the completeness and accuracy of element extraction.

[0055] In this embodiment, the AI ​​adaptation and optimization module is used to call the standard size parameters of the target doll clothing. The standard size parameters include clothing length, shoulder width, and sleeve circumference. Through the proportional adjustment algorithm, the extracted pattern is reduced in proportion to the standard size parameters, so as to maintain the warp and weft weave texture characteristics and details of the pattern and prevent the pattern from being distorted. At the same time, the shape and structure of traditional clothing are adapted to the neckline size of the doll clothing, and the width and angle of the overlapping collar are adjusted.

[0056] Specifically, when implementing the AI ​​adaptation and optimization module, a standard size library for doll clothing covering mainstream BJD doll models is first built, storing key size parameters such as clothing length, shoulder width, and sleeve circumference for different models. After the user selects a doll model, the system automatically retrieves the corresponding size parameters. For extracted traditional clothing patterns, a scaling ratio is calculated using a proportional adjustment algorithm. , The value is determined by the ratio of the doll's clothing size to the original size of the traditional clothing pattern. After normalization, it ensures dimensional consistency during scaling. The pattern is reduced according to this ratio, while preserving details such as the warp and weft weave texture and embroidery stitches to avoid blurring or distortion due to scaling. For the structure of traditional clothing, such as the cross-collar of Hanfu and the frog button layout of Tang suits, the system adjusts the overlap width and angle of the cross-collar based on parameters such as the neckline size and shoulder width of the doll's clothing. This ensures the shape fits the doll's clothing pattern, and the frog button spacing matches the length ratio of the doll's clothing, ensuring that the core features of the traditional shape are not lost. At the same time, it coordinates and unifies with the layout of the doll's clothing, solving the problem of mismatch between the traditional shape and the small size of the doll's clothing.

[0057] In this embodiment, the AI ​​adaptation and optimization module is also used to adjust the density of the pattern when the pattern is scaled down proportionally; the density adjustment is achieved through the following formula:

[0058] ;

[0059] in, This indicates the adjusted pattern density. Indicates the density of the original pattern. Indicates the area of ​​the target region. This indicates the area of ​​the original pattern region.

[0060] Specifically, pattern density adjustment is performed after the scaling step in the AI ​​adaptation and optimization module, aiming to avoid patterns becoming too dense or too sparse after scaling. During implementation, the density of the original pattern is first obtained. That is, the number of pattern elements per unit area, and then the area of ​​the original pattern region is determined. Area of ​​the target area of ​​the pattern on the doll's clothes Both are measured in square millimeters. Substitute the above parameters into the formula. The area ratio is a dimensionless parameter, and its dimension remains consistent after multiplying by the density, thus the adjusted pattern density is calculated. Based on the calculation results, the system automatically increases or decreases the number of pattern elements. For example, if the target area is reduced after scaling the original pattern, the system reduces the number of pattern elements proportionally to maintain the visual harmony and original style of the pattern, ensuring that the adjusted pattern is both suitable for the size of the doll clothes and can present the aesthetic effect of traditional patterns.

[0061] In this embodiment, the dynamic virtual try-on module is based on the Unity engine to build a 3D model of the BJD doll. The adapted doll clothing design scheme is used to generate a high-precision 3D clothing model. The skeletal binding technology is used to associate each joint of the BJD doll. The diffusion model is used to render in real time the changes in the folds of the doll clothing, the draping effect of the fabric, and the reflection of ambient light and shadow under the three actions of the BJD doll: walking, raising its hand, and sitting.

[0062] Specifically, the dynamic virtual try-on module uses the Unity engine as its core development platform. First, it constructs 3D models of BJD dolls at different scales, accurately reproducing the structure and range of motion of ball joints. The AI-optimized doll clothing design is then imported into the engine, generating 3D clothing models using high-precision modeling technology, refining details such as patterns, textures, and fabric quality of traditional clothing. Skeletal rigging technology is used to associate the 3D clothing models with each joint of the BJD doll, ensuring the clothing model deforms synchronously with skeletal movement. A diffusion model is used to render three basic movements—walking, raising an arm, and sitting—in real time, simulating the generation and dissipation of fabric wrinkles, the natural curvature of drape, and the light and shadow reflection effects under ambient light, such as the sheen of silk fabrics and the matte texture of cotton and linen fabrics. After implementation, users can intuitively view the wearing effect of the doll clothing in a dynamic state, making the dynamic presentation of traditional clothing elements closer to real-world scenarios and solving the problem that static try-ons cannot predict the actual wearing effect.

[0063] In this embodiment, the calculation formula for the skeletal animation system is:

[0064] ;

[0065] in, This indicates the vertex position of the doll's clothing in the skeletal animation. This indicates the vertex position of the doll's clothing in the bound pose. The vertex weight matrix represents the degree to which a vertex is affected by the skeleton. This represents the global transformation matrix of the skeleton, which describes the three types of transformations performed on the skeleton in three-dimensional space: translation, rotation, and scaling.

[0066] Specifically, the formula for skeletal animation systems Joint motion simulation applied to the dynamic virtual try-on module ensures that the doll's clothing deforms naturally with skeletal movement. During implementation, This indicates the vertex position of the doll's clothing in the rigged pose, that is, the three-dimensional coordinates of each vertex of the clothing model, in millimeters; This is a vertex weight matrix, where the values ​​0-1 define the degree to which each vertex is affected by the corresponding bone. For example, the weight of the cuff vertex affected by the arm bone is 0.8, and the weight of the cuff vertex affected by the shoulder bone is 0.2. This is the global transformation matrix for the skeleton. This matrix integrates three types of transformation parameters: translation, rotation, and scaling. Translation is in millimeters, rotation in radians, and scaling in dimensionless ratios. Normalization ensures consistent dimensions in matrix operations. Substituting these three parameters into the formula yields the vertex positions of the doll's clothing within the skeletal animation. The system updates the shape of the clothing model based on the new positions of all vertices, avoiding clothing distortion or penetration when joints bend, making the dynamic shape of the doll's clothes more natural and realistic.

[0067] In this embodiment, the dynamic virtual try-on module is also used to allow users to switch the BJD doll's pose and viewing angle through system interface controls, so as to preview the dynamic features of the doll wearing clothes from multiple angles and all perspectives. The dynamic features include the real-time changes of folds and textures, the natural drape of the fabric, and the subtle performance of light and shadow reflection.

[0068] Specifically, the dynamic virtual try-on module features custom-designed controls, including an action switching button and a view adjustment slider. Users can click the action switching button to switch between three preset actions: walking, raising an arm, and sitting. Dragging the view adjustment slider allows for a 360-degree panoramic view. During implementation, the system responds to control operations in real time, updating the folds and drape of the doll's clothing synchronously when switching actions. For example, when switching to the raising-arm action, the system renders the stretching folds of the sleeves and the fit of the shoulder fabric in real time. When adjusting the view, the light and shadow reflection effects adjust synchronously with the view, ensuring clear observation of the doll's clothing's dynamic features from different angles. Through multi-directional, full-view preview functionality, users can fully understand the dynamic presentation of traditional clothing elements on the doll's clothing, such as the deformation and adaptation of patterns at folds and the stretching of shapes during movements, providing ample reference for confirming customized solutions.

[0069] In this embodiment, the order generation and feedback module is used to automatically integrate all parameters of the customization plan after the user confirms the virtual try-on effect. The parameters include element combination, size data, fabric selection and receiving information, generate a standardized production order, and send the order information to the backend production system.

[0070] Specifically, the order generation and feedback module is activated after the user confirms the design effect through a dynamic virtual try-on. The system automatically extracts all key parameters from the customization plan, including the combination of traditional clothing elements, doll size data, fabric selection results, and the user's shipping information. These parameters are integrated according to the production standard format to generate a standardized production order. The order clearly indicates the technical requirements for each production stage, such as the printing position of patterns and the sewing details of the shape. The order information is sent to the backend production system in real time through a data interface to ensure that the production stage accurately obtains the customization parameters and avoids information transmission errors. After implementation, the production process is executed precisely according to the order, the customization requirements of traditional clothing elements are strictly implemented, and production efficiency and product qualification rate are improved.

[0071] In this embodiment, the order generation and feedback module is also used to save the detailed parameters of the customized solution to the user's personal account while generating the order, so that the user can review or modify the customized solution later. The system can also automatically iterate and optimize the current design solution based on the user's modification feedback.

[0072] Specifically, upon order generation, the system encrypts and stores detailed parameters of the customized design in a dedicated folder within the user's personal account. Users can access information such as element combinations, size data, and design renderings at any time by logging into their account. If a user needs to modify the design, they can find the corresponding order in their account and submit adjustment requests through the interface's modification controls, such as changing patterns or adjusting shape details. After receiving feedback, the system automatically invokes the AI ​​adaptation and optimization module and the dynamic virtual try-on module to regenerate the design and update the preview, forming an iterative optimization loop. Iterative optimization is limited to a maximum of three iterations. If the user still needs modifications after three iterations, they must resubmit the customization request to avoid endless iterations. After implementation, users can complete design modifications without repeatedly submitting basic requirements. The system's automatic iteration function ensures that the modified design maintains accurate adaptation and dynamic presentation of traditional clothing elements, improving the user's customization experience and design satisfaction.

[0073] In summary, this invention accurately collects and labels the characteristic parameters of traditional clothing elements through a traditional clothing element library module, extracts the traditional clothing element features and customization keywords from user needs through a multimodal input parsing module, and then uses an AI adaptation and optimization module to adjust the proportion and density of traditional clothing elements to accurately fit the size of the doll clothes while maintaining core details such as patterns, textures, and structural forms. This effectively solves the problem of mismatch between traditional clothing elements and small-sized doll clothes. The dynamic virtual try-on module builds a 3D model of the BJD doll based on the Unity engine, and combines a diffusion model and a skeletal animation system to render the wrinkles, drapes, and lighting effects of the doll clothes in a dynamic state in real time. It supports multi-view viewing and solves the problem that static try-on cannot predict the actual dynamic effect of wearing the clothes. The order generation and feedback module integrates parameters to generate standardized orders and supports iterative optimization, which improves the accuracy of doll clothing customization and user satisfaction, and realizes the efficient application of traditional clothing elements in doll clothing customization.

[0074] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0075] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A personalized doll clothing customization system based on traditional clothing elements, characterized in that, include: The Traditional Clothing Element Library module is used to build a database in layers according to shape, pattern and color to store traditional clothing elements. Each traditional clothing element comes with AI-annotated core feature parameters for quick retrieval and retrieval of traditional clothing elements. The multimodal input parsing module supports text descriptions, reference image uploads, and hand-drawn sketches as input methods. It extracts traditional clothing element features from user input using contrast enhancement algorithms and GrabCut image segmentation algorithms, and combines natural language processing technology to parse customized demand keywords. The AI ​​adaptation and optimization module is used to automatically adapt the size, layout, and density of traditional clothing elements based on the doll's clothing size parameters using a proportional adjustment algorithm, in order to maintain the original characteristics of the traditional clothing elements; the calculation formula for the proportional adjustment algorithm is: ； in, Sizes adapted to fit traditional clothing elements. The original dimensions of traditional clothing elements. The ratio of doll clothing size parameters to the corresponding sizes of traditional clothing elements; The AI ​​adaptation and optimization module is also used to adjust the density of the pattern when it is scaled down proportionally; the density adjustment is achieved through the following formula: ； in, This indicates the adjusted pattern density. Indicates the density of the original pattern. Indicates the area of ​​the target region. Indicates the area of ​​the original pattern region; The dynamic virtual try-on module uses diffusion model technology to construct 3D models of doll clothes, and combines a skeletal animation system to simulate the joint movement of BJD dolls, rendering the fold changes, draping effects and light and shadow reflections of traditional clothing elements in a dynamic state in real time. The order generation and feedback module is used to integrate customized solution parameters, generate standardized production orders, and support users to modify and provide feedback on the fitting effect. The system automatically iterates and optimizes the design solution.

2. The personalized doll clothing customization system based on traditional clothing elements according to claim 1, characterized in that, The traditional clothing element library module is used to acquire physical images and craft materials of traditional clothing through image acquisition equipment. After optimizing the image quality using a contrast enhancement algorithm, the traditional clothing elements are segmented and extracted using the GrabCut algorithm. Technicians annotate the core feature parameters in detail, including pattern size ratio, shape and structural details, and color value range. The annotated data is entered into a hierarchical database for regular updates and maintenance.

3. The personalized doll clothing customization system based on traditional clothing elements according to claim 2, characterized in that, The calculation formula for the contrast enhancement algorithm is as follows: ； in, For enhanced pixel grayscale values, The original pixel grayscale value. The minimum grayscale value of a pixel in the image. The maximum grayscale value of a pixel in the image. The maximum value within the target grayscale range. This is the minimum value within the target grayscale range.

4. The personalized doll clothing customization system based on traditional clothing elements according to claim 1, characterized in that, The AI ​​adaptation and optimization module is used to call the standard size parameters of the target doll clothing. The standard size parameters include clothing length, shoulder width, and sleeve circumference. Through the proportional adjustment algorithm, the extracted pattern is reduced in size according to the standard size parameters, while maintaining the warp and weft weave texture characteristics and details of the pattern and preventing pattern distortion. At the same time, the shape and structure of traditional clothing are adapted to the neckline size of the doll clothing, and the width and angle of the overlapping collar are adjusted.

5. The personalized doll clothing customization system based on traditional clothing elements according to claim 1, characterized in that, The dynamic virtual try-on module is based on the Unity engine to build a 3D model of a BJD doll. It generates a high-precision 3D clothing model from the adapted doll clothing design scheme, and associates each joint of the BJD doll with skeletal binding technology. It uses a diffusion model to render in real time the changes in the folds of the doll clothing, the draping effect of the fabric, and the reflection of ambient light and shadow under the three actions of the BJD doll: walking, raising its hand, and sitting.

6. The personalized doll clothing customization system based on traditional clothing elements according to claim 1, characterized in that, The calculation formula for the skeletal animation system is: ； in, This indicates the vertex position of the doll's clothing in the skeletal animation. This indicates the vertex position of the doll's clothing in the bound pose. The vertex weight matrix represents the degree to which a vertex is affected by the skeleton. This represents the global transformation matrix of the skeleton, which describes the three types of transformations performed on the skeleton in three-dimensional space: translation, rotation, and scaling.

7. The personalized doll clothing customization system based on traditional clothing elements according to claim 1, characterized in that, The dynamic virtual try-on module also allows users to switch the BJD doll's pose and viewing angle through system interface controls, previewing the dynamic features of the doll's clothes from multiple angles and perspectives. The dynamic features include real-time changes in folds and textures, the natural drape of the fabric, and subtle light and shadow reflections.

8. The personalized doll clothing customization system based on traditional clothing elements according to claim 1, characterized in that, The order generation and feedback module is used to automatically integrate all parameters of the customized plan after the user confirms the virtual try-on effect. The parameters include element combination, size data, fabric selection and receiving information, generate a standardized production order, and send the order information to the backend production system.

9. The personalized doll clothing customization system based on traditional clothing elements according to claim 1, characterized in that, The order generation and feedback module is also used to save the detailed parameters of the customized solution to the user's personal account when generating the order, so that the user can review or modify the customized solution later. The system can also automatically iterate and optimize the current design solution based on the user's modification feedback.

Images