Visual clothing fabric texture design method and system applying big data

Abstract

The invention discloses a visual garment fabric texture design method and system applying big data, and the method comprises the steps: collecting garment fabric texture data through multi-source big data to construct an initial database, extracting texture data features through an image translation network, and mapping feature vectors to a three-dimensional space through a neural grid renderer to construct a three-dimensional grid model; parameters such as texture density, wrinkle degree and glossiness are set and fused into the model, texture mapping is carried out through an image translation network, and finally a visual design result is output. The system correspondingly comprises a data acquisition and storage unit, a feature extraction unit, a grid construction unit, a parameter setting unit, a texture mapping unit, a result output unit and the like. According to the method and system, the image translation network and the neural grid renderer are deeply fused, efficient and accurate visual design of the clothing fabric texture is achieved, the defects of traditional design in data utilization and effect presentation are overcome, and diversified design requirements are met.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of clothing fabric texture design, and particularly relates to a visual clothing fabric texture design method and system applying big data. BACKGROUND

[0002] Under the background of the booming fashion industry and the surge in consumer individualization demand, clothing fabric texture design, as the core link determining the uniqueness and attractiveness of products, is becoming increasingly important. At the same time, the advent of the big data era has brought data-driven transformation opportunities to various industries, and the field of clothing fabric texture design also urgently needs to tap potential design inspirations with the help of big data to realize the transformation from traditional experience-driven to data-intelligent-driven, in order to meet the rapidly changing market demand and the increasingly diversified aesthetic expectations of consumers.

[0003] However, the existing clothing fabric texture design technology has significant limitations. On the one hand, the traditional design mode relies on the personal experience of designers and limited sample data, making it difficult to fully utilize massive big data resources. Designers mainly obtain design materials based on past accumulation and a small amount of market research, and cannot systematically analyze the popular trends, color preferences and texture style evolution rules contained in big data, resulting in a deviation between design results and market real demand, making it difficult to accurately grasp consumer preferences, and products are prone to be unsold after being launched. On the other hand, in terms of technical implementation, traditional methods lack advanced digitalization and intelligentization means. In terms of texture effect presentation, whether hand-drawing or using basic image processing software, it is difficult to realistically simulate the complex physical properties of fabrics. For example, the natural wrinkles of fabrics caused by gravity and movement, and the gloss changes of different materials under light, traditional technology cannot achieve high-precision and dynamic visualization effects, making the design scheme lack of realism and persuasiveness, affecting design efficiency and the quality of the final product. SUMMARY

[0004] In order to overcome the shortcomings and deficiencies of the existing technology, the present application provides a visual clothing fabric texture design method and system applying big data.

[0005] The technical solution adopted by the present application is a visual clothing fabric texture design method applying big data, comprising the following steps:

[0006] Step S1: Constructing an initial clothing fabric texture database, collecting clothing fabric texture data including different materials, patterns and colors from multi-source big data, classifying and storing the texture data according to a preset classification rule to form an initial database;

[0007] Step S2: Based on the image translation network, feature extraction is performed on the texture data in the initial clothing fabric texture database. The texture data is processed through a multi-layer convolutional neural network to obtain a multi-dimensional feature vector including texture structure and color distribution features.

[0008] Step S3: Establish a neural mesh renderer to map the multidimensional feature vectors to three-dimensional space, construct a neural mesh with spatial topology, and form a three-dimensional mesh model corresponding to the texture features of clothing fabric by adjusting the node parameters and connection parameters of the neural mesh.

[0009] Step S4: Set various parameters of the fabric texture, including texture density parameter, wrinkle degree parameter, and gloss parameter, and integrate the parameters into the three-dimensional mesh model so that the model can reflect the changes in fabric texture under different parameters;

[0010] Step S5: Use an image translation network to perform texture mapping on the 3D mesh model, mapping the texture information in the initial clothing fabric texture database onto the surface of the 3D mesh model, and optimize the texture mapping effect through iterative learning of the network;

[0011] Step S6: Output the visualized clothing fabric texture design results. Render the processed 3D mesh model to form a clothing fabric texture design scheme that can be viewed intuitively.

[0012] Furthermore, in step S2, the image translation network uses the following model formula for feature extraction: Where F represents the extracted multidimensional feature vector, n is the number of layers in the convolutional neural network, and α i Let W be the weight adjustment coefficient for the i-th layer, σ be the activation function, and W be the weight adjustment coefficient for the i-th layer. i Let X be the weight matrix of the i-th layer, and let b be the input fabric texture data. i Let be the bias vector of the i-th layer.

[0013] Furthermore, in step S3, the neural mesh renderer uses the following model formula when constructing the 3D mesh model: G=ψ(N p E c T s ), where G represents the constructed 3D mesh model, N p E is the set of node parameters for the neural mesh, including node position and node attribute parameters. c T is a set of connection parameters for a neural mesh, describing the connection relationships and strengths between nodes; s These are spatial topology parameters used to define the topological shape of the mesh in three-dimensional space.

[0014] Furthermore, in step S4, during the process of integrating the fabric texture parameters into the three-dimensional mesh model, the texture density parameter D and the wrinkle degree parameter P are processed in the following ways: establishing the correlation between texture density and mesh node distribution, adjusting the distribution density of nodes in the three-dimensional mesh model according to the texture density parameter D; constructing the mapping relationship between wrinkle degree and mesh edge curvature, changing the curvature of the edges in the three-dimensional mesh model according to the wrinkle degree parameter P, so that the three-dimensional mesh model reflects the fabric texture changes under different texture densities and wrinkle degrees.

[0015] Furthermore, in step S5, when the image translation network performs texture mapping, the network parameters are adjusted by optimizing the objective function O, wherein the objective function is: Where m is the number of samples in the texture mapping process, β j Let be the weight coefficient of the j-th sample. The texture vector V mapped in the j-th sample j Compared with the original texture vector Distance metric between them.

[0016] Furthermore, in step S6, when rendering and outputting the visualization results, the influence of the glossiness parameter S on the rendering effect is considered. By establishing the correspondence between the glossiness parameter and the light reflection coefficient, the light reflection coefficient of the surface of the three-dimensional mesh model is adjusted according to the glossiness parameter S, and the texture effects of clothing fabrics with different gloss levels are presented in the output visualization results.

[0017] Furthermore, step S3 includes the following sub-steps:

[0018] Step S3.1: Based on the dimensional information of the multidimensional feature vector, determine the initial node positions of the neural mesh in three-dimensional space. By calculating the mapping relationship between each dimension of the feature vector and the spatial coordinates, the feature information is transformed into the position coordinates of the nodes in three-dimensional space.

[0019] Step S3.2: Based on the structural characteristics of the fabric texture, set the attribute parameters of the neural network nodes, such as the texture influence range parameter of the node, which is used to represent the degree of influence of the node on the texture of the surrounding area.

[0020] Step S3.3: Based on the correlation between feature vectors, establish the connection relationship between neural network nodes. By analyzing the correlation between each dimension in the feature vectors, determine the connection strength and connection method between nodes.

[0021] Step S3.4: Perform topology optimization on the initially constructed neural mesh. By adjusting the node positions and connections, the topology of the neural mesh conforms to the spatial characteristics of the fabric texture.

[0022] Furthermore, step S4 includes the following sub-steps:

[0023] Step S4.1: Obtain the texture density parameter of the clothing fabric texture, associate this parameter with the number and distribution of nodes in the 3D mesh model, and change the texture density performance of the 3D mesh model by adjusting the number of nodes and the distribution spacing.

[0024] Step S4.2: Set the wrinkle degree parameter, calculate the bending angle and curvature change of the edges in the 3D mesh model based on the parameter, and simulate different degrees of fabric wrinkle effect by changing the geometry of the edges;

[0025] Step S4.3: Determine the glossiness parameter and associate it with the material property parameters of the 3D mesh model surface. By adjusting the material property parameters, provide basic settings for different glossiness levels for rendering.

[0026] Step S4.4: Perform a parameter consistency check on the 3D mesh model with the texture parameters set to ensure that there are no conflicts between different parameter settings and that the model can reflect the texture characteristics of the clothing fabric.

[0027] Further, step S5 includes the following sub-steps:

[0028] Step S5.1: Divide the 3D mesh model into multiple texture mapping units, and perform independent texture mapping processing on each unit;

[0029] Step S5.2: For each texture mapping unit, select matching texture data from the initial clothing fabric texture database, and determine the texture data by calculating the similarity between the unit features and the texture data features;

[0030] Step S5.3: Use an image translation network to map the selected texture data onto the corresponding texture mapping unit. Perform a preliminary mapping between the texture data and the unit through the forward propagation calculation of the network.

[0031] Step S5.4: Perform local optimization on the mapped texture effect. By analyzing the transition between adjacent texture mapping units, adjust the network parameters to make the texture mapping effect natural and smooth.

[0032] A big data-driven visualization system for designing garment fabric textures includes:

[0033] The data acquisition and storage unit is used to collect clothing fabric texture data from multi-source big data, and classify and store it according to preset classification rules to form an initial clothing fabric texture database.

[0034] The feature extraction unit is connected to the data acquisition and storage unit. It performs feature extraction on the texture data in the initial clothing fabric texture database based on the image translation network to obtain a multi-dimensional feature vector.

[0035] A mesh construction unit is connected to the feature extraction unit to establish a neural mesh renderer, which maps multi-dimensional feature vectors to three-dimensional space and constructs a three-dimensional mesh model.

[0036] The parameter setting unit, connected to the mesh construction unit, is used to set various parameters of the fabric texture and integrate the parameters into the three-dimensional mesh model.

[0037] A texture mapping unit, connected to the parameter setting unit, performs texture mapping on a 3D mesh model using an image translation network;

[0038] The result output unit is connected to the texture mapping unit to render and output the processed 3D mesh model, forming a visualized clothing fabric texture design result.

[0039] Beneficial Effects: This invention proposes a method and system for visualizing garment fabric texture design using big data. This method and system collect massive amounts of garment fabric texture data from multi-source big data sources and construct an initial database, breaking the limitations of designers relying on limited sample data and personal experience. Based on an image translation network, deep feature extraction of the texture data in the database is performed, accurately mining information such as fashion trends and color preferences contained in the big data, ensuring that designs meet market demands and reducing the risk of unsold products. Addressing the shortcomings of traditional techniques in presenting complex texture effects, the system uses a neural mesh renderer to map the extracted multi-dimensional feature vectors to three-dimensional space, constructing a three-dimensional mesh model with a spatial topological structure. By setting parameters such as texture density, wrinkle degree, and gloss, the system establishes a correlation between parameters and the model structure, realistically simulating the physical properties of the fabric. Simultaneously, the image translation network is used to perform texture mapping and iterative optimization of the three-dimensional mesh model, achieving high-precision and dynamic visualization effects, significantly improving design efficiency and quality. This method and system achieve breakthroughs in both data utilization and technical implementation, bringing innovative changes to garment fabric texture design and meeting the market's demand for diversified, personalized, and high-quality designs. Attached Figure Description

[0040] Figure 1 This is a flowchart of the method steps of the present invention;

[0041] Figure 2 This is a diagram showing the system unit composition of the present invention. Detailed Implementation

[0042] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. The application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0043] like Figure 1 As shown, the method for visualizing clothing fabric texture design using big data includes the following steps:

[0044] Step S1: Construct an initial clothing fabric texture database. Collect clothing fabric texture data including different materials, patterns, and colors from multi-source big data. Classify and store the texture data according to preset classification rules to form an initial database.

[0045] Specifically, in the field of garment fabric texture design, the richness and accuracy of data have a decisive impact on the design outcome. Step S1 aims to collect garment fabric texture data from multiple big data sources, including product display images from online e-commerce platforms, trend materials from fashion news websites, and real-life photos from offline textile exhibitions. The collected data includes garment fabric texture information for different materials (such as cotton, linen, silk, and synthetic fibers), patterns (geometric patterns, floral patterns, abstract patterns, etc.), and colors. Subsequently, according to preset classification rules, such as by material type, pattern style, and color scheme, this texture data is systematically classified and stored in a database, forming an initial database. This initial database provides a large and orderly data foundation for subsequent design work, ensuring that designers can quickly and accurately obtain the required texture data resources.

[0046] In terms of implementation, automated data acquisition programs can be developed, setting parameters such as keywords, time ranges, and data formats to periodically capture garment fabric texture data from various data sources. The acquired data is then labeled, categorized, and stored using professional data management software according to established classification rules. Simultaneously, a data update mechanism is established to regularly supplement and optimize the database, deleting outdated and duplicate data to ensure the timeliness and validity of the data, thus laying a solid data foundation for the entire garment fabric texture design process.

[0047] Step S2: Based on the image translation network, feature extraction is performed on the texture data in the initial clothing fabric texture database. The texture data is processed through a multi-layer convolutional neural network to obtain a multi-dimensional feature vector including texture structure and color distribution features.

[0048] Specifically, the image translation network plays a central role in this step, its main task being to deeply mine key feature information from the fabric texture data. A multi-layered convolutional neural network processes the texture data in the initial database layer by layer. Each layer of the convolutional neural network has a fixed function, enabling analysis of the texture data from different perspectives. During processing, the network can identify the structural information of the texture, such as the direction of the lines and the arrangement of patterns; it can also extract color distribution features, including the dominant hue, color transitions, and color combinations, ultimately obtaining multi-dimensional feature vectors containing these rich features. These multi-dimensional feature vectors accurately describe the core features of each fabric texture data point in digital form, providing targeted and guiding information for subsequent design steps.

[0049] In the specific implementation process, the image translation network must first be trained. A large amount of pre-annotated clothing fabric texture data is used as training samples. By continuously adjusting network parameters, such as the size and number of convolutional kernels and the number of network layers, the network can accurately extract features from various texture data. After training, the texture data from the initial database is input into the trained image translation network. After forward propagation calculation, the network outputs the corresponding multi-dimensional feature vectors. Furthermore, to ensure the accuracy and stability of feature extraction, the network needs to be evaluated and optimized periodically. Based on the evaluation results, network parameters are adjusted or new training data is added to ensure that the network always maintains efficient feature extraction capabilities.

[0050] Step S3: Establish a neural mesh renderer to map the multidimensional feature vectors to three-dimensional space, construct a neural mesh with spatial topology, and form a three-dimensional mesh model corresponding to the texture features of clothing fabric by adjusting the node parameters and connection parameters of the neural mesh.

[0051] Specifically, the neural mesh renderer is a key tool for constructing a 3D model from 2D texture features. This step takes the multi-dimensional feature vectors obtained in the previous step as input and transforms this feature information into a geometric structure in 3D space through a mapping algorithm. In 3D space, by setting the node parameters and connection parameters of the neural mesh, a neural mesh with a spatial topological structure is constructed. The node parameters determine the position, attributes, and other information of the mesh nodes in 3D space, while the connection parameters describe the connection relationships and connection strengths between nodes. Through fine-tuning these parameters, the constructed neural mesh can accurately reflect the features of the clothing fabric texture, ultimately forming a 3D mesh model corresponding to the texture features of the clothing fabric. This 3D mesh model provides the basic framework for subsequent visualization design and effect simulation of the fabric texture.

[0052] In terms of implementation, the approximate range and initial shape of the neural mesh in three-dimensional space are first determined based on the dimensional information and feature distribution of the multi-dimensional feature vectors. Then, for each feature dimension, a mapping relationship is established between it and the node parameters and connection parameters of the neural mesh, transforming the feature information into specific parameter values. During the construction process, professional 3D modeling software or corresponding algorithm programs are used to gradually generate the neural mesh according to the set parameters. Simultaneously, to make the constructed 3D mesh model more accurate, the model needs to be adjusted and optimized multiple times. By comparing the 3D shape of the actual fabric texture, the node parameters and connection parameters are continuously corrected to ensure that the 3D mesh model can realistically and accurately represent the spatial characteristics of the clothing fabric texture.

[0053] Step S4: Set various parameters of the fabric texture, including texture density parameter, wrinkle degree parameter, and gloss parameter, and integrate the parameters into the three-dimensional mesh model so that the model can reflect the changes in fabric texture under different parameters;

[0054] Specifically, clothing fabric textures possess multiple attributes, which can be accurately described and simulated by setting a series of parameters. In this step, the parameters set include texture density, wrinkle level, and gloss level. The texture density parameter controls the density of the fabric surface texture, the wrinkle level parameter simulates the wrinkling effect produced by the fabric under different stress conditions, and the gloss level parameter determines the fabric surface's light reflection characteristics. When integrating these parameters into a 3D mesh model, it is necessary to establish the relationship between the parameters and the model structure. For example, adjusting the texture density parameter changes the distribution density of nodes in the 3D mesh model, allowing the model to present textures with different densities; changing the curvature of the mesh edges according to the wrinkle level parameter simulates various wrinkle shapes of the fabric; and adjusting the material properties of the model surface according to the gloss level parameter lays the foundation for subsequently rendering fabrics with different gloss effects. By integrating these parameters into the 3D mesh model, the model can more realistically and comprehensively reflect the changes in fabric texture under different parameters.

[0055] During implementation, it's crucial to first define the value range and adjustment method for each parameter. For the texture density parameter, a series of values ​​from low to high can be set, allowing for different texture density effects by varying the number and spacing of nodes in the 3D mesh model. For the wrinkle level parameter, the parameter's variation range is determined based on the potential wrinkles in the actual fabric, and different degrees of wrinkles are simulated by calculating the bending angle and curvature change of the mesh edges. For the gloss parameter, its correspondence with properties such as light reflection coefficient and material roughness is established, and different gloss effects are achieved by adjusting these related attribute parameters. After integrating the parameters into the model, it's necessary to check and verify the model to ensure that the parameters are coordinated and do not conflict, guaranteeing that the model accurately reflects the defined fabric texture characteristics.

[0056] Step S5: Use an image translation network to perform texture mapping on the 3D mesh model, mapping the texture information in the initial clothing fabric texture database onto the surface of the 3D mesh model, and optimize the texture mapping effect through iterative learning of the network;

[0057] Specifically, this step again utilizes an image translation network to accurately map the texture information from the initial fabric texture database onto the surface of the 3D mesh model. First, the 3D mesh model is divided into multiple texture mapping units, allowing for refined texture mapping processing for each unit. Then, for each texture mapping unit, matching texture data is selected from the initial database. By calculating the similarity between the unit features and the texture data features, the most suitable texture data is selected. Next, using a trained image translation network, the selected texture data is mapped onto the corresponding texture mapping unit. Through forward propagation calculations, the initial mapping between texture data and units is achieved. To make the texture mapping effect more natural and realistic, local optimization of the mapped texture effect is also necessary. By analyzing the transition between adjacent texture mapping units, the parameters of the image translation network are adjusted to make the transition of texture on the model surface smoother, eliminating splicing marks, and ultimately achieving a high-quality texture mapping effect.

[0058] In practical implementation, the meshing of the 3D mesh model requires determining the appropriate meshing precision based on the model's complexity and texture detail requirements. When selecting matching texture data, a scientific similarity calculation method should be established, comprehensively considering factors such as texture structure, color, and style. When using image translation networks for mapping, the network's input and output parameters should be set appropriately to ensure that texture data can be accurately mapped to the surface of the 3D mesh model. During local optimization, deep learning-based optimization algorithms can be used to automatically analyze texture transitions and adjust network parameters. Alternatively, manual intervention can be combined, with designers fine-tuning local textures based on experience to achieve the best texture mapping effect.

[0059] Step S6: Output the visualized clothing fabric texture design results. Render the processed 3D mesh model to form a clothing fabric texture design scheme that can be viewed intuitively.

[0060] Specifically, after the preceding steps, this step transforms the 3D mesh model into a visually appealing garment fabric texture design. Before outputting the visualization, the 3D mesh model needs to be rendered, taking into full account the impact of gloss parameters on the rendering effect. By establishing the correspondence between gloss parameters and light reflection coefficients, the light reflection coefficients of the 3D mesh model's surface are adjusted according to the set gloss parameters, allowing the model to present fabric effects with different gloss levels. Simultaneously, the model's color, shadows, and other visual effects also need to be optimized and adjusted to ensure that the final visualization accurately reflects the design intent and presents a realistic garment fabric texture effect. Finally, the rendered and optimized 3D mesh model is output in the form of images, videos, or 3D interactive models, providing designers, manufacturers, and consumers with an intuitive and clear display of garment fabric texture designs, facilitating evaluation, communication, and decision-making by all parties.

[0061] In terms of implementation, professional 3D rendering software, such as 3ds Max and Maya, is selected. These software programs possess powerful rendering capabilities and rich material libraries and lighting effect settings. The processed 3D mesh model is imported into the rendering software, and rendering parameters are set according to design requirements, including rendering resolution, ray tracing algorithm, and shadow type. When considering gloss parameters during rendering, the material properties and light reflection parameters of the model surface are precisely adjusted according to the established correspondence. After rendering is completed, the output visualization results are quality checked, such as for texture distortion, color deviation, and inconsistencies in lighting and shadow. If problems are found, the rendering parameters are adjusted or the model is further optimized until a high-quality visualized clothing fabric texture design result is output.

[0062] Preferably, in step S2, the image translation network uses the following model formula for feature extraction: Where F represents the extracted multidimensional feature vector, n is the number of layers in the convolutional neural network, and α i Let W be the weight adjustment coefficient for the i-th layer, σ be the activation function, and W be the weight adjustment coefficient for the i-th layer. i Let X be the weight matrix of the i-th layer, and let b be the input fabric texture data. i Let be the bias vector of the i-th layer. Using this model formula, multi-layer feature extraction is performed on clothing fabric texture data, fully exploring the structural and color features of the texture data.

[0063] Specifically, step S2, the image translation network feature extraction process, achieves deep feature mining of clothing fabric texture data through the construction of model formulas. In terms of implementation, the definitions of each parameter in the model are first clarified. The number of layers in the convolutional neural network determines the depth of texture data feature extraction, the weight adjustment coefficients of each layer are used to balance the contribution of different layers to feature extraction, and the weight matrix and bias vector are the core parameters for feature calculation. In practice, the network is first trained using a large amount of labeled clothing fabric data with texture features. By continuously adjusting these parameters, the network can accurately identify and extract features such as texture structure and color distribution, transforming the original texture data into a multi-dimensional feature vector containing rich information. This process provides a crucial digital foundation for subsequent construction of 3D mesh models and texture design based on feature vectors, ensuring that the extracted features accurately reflect the essential attributes of the fabric texture, thereby improving the accuracy and effectiveness of the entire design method.

[0064] Preferably, in step S3, the neural mesh renderer uses the following model formula when constructing the 3D mesh model: G=ψ(N p E c T s ), where G represents the constructed 3D mesh model, N p E is the set of node parameters for the neural mesh, including node position and node attribute parameters. c T is a set of connection parameters for a neural mesh, describing the connection relationships and strengths between nodes; s These are spatial topology parameters used to define the topological shape of the mesh in three-dimensional space. Using this model formula, a three-dimensional mesh model with spatial structure and properties is constructed based on multi-dimensional feature vectors.

[0065] Specifically, step S3, constructing the 3D mesh model, introduces model formulas to clarify the construction logic of the neural mesh renderer. During implementation, the set of node parameters in the neural mesh encompasses information such as node position and attributes; these parameters determine the mesh's layout and characteristics in 3D space. The set of connection parameters describes the relationships and connection strength between nodes, influencing the overall structure of the mesh. Spatial topology parameters further define the mesh's morphology in 3D space. In practice, the initial positions of nodes in 3D space are first determined based on the information from multi-dimensional feature vectors. Then, node attribute parameters are set in conjunction with the cloth texture structure. Next, node connection relationships are established by analyzing the relationships between feature vectors. Finally, the topology is optimized based on the spatial features of the cloth texture. In this way, two-dimensional texture feature vectors are transformed into a 3D mesh model with spatial structure, laying a spatial foundation for subsequent simulation of cloth texture changes and visualization, and enabling a more realistic reproduction of the cloth's spatial form.

[0066] Preferably, in step S4, during the process of integrating the fabric texture parameters into the three-dimensional mesh model, the texture density parameter D and the wrinkle degree parameter P are processed in the following ways: establishing the correlation between texture density and mesh node distribution, adjusting the distribution density of nodes in the three-dimensional mesh model according to the texture density parameter D; constructing the mapping relationship between wrinkle degree and mesh edge curvature, changing the curvature of the edges in the three-dimensional mesh model according to the wrinkle degree parameter P, so that the three-dimensional mesh model can reflect the fabric texture changes under different texture densities and wrinkle degrees.

[0067] Specifically, step S4, which integrates the fabric texture parameters into the 3D mesh model, is refined, particularly the processing of texture density and wrinkle level parameters. During implementation, the texture density parameter is closely related to the node distribution of the 3D mesh model. By adjusting the number of nodes and their spacing, the density of the texture on the model's surface can be directly altered to simulate the dense texture effects of different fabrics. The wrinkle level parameter, on the other hand, calculates the bending angle and curvature change of the mesh edges, altering the edge geometry to simulate the wrinkling effects of the fabric under different stress conditions. After setting the parameters, a parameter consistency check is performed to ensure that the parameters are coordinated and that parameter conflicts do not prevent the model from accurately reflecting the fabric texture characteristics. This process enables the 3D mesh model to present diverse fabric texture variations based on different parameter settings, enhancing the model's ability to simulate the real state of the fabric.

[0068] Preferably, in step S5, when the image translation network performs texture mapping, the network parameters are adjusted by optimizing the objective function O, wherein the objective function is: Where m is the number of samples in the texture mapping process, β j Let be the weight coefficient of the j-th sample. The texture vector V mapped in the j-th sample j Compared with the original texture vector The distance metric between them. By minimizing the objective function O, the parameters of the image translation network are optimized to improve texture mapping performance.

[0069] Specifically, in step S5, the image translation network texture mapping process adjusts the network parameters by constructing an optimization objective function to improve the texture mapping effect. In practice, the meaning of each parameter in the objective function is first clarified: the number of samples, sample weight coefficients, and texture vector distance metric all contribute to the optimization of the network parameters. In the specific operation, after the 3D mesh model is divided into meshes, matching texture data is selected from the database for each texture mapping unit, and initial mapping is performed using the image translation network. Then, based on the optimization objective function, the distance between the mapped texture and the original texture vector is calculated, and combined with the sample weight coefficients, the network parameters are iteratively adjusted to continuously reduce the difference between the two, making the texture mapping on the model surface more accurate and natural. This process effectively avoids distortion and misalignment problems in texture mapping, ensuring that the final texture mapping effect can realistically reproduce the details and texture of the fabric.

[0070] Preferably, in step S6, when rendering and outputting the visualization results, the influence of the glossiness parameter S on the rendering effect is considered. By establishing the correspondence between the glossiness parameter and the light reflection coefficient, the light reflection coefficient of the surface of the three-dimensional mesh model is adjusted according to the glossiness parameter S, so that different glossiness of clothing fabric texture effects are presented in the output visualization results.

[0071] Specifically, this section examines the impact of glossiness parameters on the rendering effect during step S6, when rendering the visualization result. In terms of implementation, a correspondence between glossiness parameters and light reflection coefficients is first established, clarifying the values ​​of the light reflection coefficients under different glossiness parameters. During the rendering process, based on the set glossiness parameters, the light reflection coefficients of the 3D mesh model surface are precisely adjusted, thereby altering the light reflection characteristics of the fabric surface. In this way, the output visualization result accurately presents the texture effects of different gloss levels of clothing fabrics, such as the sheen of silk and the softness of cotton. Simultaneously, combined with optimizations to model colors, shadows, and other visual effects, the final output visualization design result is more realistic and visually impactful, providing users with an intuitive and accurate visual representation of fabric textures.

[0072] Preferably, step S3 includes the following sub-steps:

[0073] Step S3.1: Based on the dimensional information of the multidimensional feature vector, determine the initial node positions of the neural mesh in three-dimensional space. By calculating the mapping relationship between each dimension of the feature vector and the spatial coordinates, the feature information is transformed into the position coordinates of the nodes in three-dimensional space.

[0074] Step S3.2: Based on the structural characteristics of the fabric texture, set the attribute parameters of the neural network nodes, such as the texture influence range parameter of the node, which is used to represent the degree of influence of the node on the texture of the surrounding area.

[0075] Step S3.3: Based on the correlation between feature vectors, establish the connection relationship between neural network nodes. By analyzing the correlation between each dimension in the feature vectors, determine the connection strength and connection method between nodes.

[0076] Step S3.4: Perform topology optimization on the initially constructed neural mesh. By adjusting the node positions and connections, the topology of the neural mesh conforms to the spatial characteristics of the fabric texture.

[0077] Preferably, step S4 includes the following sub-steps:

[0078] Step S4.1: Obtain the texture density parameter of the clothing fabric texture, associate this parameter with the number and distribution of nodes in the 3D mesh model, and change the texture density performance of the 3D mesh model by adjusting the number of nodes and the distribution spacing.

[0079] Step S4.2: Set the wrinkle degree parameter, calculate the bending angle and curvature change of the edges in the 3D mesh model based on the parameter, and simulate different degrees of fabric wrinkle effect by changing the geometry of the edges;

[0080] Step S4.3: Determine the glossiness parameter and associate it with the material property parameters of the 3D mesh model surface. By adjusting the material property parameters, provide basic settings for different glossiness levels for rendering.

[0081] Step S4.4: Perform a parameter consistency check on the 3D mesh model with the texture parameters set to ensure that there are no conflicts between different parameter settings and that the model can reflect the texture characteristics of the clothing fabric.

[0082] Preferably, step S5 includes the following sub-steps:

[0083] Step S5.1: Divide the 3D mesh model into multiple texture mapping units, and perform independent texture mapping processing on each unit;

[0084] Step S5.2: For each texture mapping unit, select matching texture data from the initial clothing fabric texture database, and determine the texture data by calculating the similarity between the unit features and the texture data features;

[0085] Step S5.3: Use an image translation network to map the selected texture data onto the corresponding texture mapping unit. Perform a preliminary mapping between the texture data and the unit through the forward propagation calculation of the network.

[0086] Step S5.4: Perform local optimization on the mapped texture effect. By analyzing the transition between adjacent texture mapping units, adjust the network parameters to make the texture mapping effect natural and smooth.

[0087] like Figure 2 As shown, the big data-based visualization system for designing garment fabric textures includes:

[0088] The data acquisition and storage unit is used to collect clothing fabric texture data from multi-source big data, and classify and store it according to preset classification rules to form an initial clothing fabric texture database.

[0089] The feature extraction unit is connected to the data acquisition and storage unit. It performs feature extraction on the texture data in the initial clothing fabric texture database based on the image translation network to obtain a multi-dimensional feature vector.

[0090] A mesh construction unit is connected to the feature extraction unit to establish a neural mesh renderer, which maps multi-dimensional feature vectors to three-dimensional space and constructs a three-dimensional mesh model.

[0091] The parameter setting unit, connected to the mesh construction unit, is used to set various parameters of the fabric texture and integrate the parameters into the three-dimensional mesh model.

[0092] A texture mapping unit, connected to the parameter setting unit, performs texture mapping on a 3D mesh model using an image translation network;

[0093] The result output unit is connected to the texture mapping unit to render and output the processed 3D mesh model, forming a visualized clothing fabric texture design result.

[0094] This system and method for visualizing garment fabric texture design using big data addresses the challenge of traditional design methods struggling to extract value from big data. By constructing an initial garment fabric texture database, it collects and categorizes massive amounts of texture data from multiple sources, overcoming the limitations of designers relying on personal experience and limited samples. Simultaneously, by utilizing an image translation network to extract deep features from the texture data in the database, it accurately captures hidden information such as fashion trends and color preferences within the big data. This data is transformed into key elements that guide design, making designs more aligned with market demands and effectively reducing the risk of products becoming unsaleable due to a disconnect with consumer preferences.

[0095] To address the challenge of rendering complex textures using traditional techniques, the system employs a neural mesh renderer to map extracted multidimensional feature vectors into three-dimensional space, constructing a 3D mesh model with a spatial topological structure. By setting parameters such as texture density, wrinkle level, and gloss, and establishing a correspondence between these parameters and the model structure, the system can realistically simulate the physical properties of fabric under different conditions. For example, adjusting the mesh node distribution based on the texture density parameter, changing the mesh edge curvature based on the wrinkle level parameter, and adjusting the light reflection coefficient of the model surface based on the gloss parameter, the system achieves high-precision, dynamic visualization effects, making the design more realistic and persuasive.

[0096] Furthermore, this invention boasts advantages in process continuity and efficiency. From data acquisition and feature extraction to mesh construction, parameter setting, texture mapping, and finally, visualization output, each step is closely integrated and operates collaboratively. The functional units within the system have clearly defined roles yet cooperate effectively. For example, the data acquisition and storage unit provides the data foundation for subsequent steps, the feature extraction unit provides core information for mesh construction, and the texture mapping and output units ensure high-quality presentation of the design results. This integrated design significantly improves design efficiency, reduces manual intervention, and brings innovative changes to the field of garment fabric texture design, powerfully driving the industry towards intelligent and precise development.

[0097] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," "link," and "fix" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal communication between two components. Those skilled in the art will understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0098] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various equivalent 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 method for visualizing clothing fabric texture design using big data, characterized in that: Includes the following steps: Step S1: Construct an initial clothing fabric texture database. Collect clothing fabric texture data including different materials, patterns, and colors from multi-source big data. Classify and store the texture data according to preset classification rules to form an initial database. Step S2: Based on the image translation network, feature extraction is performed on the texture data in the initial clothing fabric texture database. The texture data is processed through a multi-layer convolutional neural network to obtain a multi-dimensional feature vector including texture structure and color distribution features. Step S3: Establish a neural mesh renderer to map the multidimensional feature vectors to three-dimensional space, construct a neural mesh with spatial topology, and form a three-dimensional mesh model corresponding to the texture features of clothing fabric by adjusting the node parameters and connection parameters of the neural mesh. Step S4: Set various parameters of the fabric texture, including texture density parameter, wrinkle degree parameter, and gloss parameter, and integrate the parameters into the three-dimensional mesh model so that the model can reflect the changes in fabric texture under different parameters; Step S5: Use an image translation network to perform texture mapping on the 3D mesh model, mapping the texture information in the initial clothing fabric texture database onto the surface of the 3D mesh model, and optimize the texture mapping effect through iterative learning of the network; Step S6: Output the visualized clothing fabric texture design results. Render the processed 3D mesh model to form a clothing fabric texture design scheme that can be viewed intuitively.

2. The method for visualizing clothing fabric texture design using big data according to claim 1, characterized in that, In step S2, the image translation network uses the following model formula for feature extraction: Where F represents the extracted multidimensional feature vector, n is the number of layers in the convolutional neural network, and α i Let W be the weight adjustment coefficient for the i-th layer, σ be the activation function, and W be the weight adjustment coefficient for the i-th layer. i Let X be the weight matrix of the i-th layer, and let b be the input fabric texture data. i Let be the bias vector of the i-th layer.

3. The method for visualizing clothing fabric texture design using big data according to claim 1, characterized in that, In step S3, the neural mesh renderer uses the following model formula when constructing the 3D mesh model: G=ψ(N p E c T s ), where G represents the constructed 3D mesh model, N p E is the set of node parameters for the neural mesh, including node position and node attribute parameters. c T is the set of connection parameters for a neural mesh, describing the connection relationships and strengths between nodes; s These are spatial topology parameters used to define the topological shape of the mesh in three-dimensional space.

4. The method for visualizing clothing fabric texture design using big data according to claim 1, characterized in that, In step S4, during the process of integrating the fabric texture parameters into the three-dimensional mesh model, the texture density parameter D and the wrinkle degree parameter P are processed in the following ways: establishing the correlation between texture density and mesh node distribution, adjusting the distribution density of nodes in the three-dimensional mesh model according to the texture density parameter D; constructing the mapping relationship between wrinkle degree and mesh edge curvature, changing the curvature of the edges in the three-dimensional mesh model according to the wrinkle degree parameter P, so that the three-dimensional mesh model reflects the fabric texture changes under different texture densities and wrinkle degrees.

5. The method for visualizing clothing fabric texture design using big data according to claim 1, characterized in that, In step S5, when the image translation network performs texture mapping, it adjusts the network parameters by optimizing the objective function O. The objective function is: Where m is the number of samples in the texture mapping process, β j Let be the weight coefficient of the j-th sample. The texture vector V mapped in the j-th sample j Compared with the original texture vector Distance metric between.

6. The method for visualizing clothing fabric texture design using big data according to claim 1, characterized in that, In step S6, when rendering and outputting the visualization results, the influence of the glossiness parameter S on the rendering effect is considered. By establishing the correspondence between the glossiness parameter and the light reflection coefficient, the light reflection coefficient of the surface of the three-dimensional mesh model is adjusted according to the glossiness parameter S, and the texture effects of clothing fabrics with different gloss levels are presented in the output visualization results.

7. The method for visualizing clothing fabric texture design using big data according to claim 1, characterized in that, Step S3 includes the following sub-steps: Step S3.1: Based on the dimensional information of the multidimensional feature vector, determine the initial node positions of the neural mesh in three-dimensional space. By calculating the mapping relationship between each dimension of the feature vector and the spatial coordinates, the feature information is transformed into the position coordinates of the nodes in three-dimensional space. Step S3.2: Based on the structural characteristics of the fabric texture, set the attribute parameters of the neural network nodes, such as the texture influence range parameter of the node, which is used to represent the degree of influence of the node on the texture of the surrounding area. Step S3.3: Based on the correlation between feature vectors, establish the connection relationship between neural network nodes. By analyzing the correlation between each dimension in the feature vectors, determine the connection strength and connection method between nodes. Step S3.4: Perform topology optimization on the initially constructed neural mesh. By adjusting the node positions and connections, the topology of the neural mesh conforms to the spatial characteristics of the fabric texture.

8. The method for visualizing clothing fabric texture design using big data according to claim 1, characterized in that, Step S4 includes the following sub-steps: Step S4.1: Obtain the texture density parameter of the clothing fabric texture, associate this parameter with the number and distribution of nodes in the 3D mesh model, and change the texture density performance of the 3D mesh model by adjusting the number of nodes and the distribution spacing. Step S4.2: Set the wrinkle degree parameter, calculate the bending angle and curvature change of the edges in the 3D mesh model based on the parameter, and simulate different degrees of fabric wrinkle effect by changing the geometry of the edges; Step S4.3: Determine the glossiness parameter and associate it with the material property parameters of the 3D mesh model surface. By adjusting the material property parameters, provide basic settings for different glossiness levels for rendering. Step S4.4: Perform a parameter consistency check on the 3D mesh model with the texture parameters set to ensure that there are no conflicts between different parameter settings and that the model can reflect the texture characteristics of the clothing fabric.

9. The method for visualizing clothing fabric texture design using big data according to claim 1, characterized in that, Step S5 includes the following sub-steps: Step S5.1: Divide the 3D mesh model into multiple texture mapping units, and perform independent texture mapping processing on each unit; Step S5.2: For each texture mapping unit, select matching texture data from the initial clothing fabric texture database, and determine the texture data by calculating the similarity between the unit features and the texture data features; Step S5.3: Use an image translation network to map the selected texture data onto the corresponding texture mapping unit. Through the forward propagation calculation of the network, perform a preliminary mapping between the texture data and the unit. Step S5.4: Perform local optimization on the mapped texture effect. By analyzing the transition between adjacent texture mapping units, adjust the network parameters to make the texture mapping effect natural and smooth.

10. A visualization-based garment fabric texture design system utilizing big data, characterized in that: include: The data acquisition and storage unit is used to collect clothing fabric texture data from multi-source big data, and classify and store it according to preset classification rules to form an initial clothing fabric texture database. The feature extraction unit is connected to the data acquisition and storage unit. Based on the image translation network, it extracts features from the texture data in the initial clothing fabric texture database to obtain a multi-dimensional feature vector. A mesh construction unit is connected to the feature extraction unit to establish a neural mesh renderer, which maps multi-dimensional feature vectors to three-dimensional space and constructs a three-dimensional mesh model. The parameter setting unit, connected to the mesh construction unit, is used to set various parameters of the fabric texture and integrate the parameters into the three-dimensional mesh model. A texture mapping unit, connected to the parameter setting unit, performs texture mapping on a 3D mesh model using an image translation network; The result output unit is connected to the texture mapping unit to render and output the processed 3D mesh model, forming a visualized clothing fabric texture design result.

Images