Artware three-dimensional dynamic display method and system and storage medium

Abstract

The invention relates to the technical field of three-dimensional image processing, in particular to an artware three-dimensional dynamic display method and system and a storage medium. The method comprises the following steps: extracting initial handicraft point cloud data from a flexible handicraft sample, and carrying out flexible handicraft microstructure quantification to generate microstructure quantification data; analyzing a macroscopic deformation mode of the drape according to the microstructure quantitative data, updating a real-time deformation geometric grid, and generating artware deformation geometric grid data; performing surface visual feature mapping and visual feature rendering on the artware deformation geometric grid data to generate an artware three-dimensional dynamic picture; and carrying out three-dimensional interactive dynamic display picture generation on the three-dimensional dynamic picture of the artware to obtain an interactive visual presentation frame. Accurate simulation of real wrinkle deformation and microscopic optical characteristics of the flexible handicraft in a three-dimensional space is realized, so that a highly vivid dynamic display effect is provided.

Description

Technical Field

[0001] The present invention relates to the technical field of three-dimensional image processing, and in particular to a method, system and storage medium for dynamic three-dimensional display of handicrafts. Background Art

[0002] Flexible craft materials inherently possess dynamic properties such as flexibility, drape, folding deformation, and a flowing luster generated by interaction with light. However, existing 3D display systems often simplify these crafts into fixed flat surfaces or display them in only a limited number of preset states, failing to capture the material's natural deformation and light and shadow changes under varying stress conditions. This simplification severely suppresses silk's unique drape, fold dynamics, and the micro-texture of yarn weaving, while failing to accurately present the material's unique glossy fluidity and three-dimensional structure. Traditional 3D display technology, which primarily relies on static scanning and texture mapping, can only restore the basic geometric form and surface color of the craft, failing to simulate the key characteristics of flexible crafts. This results in a noticeably "hard" feel to digital exhibits, lacking the original craft's soft drape, natural fold changes, and dynamic gloss changes under light. Summary of the Invention

[0003] Based on this, the present invention provides a three-dimensional dynamic display method, system and storage medium for handicrafts to solve at least one of the above technical problems.

[0004] To achieve the above object, a three-dimensional dynamic display method for handicrafts includes the following steps:

[0005] Step S1: Perform multi-angle ultra-depth microscopic scanning on the flexible handicraft sample to generate handicraft spectral image data; extract initial handicraft point cloud data based on the handicraft spectral image data, and quantify the microstructure of the flexible handicraft to generate microstructure quantitative data;

[0006] Step S2: Calculate the self-weight deformation gradient generated by the stacked drape based on the quantitative data of the microstructure, analyze the macroscopic deformation mode of the drape wrinkles, and generate macroscopic flexible behavior characteristic data; update the initial handicraft point cloud data in real time using the macroscopic flexible behavior characteristic data to generate the handicraft deformation geometric mesh data;

[0007] Step S3: performing surface visual feature mapping on the deformed geometric mesh data of the handicraft using the microstructure quantification data to generate multi-dimensional surface visual feature data; performing visual feature rendering on the deformed geometric mesh data of the handicraft based on the multi-dimensional surface visual feature data, and performing a three-dimensional dynamic image display of the flexible handicraft to generate a three-dimensional dynamic image of the handicraft;

[0008] Step S4: capturing user interaction information input by the user through the terminal device, and generating a three-dimensional interactive dynamic display picture for the three-dimensional dynamic picture of the handicraft to obtain an interactive visual performance frame.

[0009] Preferably, the present invention further provides a three-dimensional dynamic display system for handicrafts, which executes the three-dimensional dynamic display method for handicrafts as described above, and the three-dimensional dynamic display system for handicrafts comprises:

[0010] The craft microscopic scanning module is used to perform multi-angle ultra-depth microscopic scanning on flexible craft samples to generate craft spectral image data; extract initial craft point cloud data based on the craft spectral image data, and quantify the microstructure of the flexible craft to generate microstructure quantitative data;

[0011] The flexible deformation simulation module is used to calculate the self-weight deformation gradient generated by stacked draping based on the quantitative data of the microstructure, analyze the macro deformation mode of the draping wrinkles, and generate macro flexible behavior characteristic data. The macro flexible behavior characteristic data is used to perform real-time deformation geometry grid updates on the initial handicraft point cloud data to generate handicraft deformation geometry grid data.

[0012] A three-dimensional visual rendering module is used to map the surface visual features of the handicraft's deformed geometric grid data using the microscopic structural quantitative data to generate multi-dimensional surface visual feature data; based on the multi-dimensional surface visual feature data, the module performs visual feature rendering on the handicraft's deformed geometric grid data and displays a three-dimensional dynamic image of the flexible handicraft to generate a three-dimensional dynamic image of the handicraft;

[0013] The user interaction display module is used to capture the user interaction information input by the user through the terminal device, and generate a three-dimensional interactive dynamic display screen for the three-dimensional dynamic screen of the handicraft, so as to continuously update the screen to form a continuous and smooth three-dimensional interactive dynamic display and obtain an interactive visual performance frame.

[0014] Preferably, the present invention further provides a computer-readable storage medium storing a computer program, wherein when the computer program is executed, the method for three-dimensional dynamic display of handicrafts as described above is implemented.

[0015] The present invention uses high-precision spectral image data acquired through multi-angle, ultra-depth-of-field microscopy to accurately capture the microscopic weave structure and surface characteristics of silk threads. This microstructural quantification overcomes the limitations of traditional texture mapping, enabling precise digitization of the internal structure of silk fabrics and accurately reproducing the fabric's unique optical properties and texture. A drape self-weight deformation gradient calculation model, built on this microscopic foundation, enables the system to accurately predict deformation variations in different parts due to differences in material density and structure, thereby presenting the precise drape and wrinkle morphology of flexible crafts in their natural state—a dynamic feature unattainable with traditional static displays. The system can dynamically adjust the three-dimensional form of the craft in response to external forces, no longer limited to a limited number of preset states, but capable of displaying the craft's natural deformation process under any load conditions. The introduction of multi-dimensional surface visual feature data addresses the lack of glossy fluidity in traditional displays, recreating the unique "glittering" effect of silk crafts, allowing viewers to experience the unique optical effects of the material as the viewing angle changes. This coordinated change in macroscopic morphology and microscopic details frees digital exhibits from the limitations of their previous "plastic feel" and presents a visual presentation that is highly similar to the real thing. The real-time updated three-dimensional model is combined with an interactive dynamic display mechanism, allowing viewers to interact naturally and smoothly with digital crafts through intuitive operation, and experience the real reaction of the material under different stress conditions, which greatly enhances the immersiveness and educational value of the digital display. Therefore, a three-dimensional dynamic display method for crafts of the present invention accurately extracts single-filament perturbation geometric data and stacking area identification data by constructing a multi-scale correlation model of the fabric microstructure and macroscopic flexible behavior, and realizes adaptive determination of the critical angle of wrinkles and physics-based calculation of self-weight deformation gradient; adopts dynamic wrinkle local geometric superposition technology and microscopic optical texture generation method, combines the tiny highlight dot matrix formed by light reflection on the yarn surface with the stacking correction displacement data, and realizes accurate distinction and presentation of sharp wrinkles and soft wrinkles; at the same time, it integrates surface light scattering characteristic data with real-time deformation geometric grid, and realizes realistic visual performance and smooth interactive experience of flexible crafts under different lighting conditions through sub-surface scattering effect processing and dynamic shadow calculation. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 Schematic diagram of the process flow of the three-dimensional dynamic display method of handicrafts of the present invention;

[0017] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0018] The following is a clear and complete description of the technical method of the present invention in conjunction with the accompanying drawings. It is obvious that the embodiments described are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative efforts are within the scope of protection of the present invention.

[0019] In addition, the accompanying drawings are merely schematic illustrations of the present invention and are not necessarily drawn to scale. Identical reference numerals in the figures denote identical or similar parts, and thus repetitive descriptions thereof will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities that do not necessarily correspond to physically or logically separate entities. These functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor and / or microcontroller approaches.

[0020] It should be understood that although the terms "first," "second," and the like may be used herein to describe various elements, these elements should not be limited by these terms. These terms are used solely to distinguish one element from another. For example, a first element may be referred to as a second element, and similarly, a second element may be referred to as a first element, without departing from the scope of the exemplary embodiments. The term "and / or" as used herein includes any and all combinations of one or more of the listed associated items.

[0021] To achieve this, please refer to Figure 1 The present invention provides a three-dimensional dynamic display method for handicrafts, comprising the following steps:

[0022] Step S1: Perform multi-angle ultra-depth microscopic scanning on the flexible handicraft sample to generate handicraft spectral image data; extract initial handicraft point cloud data based on the handicraft spectral image data, and quantify the microstructure of the flexible handicraft to generate microstructure quantitative data;

[0023] Step S2: Calculate the self-weight deformation gradient generated by the stacked drape based on the quantitative data of the microstructure, analyze the macroscopic deformation mode of the drape wrinkles, and generate macroscopic flexible behavior characteristic data; update the initial handicraft point cloud data in real time using the macroscopic flexible behavior characteristic data to generate the handicraft deformation geometric mesh data;

[0024] Step S3: performing surface visual feature mapping on the deformed geometric mesh data of the handicraft using the microstructure quantification data to generate multi-dimensional surface visual feature data; performing visual feature rendering on the deformed geometric mesh data of the handicraft based on the multi-dimensional surface visual feature data, and performing a three-dimensional dynamic image display of the flexible handicraft to generate a three-dimensional dynamic image of the handicraft;

[0025] Step S4: capturing user interaction information input by the user through the terminal device, and generating a three-dimensional interactive dynamic display picture for the three-dimensional dynamic picture of the handicraft to obtain an interactive visual performance frame.

[0026] In an embodiment of the present invention, the method for three-dimensional dynamic display of handicrafts includes the following steps:

[0027] Step S1: Perform multi-angle ultra-depth microscopic scanning on the flexible handicraft sample to generate handicraft spectral image data; extract initial handicraft point cloud data based on the handicraft spectral image data, and quantify the microstructure of the flexible handicraft to generate microstructure quantitative data;

[0028] In an embodiment of the present invention, a flexible craft sample first undergoes multi-angle ultra-depth-of-field microscopy scanning, using a confocal laser scanning microscope with a resolution of 5 microns per pixel for data acquisition. During the scanning process, the craft is placed on a precision six-axis rotating platform with a platform accuracy of 0.01 degrees. A set of data is collected every 30 degrees, for a total of 12 angles. Five depth-level scans are performed at each angle, with a layer spacing of 20 microns, to synthesize an ultra-depth-of-field image. The spectral range collected is 400-700 nanometers, with a step size of 10 nanometers, forming a 31-channel spectral image data. Initial craft point cloud data is extracted from the spectral image data, and depth information is acquired using phase-shifted structured light technology, with a point cloud density of 10,000 points per square centimeter. The point cloud data from 12 angles is fused using a 3D point cloud registration algorithm with a registration accuracy of 0.05 mm. Fabric surface texture matching is performed on the point cloud data. Texture features are extracted using a grayscale co-occurrence matrix and matched against a fabric weave pattern library to identify the fabric weave type. Analyze individual yarn diameter (accurate to 0.001 mm), twist (accurate to 10 twists / meter), density (accurate to 5 strands / cm), and microscopic void distribution to generate a microscopic void density map. This ultimately quantifies the fabric structure, generating microscopic structural quantitative data including fabric type code, warp and weft yarn density, yarn diameter, twist, interlacing angle, and microscopic void pore size.

[0029] Step S2: Calculate the self-weight deformation gradient generated by the stacked drape based on the quantitative data of the microstructure, analyze the macroscopic deformation mode of the drape wrinkles, and generate macroscopic flexible behavior characteristic data; update the initial handicraft point cloud data in real time using the macroscopic flexible behavior characteristic data to generate the handicraft deformation geometric mesh data;

[0030] In an embodiment of the present invention, based on the quantitative data of the microstructure, the mechanical parameters of the flexible crafts are extracted through a preset material property association table, including the initial Young's modulus (range 0.5-5.0GPa), Poisson's ratio (range 0.1-0.4) and shear modulus. The initial point cloud data is parameter mapped to form basic mechanical parameter mapping data. Based on the fabric tissue type code and Young's modulus, dynamic drape characteristics analysis is performed, the influencing factors of the local area of ​​the crafts are evaluated, and dynamic drape factor data is generated. The local areas of the vertical stacking relationship in the crafts are identified through spatial depth analysis, and the number of layers (accurately recording the thickness of each layer), the distance between layers and the effective contact area are counted to obtain the stacking geometric characteristic data. The local drape deformation tendency is calculated based on the stacking geometric characteristics and the dynamic drape factor, and the self-weight distribution of each layer in the multi-layer structure is evaluated to obtain the interlayer pressure distribution data. The pressure distribution data is spatially gradient calculated, the self-weight deformation gradient is derived, and the macroscopic deformation pattern of the drape wrinkles is analyzed. A 3D geometric mesh structure is constructed based on the initial point cloud data, with a mesh edge length of 2-5 mm and containing approximately 200,000 triangular facets. A virtual gravity field is applied to the mesh structure, with a surface density ranging from 0.05 to 0.35 g / cm². The initial stress state is analyzed, and the stress-strain tensor and node displacements are calculated. An iterative solution is used to reach equilibrium, generating static deformation morphological data. Vertical displacement corrections are performed on the mesh vertices in the stacked area, local bending angles are detected, and the wrinkle activation state is determined. Detailed wrinkle displacements are then superimposed to generate the artifact's deformed geometric mesh data.

[0031] Step S3: performing surface visual feature mapping on the deformed geometric mesh data of the handicraft using the microstructure quantification data to generate multi-dimensional surface visual feature data; performing visual feature rendering on the deformed geometric mesh data of the handicraft based on the multi-dimensional surface visual feature data, and performing a three-dimensional dynamic image display of the flexible handicraft to generate a three-dimensional dynamic image of the handicraft;

[0032] In this embodiment of the present invention, the area-weighted average method is used to calculate vertex normal vectors with an accuracy of four decimal places to generate artifact texture normal data. Craft illumination characteristic test data is obtained, and a multi-angle spectral reflectance measurement system is used with 72 independently controlled light sources evenly distributed on a hemispherical surface to measure the bidirectional reflectance distribution function. Light reflection and scattering characteristics are analyzed based on the illumination characteristic test data and the microstructure quantitative data, including the average aperture of microscopic voids and warp and weft yarn density. Monte Carlo ray tracing is used to generate surface light scattering characteristic data by emitting 10,000 light rays from each incident direction. Based on the average single yarn diameter, average twist, and average interlacing angle, the surface light scattering characteristic data is used to calculate the microscopic geometric fluctuations of the yarn surface. Light reflection on the yarn surface is simulated, and the microscopic regions that produce highlights are determined. The degree of light obstruction or enhancement at interlacing points is inferred to generate the yarn microscopic optical texture. The artifact texture normal data is perturbed and superimposed with the wrinkle texture normal to generate wrinkle detail normal data. Visual feature mapping is performed on the surface feature data to generate multi-dimensional surface visual feature data, including four channels: basic color texture, specular reflection-roughness, normal map, and sub-surface scattering intensity. The deformable geometric mesh is treated as a geometric entity in the 3D scene, and virtual light sources (one primary light source and three auxiliary light sources) are configured. The multi-dimensional surface visual feature data is loaded, and the reflection, scattering, and shadow effects of light on the fabric surface are calculated. The attenuation changes after light penetrates the fabric are analyzed to achieve sub-surface scattering effects, ultimately generating a 3D dynamic image of the flexible craft.

[0033] Step S4: Capture the user interaction information input by the user through the terminal device, and generate a three-dimensional interactive dynamic display screen for the three-dimensional dynamic screen of the handicraft, so as to continuously update the screen to form a continuous and smooth three-dimensional interactive dynamic display and obtain an interactive visual performance frame.

[0034] In an embodiment of the present invention, user interaction information is captured through the input interface of the terminal device, including touch screen gestures, mouse movements, keyboard keys, and somatosensory device data. Touch screen gesture recognition includes operations such as single-finger drag (controlling perspective rotation), two-finger pinch (controlling zoom), three-finger slide (controlling translation), and four-finger rotation (controlling model rotation). The gesture recognition accuracy is ±2 mm and the delay is less than 16 milliseconds. The mouse input sampling rate is 125Hz, and the displacement conversion accuracy is 0.01 degrees / pixel. The interaction information is smoothed by Kalman filtering with a filter coefficient of 0.85 to eliminate jitter. The system maps user interaction information into operating parameters in the virtual scene, including the perspective transformation matrix, the model transformation matrix, and the ambient lighting parameters. The perspective transformation calculates the camera position and orientation in real time, with a field of view range of 30-90 degrees. When the user drags the craft, the flexible deformation is calculated based on the physical model, and the craft deformation geometric mesh data is updated in real time. The deformation calculation frequency is 30Hz. The dynamic display screen is generated using double buffering technology. After the background rendering is completed, it is exchanged with the foreground display buffer to avoid screen tearing. To ensure a smooth 60 frames per second display, adaptive level-of-detail technology is employed, using a simplified mesh model for long-distance viewing. High-frequency detail is achieved through normal and displacement mapping, while low-frequency deformation is achieved through geometric mesh deformation. The system continuously monitors rendering performance and automatically reduces shadow quality and reflection sampling when the frame rate drops below 45 frames per second. The resulting interactive visual presentation frame undergoes image enhancement processing, including temporal anti-aliasing and ambient occlusion, to achieve high-quality interactive 3D dynamic presentations.

[0035] Preferably, step S1 includes the following steps:

[0036] Constructing an initial three-dimensional geometric point cloud of the flexible craft surface according to the craft spectral image data to obtain initial craft point cloud data;

[0037] Performing fabric surface texture matching based on the initial handicraft point cloud data and the handicraft spectral image data to identify the fabric structure type and obtain fabric structure type identification data;

[0038] The initial handicraft point cloud data and fabric structure type identification data are used to analyze the diameter, initial twist, thread density, and microscopic void size and distribution of individual silk threads, generating a microscopic void density map.

[0039] The microstructure of flexible handicrafts is quantified based on fabric type identification data and micropore density maps to generate microstructure quantitative data. The microstructure quantitative data includes fabric type code, warp and weft yarn density, average single silk thread diameter, average twist, average interlacing angle and average pore size of micro voids.

[0040] In this embodiment of the present invention, a binocular structured light 3D scanner was used to collect omnidirectional spectral image data of flexible crafts. The scanner was equipped with two 5-megapixel industrial cameras, a projector with a resolution of 1920×1080, and a working distance of 400 mm. During the acquisition process, the craft was placed on a rotating platform, which was scanned every 30 degrees of rotation, for a total of 12 sets of data. Depth information was acquired using phase-encoded structured light fringe projection, with a fringe period of 8 pixels and a four-step phase-shifted pattern projected. A point cloud reconstruction algorithm was applied to each set of depth images. The 12 point clouds were stitched together using an iterative closest point registration method with a registration threshold of 0.5 mm and 20 iterations. This ultimately generated a complete initial point cloud data set for the craft, with a point cloud density of 5000 points per square centimeter. The initial point cloud data was segmented into 10 cm × 10 cm local regions. Surface normal features were extracted from each region, and a local normal distribution histogram was constructed. Texture features, including gray-level co-occurrence matrix, local binary pattern, and Fourier spectrum features, were also extracted from the craft spectral image. The extracted features are matched against a pre-established fabric structure feature library, which contains standard feature templates for 20 basic fabric structures, including plain, twill, and satin. A support vector machine classifier is used for matching, with the radial basis function selected as the kernel function and the relaxation factor set to 1.0. When the matching similarity exceeds 85%, the fabric structure type is determined, and fabric structure type identification data is generated, including the structure code and the warp and weft interweaving pattern. In the initial handicraft point cloud data, areas with drastic surface height changes are extracted as silk thread interweaving points. The Gaussian curvature analysis method is used to determine the profile of a single silk thread, and the diameter of a single silk thread is measured. Using the interweaving pattern in the fabric structure type identification data, the center lines of the silk threads are extracted along the warp and weft directions, and the distance between adjacent silk thread center lines is calculated to obtain the silk thread density. By analyzing the local waveform of the point cloud data, the surface texture variation period of the silk thread is extracted, and the initial twist is calculated. The point cloud data was triangulated and the mesh side length was set to 0.05 mm. The grid gap points were identified as microscopic gaps, and the gap size was calculated using the maximum inscribed sphere algorithm. The size and distribution of all gaps were counted to generate a microscopic pore density map with a resolution of 0.1 mm / pixel. Based on the fabric tissue type identification data, the identified fabric tissue structure was converted into a four-digit code according to the international textile coding standard. The first two digits of the code represent the tissue type, and the last two digits represent the variation form. Using the spatial distribution of warp and weft threads in the point cloud data, the number of warp and weft threads per unit area (1 square centimeter) was calculated to obtain the warp and weft yarn density. By statistically averaging all the extracted thread diameter data, the average single thread diameter was obtained, accurate to 0.001 mm. The average twist was calculated using the thread surface texture period, in twists / cm. The spatial direction vectors of the intersecting threads were extracted in the three-dimensional point cloud, and the vector angle was calculated to obtain the average interlacing angle, accurate to 0.1 degrees.All pore areas were extracted from the microscopic pore density map, and the equivalent circle diameter was calculated as the average pore size of the microscopic voids, accurate to 0.001 mm.

[0041] Preferably, performing fabric surface texture matching based on the initial handicraft point cloud data and the handicraft spectral image data to identify the fabric structure type includes:

[0042] Gray-level co-occurrence matrix processing is performed on the craft spectral image data. When the contrast value in the gray-level co-occurrence matrix is ​​greater than 0.3 and the entropy value is greater than 4.5, local texture feature data is extracted.

[0043] Perform pattern matching on local texture feature data using a preset fabric weave pattern library to identify the fabric weave type and obtain preliminary weave type identification data;

[0044] When the preliminary weave type identification data is plain, twill, or satin regular fabric type, the spacing between adjacent silk thread interlacing points and the silk thread floating and sinking period in the initial handicraft point cloud data are quantified to obtain the regular fabric quantitative parameters;

[0045] When the preliminary weave type identification data is a jacquard or embroidered complex fabric type, the thread arrangement variation pattern and color interweaving area boundary within each repeating unit in the initial handicraft point cloud data are marked to construct a complex fabric unit code;

[0046] Based on regular fabric quantization parameters and complex fabric unit coding, fabric structure type identification data including fabric structure type coding, warp and weft yarn density, and average yarn interlacing angle are generated.

[0047] In an embodiment of the present invention, the spectral image data of the handicraft is processed in blocks using a 512×512 pixel resolution, and a grayscale co-occurrence matrix calculation is performed for each block. During the grayscale co-occurrence matrix calculation process, the grayscale level is set to 256, the displacement distance is set to 1 pixel, and the directional angles are 0°, 45°, 90°, and 135°, respectively. Four types of texture statistical features, namely contrast, entropy, correlation, and homogeneity, are extracted from the calculated grayscale co-occurrence matrices in the four directions. When the contrast value of the grayscale co-occurrence matrix in any direction is greater than 0.3 and the entropy value is greater than 4.5, it indicates that the area has obvious fabric texture characteristics. The system marks the area and extracts local texture feature data, including the second-order matrix characteristics of the grayscale co-occurrence matrix, wavelet transform coefficients, and texture directionality indicators, to form a 128-dimensional texture feature vector. The preset fabric weave pattern library is constructed by offline modeling of 500 typical fabric samples. For each fabric sample, five high-definition images at different angles are collected, and texture features are extracted and feature vectors are calculated, forming 2,500 sets of standard feature vectors and corresponding fabric weave type labels. The fabric pattern library includes 20 basic types and their variations, including plain, twill, satin, jacquard, and embroidery. Cosine similarity is calculated between the local texture feature data and the standard feature vectors in the pattern library, with a similarity threshold of 0.85. When the similarity exceeds the threshold, the nearest neighbor classification method is used to determine the most matching fabric type. A Bayesian probability model is also used to calculate the matching confidence. When the confidence exceeds 80%, preliminary weave type identification data is generated, including the weave type name, confidence value, and a description of typical weave features. When the preliminary weave type is identified as a regular weave type such as plain, twill, or satin, surface height variation features are extracted from the initial artifact point cloud data to identify thread interlacing points. Interlacing point features are enhanced using a three-dimensional morphological filtering method, with a filter window size of 3×3×3 and a filter threshold of 15% of the average point cloud height. Interlacing point spacing is determined by calculating the Euclidean distance between adjacent interlacing points, with an accuracy of 0.01 mm. Fast Fourier transforms are used to analyze the periodic variations in point cloud height data and extract the periods of silk thread rise and fall, with a period length accuracy of 0.1 mm. For twill fabrics, the twill angle is measured to an accuracy of 1 degree. For satin fabrics, the distribution density and regularity of satin points are calculated. These parameters are combined to form a regular fabric quantitative parameter data structure, including parameters such as the warp interlacing point spacing, weft interlacing point spacing, float length ratio, and weave unit size. When the initial weave type is identified as a complex fabric type such as jacquard or embroidery, color segmentation is first performed on the craft spectral image data. A region growing algorithm is used to identify different color regions, with a segmentation threshold of 5 between adjacent pixels in color space. Combined with the height information in the point cloud data, a watershed algorithm is used to mark raised areas such as embroidery and accurately locate the boundaries of the color interlacing regions. Spatial overlay analysis of the point cloud data and color regions is performed to identify the boundaries of repeating units, which are generally 5 cm x 5 cm in size.Within each repeating unit, the yarn arrangement variation pattern, including the warp variation sequence, weft variation sequence, and color variation sequence, is recorded to generate a two-dimensional variation matrix. Run-length encoding is used to compress the variation matrix to form a complex fabric unit code. The code length is controlled within 512 bytes to ensure that the code contains complete complex fabric structure information. Fabric weave type identification data is generated based on the regular fabric quantization parameters or complex fabric unit code. For regular fabrics, the fabric weave type is converted into a four-digit code according to international textile standards, such as 1001 for plain weave and 2001 for three-wire twill. Warp density is calculated by counting the number of warp threads per unit length (10 cm), and weft density is calculated by counting the number of weft threads per unit length (10 cm). The density unit is roots / 10 cm. The average yarn interlacing angle is calculated by extracting the three-dimensional direction vectors of the warp and weft threads from the point cloud data. First, the warp and weft direction vectors are determined, and then the angle between the two vectors is calculated. The average angle of the interlacing points is taken as the average yarn interlacing angle, accurate to 0.1 degree. The final fabric type identification data is stored in the form of structured data, including fabric type code, warp and weft yarn density values, interlacing angle values, and fabric structure description text.

[0048] Preferably, in step S2, calculating the self-weight deformation gradient generated by the stacked draping according to the quantitative data of the microstructure, and analyzing the macroscopic deformation mode of the draping wrinkles includes:

[0049] The initial Young's modulus, Poisson's ratio and shear modulus of the flexible craft are extracted from the microstructure quantitative data through the preset material property association table, and the initial craft point cloud data is parameter mapped to obtain the basic mechanical parameter mapping data;

[0050] Based on the fabric type code in the microstructure quantitative data and the Young's modulus in the basic mechanical parameter mapping data, dynamic drape characteristics analysis is performed, and the corresponding local area impact factor evaluation of the flexible craft is performed to obtain dynamic drape factor data;

[0051] Based on the dynamic drape factor data, the initial craft point cloud data is used to calculate the self-weight deformation gradient caused by the stacking drape in the local area to obtain the drape gradient quantitative data;

[0052] Based on the Young's modulus and Poisson's ratio in the basic mechanical parameter mapping data, the minimum bending angle value for the flexible craft to produce stable dynamic wrinkles when bent is set to obtain the wrinkle critical angle value; among them, when the Young's modulus is less than 2.0 GPa and the Poisson's ratio is greater than 0.28, the wrinkle critical angle is set to 15°-25°, when the Young's modulus is in the range of 2.0-3.5 GPa and the Poisson's ratio is in the range of 0.20-0.28, the wrinkle critical angle is set to 25°-40°, and when the Young's modulus is greater than 3.5 GPa and the Poisson's ratio is less than 0.20, the wrinkle critical angle is set to 40°-65°, and the wrinkle critical angle value is obtained;

[0053] The macro deformation pattern of the drape fold is analyzed based on the quantitative data of the drape gradient and the critical angle of the fold to generate the macro deformation pattern data;

[0054] The macro deformation mode data, basic mechanical parameter mapping data, overhang gradient quantification data and wrinkle critical angle values ​​are associated with the macro deformation properties to generate macro flexible behavior characteristic data.

[0055] In this embodiment of the present invention, a preset material attribute association table is constructed by performing standard mechanical tests on 500 typical textile materials. Each material contains microstructure parameters and corresponding mechanical performance parameters. The test was performed using an electronic universal testing machine for uniaxial tensile testing, with a loading rate of 5 mm / min, a test temperature of 23±2°C, and a relative humidity of 65±3%. Based on the quantitative microstructure data of the artifact to be analyzed, the most similar fabric type parameter combination was retrieved from the material attribute association table. When the fabric type code in the quantitative microstructure data is plain weave (1001), the average yarn diameter is 0.15 mm, and the warp and weft yarn density is 300 yarns / 10 cm, the corresponding initial Young's modulus is 1.8 GPa, the Poisson's ratio is 0.32, and the shear modulus is 0.68 GPa. The extracted mechanical parameters are mapped to each grid cell of the initial artifact point cloud data at a mapping resolution of 5 mm × 5 mm, forming a basic mechanical parameter mapping data matrix covering the entire artifact surface. Based on the fabric type encoding and basic mechanical parameter mapping data from the microstructure quantitative data, a dynamic drape model for a localized region of a craft was constructed. The craft surface was first divided into 10 cm x 10 cm analysis cells, and drape characteristics were calculated for each analysis cell. The drape calculation employed a modified FAST fabric testing method, measuring the drape produced by the craft under its own weight to an accuracy of 0.1 mm. For regions with a Young's modulus less than 2.0 GPa, the drape influence factor was set to 1.2-1.5; for regions with a Young's modulus between 2.0 and 3.5 GPa, the drape influence factor was set to 0.5-0.8; and for regions with a Young's modulus greater than 3.5 GPa, the drape influence factor was set to 0.5-0.8. Furthermore, considering the influence of fabric type, a correction factor of 1.0 was applied to plain weave regions, 0.9 to twill weave regions, and 0.8 to satin weave regions. Dynamic drape factor data was generated, including the drape values ​​and influence factors for the localized regions. Based on dynamic drape factor data, the self-weight deformation gradient resulting from stacked drape is calculated for the initial artifact point cloud data. Finite element analysis is used to convert the point cloud data into a triangular mesh model with a 2 mm mesh element side length, constructing approximately 50,000 triangular elements. A fixed constraint is set at the top edge of the model, and a vertical downward force field with a gravitational acceleration of 9.8 m / s² is applied. Based on the Young's modulus, Poisson's ratio, and shear modulus from the basic mechanical parameter mapping data, the displacement and strain distribution of each mesh element under gravity are calculated. The model is iteratively calculated to reach static equilibrium, and the displacement vector of each mesh node is recorded. The displacement gradient is then calculated at 5 cm intervals along the vertical direction. For multi-layered fabrics, the interlayer friction coefficient is considered, with a range of 0.15-0.35. This results in quantitative drape gradient data, containing the displacement gradient values ​​for each region, expressed in mm / cm.Based on the Young's modulus and Poisson's ratio in the basic mechanical parameter mapping data, the critical angle for flexible crafts to produce stable dynamic wrinkles is set. For craft areas with a Young's modulus less than 2.0GPa and a Poisson's ratio greater than 0.28, such as silk and light cotton fabrics, their molecular chain structure is highly flexible and has strong deformation ability, and the critical wrinkle angle is set to 15°-25°. The specific calculation formula is: critical wrinkle angle = 15° + 10° × (1-Poisson's ratio / 0.4). For craft areas with a Young's modulus in the range of 2.0-3.5GPa and a Poisson's ratio in the range of 0.20-0.28, such as thick cotton and linen fabrics, the critical wrinkle angle is set to 25°-40°, and the calculation formula is: critical wrinkle angle = 25° + 15° × (1-(Young's modulus - 2.0) / 1.5). For areas of handicrafts with a Young's modulus greater than 3.5 GPa and a Poisson's ratio less than 0.20, such as coarse linen and thick wool fabrics, the critical wrinkle angle is set to 40°-65°, and the calculation formula is: critical wrinkle angle = 40° + 25° × (1-(0.20-Poisson's ratio) / 0.20). Through these calculations, the critical wrinkle angle value of each area is obtained. Based on the drape gradient quantitative data and the critical wrinkle angle value, the macroscopic deformation pattern of the handicraft drape folds is analyzed. First, a three-dimensional curvature field of the handicraft surface is constructed, and the Gaussian curvature and average curvature are calculated using discrete differential geometry methods. The curvature calculation radius is 10 mm. When the local curvature change exceeds the curvature value corresponding to the critical wrinkle angle, it is marked as a wrinkle formation area. Based on the fold density, deformation patterns were classified into four categories: uniformly distributed folds (fold spacing variation coefficient less than 0.2), centrally clustered folds (fold density center-to-edge decay rate greater than 0.5), striped folds (fold directionality index greater than 0.7), and irregular folds (fold distribution entropy greater than 3.0). Image processing methods were used to extract fold line features, including fold line length, width, depth, direction, and distribution density. Comprehensive analysis was then performed to determine the macroscopic deformation pattern data of the artifact under draping. Multidimensional correlation analysis was performed on the macroscopic deformation pattern data, basic mechanical parameter mapping data, drape gradient quantification data, and critical fold angle values ​​to generate macroscopic flexible behavior characteristic data. Principal component analysis was used to extract the main influencing factors and establish a predictive model for the artifact's flexible behavior. For each macroscopic deformation pattern, characteristic parameters extracted included fold formation time (in seconds), fold stability depth (in millimeters), fold recovery rate (in percentage), and critical deformation angle (in degrees). For different macroscopic deformation modes, a library of wrinkle morphological evolution patterns was constructed, recording data from the entire process of wrinkle formation to stabilization. The resulting macroscopic flexible behavior characteristic data was organized into a hierarchical structure, including a static characteristic layer (basic mechanical parameters), a dynamic characteristic layer (drape gradient data), a critical characteristic layer (critical wrinkle angle), and a morphological characteristic layer (macroscopic deformation mode), forming a complete macroscopic flexible behavior characteristic data set.

[0056] What is particularly important is that the self-weight deformation gradient generated by the stacking and draping of the local area is calculated based on the initial craft point cloud data based on the dynamic draping factor data:

[0057] Perform spatial depth analysis and surface adjacency detection on the initial handicraft point cloud data to identify local areas with vertical stacking relationships in the initial handicraft point cloud data and obtain stacking area recognition data;

[0058] Based on the stacking area recognition data, the number of layers, inter-layer distance and effective contact area of ​​each identified stacking area in the initial craft point cloud data are quantified to obtain the stacking geometric characteristic data;

[0059] Calculate the local drape deformation tendency in each stacking area due to the inherent flexibility of the material based on the stacking geometric characteristic data and the dynamic drape factor data to obtain the local drape tendency data;

[0060] The weight distribution of each layer in the multi-layer structure and the vertical pressure it exerts on the layers below are evaluated based on the stacking geometric characteristics data and the local overhang tendency data, and the inter-layer pressure distribution data are obtained;

[0061] The spatial gradient calculation and normalization of the interlayer pressure distribution data are performed to deduce the self-weight deformation gradient caused by the stacking overhang in the local area and obtain the quantitative data of the overhang gradient.

[0062] In an embodiment of the present invention, the initial handicraft point cloud data is first subjected to octree spatial segmentation, and the segmentation accuracy is set to 2 mm. The K nearest neighbor algorithm is used to estimate the normal vector of the point cloud, the K value is set to 20, and the normal vector calculation radius is 5 mm. The surface continuity is judged by normal vector consistency analysis, and when the angle between the normal vectors of two adjacent points is less than 15 degrees, they are considered to be the same plane. By setting the Z-axis (vertical direction) projection distance threshold to 5 mm, the area with vertical projection overlap in the point cloud is identified. The overlapping area is subjected to Euclidean clustering segmentation, the cluster distance threshold is set to 10 mm, and the minimum point number threshold is 100 points to obtain preliminary stacking area candidates. The main plane of each candidate area is extracted by the RANSAC plane fitting algorithm, and the plane fitting threshold is set to 2 mm. When the distance between the two planes is less than 30 mm and the angle between the plane normal vectors is less than 30 degrees, it is confirmed to be a valid stacking relationship. Finally, the stacking area recognition data is generated, which includes the spatial boundaries, point cloud indexes and stacking relationship descriptions of each stacking area. Geometric characteristics quantification is performed on each identified stacking area. First, the point cloud data was sliced ​​horizontally at 2 mm intervals along the vertical direction. The appearance and disappearance patterns of points in consecutive slices were used to identify material layer boundaries. Interlayer boundaries were marked when the density change between two adjacent slices exceeded 50%. The number of layers in each stacking region was counted, and the thickness of each layer was recorded to the nearest 0.1 mm. The interlayer distance was calculated by calculating the Euclidean distance between the center planes of two adjacent layers, to the nearest 0.5 mm. The effective contact area was quantified using a convex hull algorithm to calculate the horizontal projection area of ​​each layer. The interlayer contact area was then determined using a spatial intersection operation to the nearest 1 square millimeter. For irregular stacking morphologies, the alpha shape algorithm was used to optimize the contact area calculation, with an alpha value of 10 mm. A stacking geometry data structure was constructed, containing a complete geometric description of each stacking region, combining the number of layers, thickness, interlayer distance, and contact area data. The local drape deformation tendency was calculated based on the stacking geometry data and dynamic drape factor data. For each stacking region, the Young's modulus, Poisson's ratio, and shear modulus at the corresponding location in the basic mechanical parameter mapping data were extracted. When the handicraft material is silk (Young's modulus 1.5GPa, Poisson's ratio 0.35), its inherent drape coefficient is set to 1.8; when the material is cotton and linen blend (Young's modulus 2.8GPa, Poisson's ratio 0.25), the drape coefficient is set to 1.2; when the material is woolen cloth (Young's modulus 3.8GPa, Poisson's ratio 0.18), the drape coefficient is set to 0.7. Combined with the sag influencing factors in the dynamic drape factor data, the natural drape curvature of each layer without external force is calculated, with the curvature accuracy of 0.001 / mm. Based on the layer shape data in the stacking geometric characteristics, the natural drape curve of each layer is analyzed by Bezier surface fitting, and the drape curve is described by 10 control points. A drape simulation calculation is performed on the center line of each layer to obtain local drape tendency data, including the predicted drape displacement value and the main drape direction of each point. The interlayer pressure distribution is evaluated based on the stacking geometric characteristics data and the local drape tendency data.First, the areal density of each layer is calculated in grams per square centimeter. This density is derived from the fabric weave type, warp and weft density, and average yarn diameter from the microstructure quantitative data. The calculation formula involves the yarn density and interweaving coefficient. For silk, the typical areal density is 0.08 g / square centimeter; for cotton and linen blends, 0.15 g / square centimeter; and for wool, 0.25 g / square centimeter. The layer weight is calculated based on the area and density of each layer. The stacked structure is then simulated using a spring-mass network model with a mesh resolution of 5 mm × 5 mm. The spring stiffness is calculated from the Young's modulus. Force equilibrium is calculated iteratively to determine the force at each grid point. Interlayer pressure is calculated using the normal force at the interlayer contact point. A pressure distribution matrix is ​​constructed with a resolution of 5 mm × 5 mm and pressure units in Pascals. This generates interlayer pressure distribution data, which includes the pressure exerted by each layer and the pressure on the underlying layer. Spatial gradient calculation and normalization are performed on the interlayer pressure distribution data. The finite difference method is used to calculate the spatial gradient of the pressure field, with a differential step of 5 mm. The pressure change rate in the X, Y, and Z directions is calculated respectively, with the unit being Pascal / mm. According to the predicted value of the overhang displacement in the local overhang tendency data, combined with the pressure gradient and basic mechanical parameters, the linear elastic mechanics model is used to calculate the actual displacement response of each point. The displacement in the calculation formula is proportional to the pressure, inversely proportional to the Young's modulus, and related to local geometric features (such as thickness and area). The obtained displacement field is normalized, the maximum displacement is normalized to 1.0, and other positions are scaled proportionally. At the same time, the displacement gradient is calculated, that is, the rate of change of displacement within a unit distance, in units of mm / cm. Finally, the overhang gradient quantitative data is generated, including the normalized displacement field, the displacement gradient field, and the predicted three-dimensional deformation surface, with a data accuracy of 0.01 mm, and a deformation gradient thermal map is drawn to intuitively display the deformation trend of the craft under the action of its own weight.

[0063] Preferably, in step S2, performing real-time deformation geometric mesh updating on the initial handicraft point cloud data using the macro-flexible behavior characteristic data includes:

[0064] Constructing a three-dimensional geometric grid structure of the flexible craft based on the initial craft point cloud data;

[0065] Based on the three-dimensional geometric grid structure of the flexible craft, a virtual gravity field is loaded to obtain the initial deformation grid;

[0066] Based on the macroscopic flexible behavior characteristic data, the initial stress state of the initial deformation grid is analyzed to obtain the initial stress and strain tensor of the unit;

[0067] Calculate the initial displacement of each node in the initial deformation grid according to the initial stress and strain tensor of the unit, and generate a set of initial node displacement vectors;

[0068] Perform iterative solution processing based on the preliminary node displacement vector set until the preset convergence judgment conditions are met to generate static deformation morphological data;

[0069] The static deformation morphology data is corrected for the vertical displacement of the grid vertices in the stacking area by using the macroscopic flexible behavior characteristic data to obtain the stacking corrected displacement data;

[0070] Performing local wrinkle displacement superposition processing on static deformation morphology data by stacking corrected displacement data to generate local wrinkle displacement superposition data;

[0071] Based on the stacked corrected displacement data, the static deformation morphology data is dynamically superimposed on the local wrinkle displacement superposition data, and the real-time deformation geometric grid is updated to generate the handicraft deformation geometric grid data.

[0072] In an embodiment of the present invention, the initial handicraft point cloud data is constructed into a three-dimensional geometric grid structure using the March cube algorithm, and voxel processing with a grid resolution of 2 mm is adopted. Point cloud preprocessing includes statistical outlier filtering, and the filter parameters are set to a standard deviation multiple of 1.5 and 20 neighbor points. The grid topology optimization adopts a quadratic error metric grid simplification algorithm, and the number of facets is controlled at 150,000 to ensure a balance between geometric accuracy and computational efficiency. To enhance the detail performance of the wrinkle area, an adaptive grid subdivision strategy is adopted, and loop subdivision is performed in areas where the curvature change is greater than 0.05 / mm, with a subdivision depth of 1 level. Laplace smoothing and edge flipping operations are used for grid quality control, and the edge flipping angle threshold is 15 degrees. Boundary preservation processing is used for edge areas to keep the boundary curve accuracy within 0.5 mm. The final grid storage adopts a half-edge data structure, which includes a vertex coordinate array (X, Y, Z floating point values ​​accurate to 0.001 mm), a face index array, edge association information, and a vertex normal vector array. When loading a virtual gravity field based on a three-dimensional geometric mesh structure, the gravity vector is set to (0, 0, -9.8) m / s². The mass distribution of the artifacts is calculated based on quantitative microstructural data. The regional density for silk fabrics is set to 80 g / m², the regional density for cotton fabrics is set to 150 g / m², and the regional density for wool fabrics is set to 250 g / m². The mass distribution uses a vertex-lumped mass model, with the mass of each vertex calculated as 1 / 3 of its control domain area multiplied by the surface density. The suspension points are set using geometric boundary detection, with the continuous edge vertex with the highest Z coordinate as the fixed point. Rigid constraints are set at these fixed points, and displacement is set to zero. Preset suspension modes include single-point, two-point, multi-point, and tiled modes to demonstrate different suspension states. For single-point suspension, a rigid constraint is set at the highest point; for two-point suspension, rigid constraints are set at the two vertices with the highest Z coordinates; for multi-point suspension, rigid constraints are set at eight evenly distributed vertices along the top edge; and for tiled mode, semi-rigid constraints are set at all vertices on the top surface. After loading the gravity field, the initial deformable mesh data structure is generated. The macroscopic flexible behavior characteristic data is used to analyze the initial stress state of the starting deformation grid. The material model adopts the anisotropic linear elastic model. The warp Young's modulus and weft Young's modulus are set according to the fabric structure type. The warp Young's modulus of plain fabric is 2.5GPa and the weft Young's modulus is 2.2GPa; the warp Young's modulus of twill fabric is 2.8GPa and the weft Young's modulus is 2.4GPa; the warp Young's modulus of satin fabric is 2.1GPa and the weft Young's modulus is 1.8GPa. The shear modulus is set to 0.4 times the Young's modulus. The strain calculation adopts the nonlinear Green-Lagrange strain tensor to consider the large deformation effect. The strain tensor E is calculated as E=(F TFI) / 2, where F is the deformation gradient tensor, I is the unit tensor, and T is the matrix transpose identifier. The stress tensor S is calculated by the generalized Hooke's law, S=C:E, where C is the fourth-order stiffness tensor, constructed by Young's modulus, Poisson's ratio and shear modulus. For each triangular element, the stress and strain state is calculated at three Gaussian integration points and then interpolated to the element vertices through shape functions. Finally, the element initial stress and strain tensor data set is generated, which contains 6 independent stress components and 6 independent strain components for each element. The element initial stress and strain tensor is used to calculate the preliminary displacement of each node of the starting deformation mesh. Using a physics-based finite element solution framework, the global stiffness matrix is ​​assembled by the stiffness matrices of each element. The element stiffness matrix K e Calculated as K e =∫B T ·D·B·dV, where B is the strain-displacement matrix, D is the material stiffness matrix, and the integration is performed on the unit volume. The global equation group is K·u=F, where K is the global stiffness matrix, u is the unknown displacement vector, and F is the external force vector. In order to deal with the suspension point constraints, the Lagrange multiplier method is used to introduce the constraint equation G·u=c, where G is the constraint matrix and c is the constraint value. The extended system equation adopts the block matrix form to solve the displacement vector u and the Lagrange multiplier vector λ. The sparse matrix direct solver PARDISO is used for solution, and the matrix reordering adopts the approximate minimum degree algorithm, and the filling factor is controlled within 2.5. The preliminary displacement vector obtained contains the displacement components of each mesh vertex in the X, Y, and Z directions, with an accuracy of 0.001 mm, forming a preliminary displacement vector set of the node. The preliminary displacement vector set of the node needs to be solved iteratively to reach convergence. The dynamic implicit integration method is used to introduce the mass matrix M and the damping matrix C to construct the dynamic equation in is the acceleration vector, is the velocity vector. The time integration uses the Newmark-β method, with the β parameter set to 0.25 and the γ parameter set to 0.5 to ensure unconditional stability. The time step is initially set to 0.01 seconds and automatically adjusted based on the system stiffness. The iteration termination criteria are set when the norm of the residual force is less than 1E-5 Newtons, the norm of the displacement change between two consecutive iterations is less than 0.005 mm, or the maximum number of iterations, 30, is reached. During each iteration, the deformed geometry and stiffness matrix are updated, accounting for geometric nonlinear effects. The local Newton iteration uses a line search strategy, with the search direction determined by the conjugate gradient method and the step size factor calculated by the backplane tracking method to ensure monotonically decreasing energy. For regions with large torsions, a rotational inertia term is introduced to prevent excessive deformation. When the iteration converges, the final equilibrium position of each node is recorded to generate static deformation morphological data. Mesh vertices in overlapping regions in the static deformation morphological data are corrected for vertical displacement. Overlapping regions are identified using a collision detection algorithm accelerated by a spatial octree, with a collision distance threshold set to 1.2 times the fabric thickness. For the identified vertex pairs in the stacking area, the interlayer contact pressure and compression are calculated. The pressure calculation formula is P = K·d·exp(-d / d0), where K is the contact stiffness coefficient, ranging from 50 to 200 Newtons / square meters, d is the penetration depth, and d0 is the characteristic compression distance, which is 0.5 times the thickness of the fabric. The displacement correction δ is nonlinearly related to the pressure, δ = d·(1-exp(-P / P0)), where P0 is the characteristic pressure, which is 50 Pascals. The correction strategy is layered processing, adjusting layer by layer from top to bottom, and the Z-direction displacement correction of each layer is calculated according to the compression model to ensure that there is no penetration between layers. The width of the displacement smooth transition area is set to 30 mm, and the transition function uses the cubic Hermite interpolation polynomial to ensure C 1 Continuity. For multi-layer superposition areas, the superposition compression rate decreases as the number of layers increases, and the compression rate of the nth layer is the basic compression rate multiplied by (1-0.1·(n-1)). After correction, the stacking correction displacement data is generated. The stacking correction displacement data is used for local wrinkle displacement superposition processing of static deformation morphological data. The wrinkle area identification is based on the principal curvature analysis. When the maximum principal curvature k1 is greater than 0.05 / mm and the minimum principal curvature k2 is less than -0.05 / mm, it is judged as a wrinkle area. The wrinkle line extraction adopts the ridge detection algorithm and is obtained by tracking the principal curvature direction field. The wrinkle characteristic parameters include wrinkle depth h, wrinkle width w and wrinkle length l, which are estimated by curvature integral. The wrinkle morphological function is described by the modified Gaussian function, and the expression is f(r) = h·exp(-(r / w) 2)·(1-α·r / l), where r is the vertical distance to the wrinkle line, and α is the shape modulation parameter with a value range of 0.2-0.8, which is determined by the hardness of the fabric. For silk material, α is set to 0.8, for cotton fabric, α is set to 0.5, and for wool fabric, α is set to 0.3. The wrinkle wave depth h is related to the material thickness t and the curvature k, h = t·(1-exp(-β·k)), where β is the material correlation coefficient, which is 1.8 for silk, 1.2 for cotton fabric, and 0.8 for wool fabric. For each mesh vertex, its distance to all wrinkle lines is calculated, and the additional displacement is calculated based on the nearest wrinkle line feature, with the displacement direction along the vertex normal vector. Brunstein polynomial interpolation is used to achieve smooth transition in multi-wrinkle areas, and local wrinkle displacement superposition data is generated. The stacked corrected displacement data and local wrinkle displacement superposition data are used for dynamic wrinkle local geometric superposition and real-time update of static deformation morphological data. Dynamic wrinkle simulation uses a simplified physical model, with the wrinkle frequency f set between 0.5 and 2 Hz, and the amplitude variation range within ±20% of the base amplitude. The time-varying wrinkle displacement δ(t) is calculated as δ(t) = δ0·(1+0.2·sin(2πft)), where δ0 is the static wrinkle displacement. Real-time updates utilize GPU-accelerated geometry shader technology, directly modifying vertex positions within the rendering pipeline to avoid data transfer bottlenecks between the CPU and GPU. Mesh data is organized into vertex buffer objects and index buffer objects. Vertex attributes include position, normal, texture coordinates, and displacement. Displacements are synthesized using weighting factors that account for cascading effects, wrinkle effects, and dynamic changes. To improve performance, frustum culling and level-of-detail control are employed. A simplified mesh model is used when the viewing distance is greater than 3 meters, and geometric interpolation is used for level-of-detail transitions. The frame rate is maintained at above 30 frames per second, with displacement updates occurring 10 times per second, and smooth transitions achieved through interpolation. The resulting deformable geometric mesh data contains a complete topology and a real-time update mechanism.

[0073] It is particularly important to perform local wrinkle displacement superposition processing on static deformation morphology data by stacking corrected displacement data. Specifically:

[0074] detecting the bending angle of each local area in the initial deformed mesh according to the stacked corrected displacement data, and generating a local bending angle;

[0075] The wrinkle activation condition is determined according to the local bending angle and the wrinkle critical angle value in the macro-flexible behavior characteristic data. When the local bending angle does not exceed the wrinkle critical angle value in the macro-flexible behavior characteristic data, it is determined to be in an inactive state. When the local bending angle exceeds the wrinkle critical angle value in the macro-flexible behavior characteristic data, it is determined to be in an active state, and the wrinkle activation state data is generated.

[0076] When the wrinkle activation state data is in the activated state, the preset small surface displacement and wrinkle formation state pattern are matched according to the microstructure quantitative data and the local bending angle to obtain the matching wrinkle unit data; wherein the wrinkle formation state pattern includes a sharp wrinkle formation state and a soft wrinkle potential state;

[0077] The matched wrinkle unit data is used as the superimposed displacement details and superimposed on the corresponding curved area of ​​the current grid to obtain the local wrinkle displacement superimposed data.

[0078] In an embodiment of the present invention, the bending angle of each local area in the starting deformation grid is detected based on the stacked corrected displacement data. The grid is divided into square analysis areas with a side length of 10 mm, and 25 sampling points are extracted from each analysis area. For each analysis area, a local coordinate system is constructed, with the origin being the center point of the area, the Z axis along the direction of the average normal vector of the area, and the X axis and Y axis determined by Schmidt orthogonalization. The regional curvature characteristics are calculated using the principal curvature analysis method. First, the curvature tensor of each sampling point is constructed and obtained by fitting a quadratic surface. The eigenvalue decomposition of the curvature tensor obtains the principal curvatures κ1 and κ2 and their corresponding main directions. The bending angle θ is calculated by multiplying the principal curvature by the area size, θ = arcsin(κ1·L), where L is the characteristic length of the area, which is 10 mm. For strongly curved areas, multi-scale analysis is used to capture bending features of different scales, with a scale range of 5-20 mm. In particular, for soft fabrics with a Young's modulus less than 2.0 GPa, a material correction factor of 1.2 is introduced into the bending angle calculation; for hard fabrics with a Young's modulus greater than 3.5 GPa, the correction factor is set to 0.8. Finally, local bending angle data is generated, accurate to 0.1 degree. The local bending angle is compared with the wrinkle critical angle value in the macroscopic flexible behavior characteristic data to perform wrinkle activation condition judgment. The judgment process adopts a regional adaptive threshold method, taking into account material heterogeneity. For the silk area (Young's modulus 1.5 GPa, Poisson's ratio 0.35), the wrinkle critical angle is 18 degrees; for the cotton fabric area (Young's modulus 2.5 GPa, Poisson's ratio 0.25), the wrinkle critical angle is 32 degrees; for the woolen area (Young's modulus 4.0 GPa, Poisson's ratio 0.18), the wrinkle critical angle is 55 degrees. When the local bending angle is less than 80% of the critical angle, the state is determined to be inactive; when the local bending angle is between 80% and 100% of the critical angle, the state is determined to be pre-activated; and when the local bending angle exceeds the critical angle, the state is determined to be active. For stacked regions, the critical angle is multiplied by a correction factor of 0.85 due to interlayer interactions. The activation state determination also considers the bending continuity of adjacent regions. When three or more of the eight surrounding adjacent regions are in an active state, the activation threshold for the current region is lowered by 5 degrees. The determination results form wrinkle activation state data, which contains the activation state identifier (0 for inactive, 1 for pre-activated, and 2 for activated) and the activation probability value (a floating-point number between 0 and 1) for each analyzed region. When the wrinkle activation state data indicates an active state, wrinkle formation state pattern matching is performed. The preset wrinkle formation state pattern library was established through physical simulation and physical scanning and contains 500 typical wrinkle templates, each with a 3D geometric description and material parameters. The templates are divided into two categories: sharp wrinkle formation state (bending angle is greater than 1.5 times of the critical angle, and curvature changes dramatically) and soft wrinkle potential state (bending angle is between 1.0-1.5 times of the critical angle, and curvature changes gently).The matching process first selects a subset of templates with similar materials based on quantitative microstructural data. The selection criteria include fabric type code, average yarn diameter (±0.02 mm), and warp and weft density (±30 yarns / 10 cm). The similarity between the local bending angle and the template is then calculated, using similarity metrics including the principal curvature ratio, principal direction angle, and curvature distribution entropy. A weighted Euclidean distance is used as the matching metric, with weights set to: principal curvature ratio 0.5, principal direction angle 0.3, and curvature distribution entropy 0.2. When the similarity of the best matching template exceeds 0.85, it is selected as the matching pleat element. When the similarity is between 0.7 and 0.85, a new template is synthesized through multi-template interpolation. When the similarity is less than 0.7, a parameterized pleat model is used for direct generation. The matching results include a geometric description of the pleat element, a displacement field, and texture deformation parameters. The matched pleat element data is used as superimposed displacement details and overlaid onto the corresponding curved region of the current mesh. The superposition process uses the spatial deformation field method to construct a mapping function from the matching wrinkle unit to the target mesh area. The mapping function uses radial basis function interpolation, and the basis function uses polyharmonic splines with an order of 3 to ensure C. 2 Continuity. First, control points are set on the boundary of the matching wrinkle unit. The number of control points is 32 and evenly distributed on the boundary. Then, corresponding control points are set on the boundary of the target grid area, and the corresponding relationship is determined by the boundary alignment algorithm. The internal deformation field is obtained by solving the Poisson equation and discretized using the finite element method. The grid accuracy is 1 mm. For the sharp wrinkle formation state, the displacement superposition intensity is 100%; for the soft wrinkle potential state, the displacement superposition intensity is the basic intensity (60%) multiplied by the activation probability value. The edge of the displacement superposition area uses a distance field for smooth transition. The transition zone width is 15 mm, and the transition function is a cubic spline function. For multiple adjacent wrinkle areas, the Brende function is used to achieve smooth blending, and the blending weight is inversely proportional to the distance. When the frequency of wrinkle geometric details exceeds the basic grid resolution, the normal vector texture mapping technology is used to enhance the visual effect. Finally, the local wrinkle displacement superposition data is generated, which contains the additional displacement vector and normal vector correction for each vertex to achieve accurate expression of the wrinkle details of the handicraft.

[0079] Preferably, in step S3, performing surface visual feature mapping on the handicraft deformation geometric grid data using the microstructure quantification data includes:

[0080] Surface normals and color textures are extracted based on the deformed geometric mesh data of the craft to obtain texture normal data of the craft;

[0081] Obtain test data on the lighting characteristics of handicrafts;

[0082] Based on the test data of the illumination characteristics of handicrafts, the light reflection and scattering characteristics are analyzed by the average pore size of the microscopic voids and the density of the warp and weft yarns in the quantitative data of the microscopic structure to generate the surface light scattering characteristics data;

[0083] Based on the average single yarn diameter, average twist, and average interlacing angle in the quantitative data of the microscopic structure, the surface light scattering characteristic data is used to calculate the tiny highlight lattice formed by the reflection of light on the yarn surface, and generate the yarn microscopic optical texture;

[0084] Perform normal perturbation processing on the handicraft texture normal data, and perform wrinkle texture normal superposition on the handicraft deformation geometric mesh data to obtain wrinkle detail normal data;

[0085] Surface visual feature mapping is performed based on the craft texture normal data, surface light scattering characteristic data, yarn microscopic optical texture and wrinkle detail normal data to generate multi-dimensional surface visual feature data; among which, the surface visual feature mapping includes the surface basic color texture, specular reflection-roughness, wrinkle detail normal and sub-surface scattering intensity.

[0086] In this embodiment of the present invention, the deformable geometric mesh data of the artifact is first constructed in tangent space, and the normal vector of each vertex is calculated using differential geometry methods. Normal vector calculation uses the area-weighted average method, summing the normal vectors of all triangles connected to the vertex according to the area weight and then normalizing them. Normal vector accuracy is maintained to four decimal places. Tangents and bitangents are fitted to a local parameterized coordinate system using the least-squares method, ensuring orthogonality error less than 0.001. For regions with mesh singularities, the tangent space is reconstructed using the polar parameterization method. Color and texture extraction utilizes artifact spectral image data and is mapped to the deformable geometric mesh through texture parameterization. The parameterization uses a minimum-warping isometric mapping algorithm, keeping mapping distortion within 5%. Texture resolution is set to 256×256 pixels per square centimeter, and color sampling is performed using bicubic interpolation. For overlapping areas, occlusion relationships are automatically detected to ensure texture penetration. Ultimately, artifact texture normal data is generated, containing vertex normal vectors, tangent vectors, texture coordinates, and color information. The illumination characteristics of handicrafts are tested using a multi-angle spectral reflectance measurement system. The measurement equipment includes a hemispherical multi-light source array and a high dynamic range imaging sensor. The light source array consists of 72 independently controlled white light-emitting diodes evenly distributed across the hemisphere with an angular resolution of 15 degrees. The spectral range of each light source is 380-780 nanometers, with a step size of 5 nanometers. During the measurement process, the handicraft sample is placed at the center of a rotating platform with a rotation accuracy of 0.1 degrees. The sample is scanned in all directions, and a set of high dynamic range images is acquired for each combination of incident and observation angles. The exposure time is set in 10 levels from 1 / 1000 second to 1 second. The captured images are photometrically calibrated using a diffuse white board with a reflectance of 99%. The measurement data includes the bidirectional reflectance distribution function (BRDF) with an angular resolution of 5 degrees in the incident and 5 degrees in the outgoing direction, and a wavelength resolution of 10 nanometers. Specifically, the fabric's reflectance characteristics at grazing angles (greater than 75 degrees) and its transmittance under backlight conditions are recorded. Data is stored as a four-dimensional tensor, with dimensions including incident angle, exit angle, azimuth, and wavelength. The illumination property test data of the handicrafts are combined with the quantitative data of the microstructure to analyze the light reflection and scattering characteristics. First, a micropore light transmission model is established, and the pore diameter parameters are extracted from the quantitative data of the microstructure. The average pore diameter of the micropores in silk fabrics is 0.015 mm, 0.025 mm for cotton and linen fabrics, and 0.035 mm for wool fabrics. The warp and weft yarn density parameters determine the periodicity of the surface microstructure. For a plain weave fabric with a warp density of 350 yarns / 10 cm and a weft density of 280 yarns / 10 cm, the surface microstructure period is 0.29 mm × 0.36 mm. The Monte Carlo ray tracing method is used to calculate light scattering. 10,000 light rays are emitted for each incident direction, and the scattering path of the light in the fabric structure is recorded. The interaction between light and fibers includes surface reflection, volume scattering, and multiple scattering between fibers.Scattering parameters were extracted from measured data using inverse rendering. The surface reflectance was set to 0.1-0.3, the volume scattering coefficient to 5-15 / mm, and the asymmetry factor to 0.7-0.9. Scattering properties were calculated for different wavelengths and synthesized into spectral response curves. Surface light scattering data were generated, including diffuse reflectance, specular reflectance, scattering depth, and scattering directional distribution function. The yarn microscopic optical texture was calculated based on the average individual yarn diameter, average twist, and average interlacing angle from the microstructural quantitative data, combined with the surface light scattering data. The yarn cross-section was modeled using a modified ellipse model, with the major-to-minor axis ratio determined by the yarn diameter and twist. For a yarn with a diameter of 0.15 mm and a twist of 500 twists / m, the major-to-minor axis ratio was set to 1.2. The yarn helical twist was modeled using a parameterized helical curve, with the helical period determined by the twist. For a yarn with a twist of 800 twists / m, the helical period was 1.25 mm. The interlacing angle determines the spatial configuration of the warp and weft interlacing points. For fabrics with an average interlacing angle of 45 degrees, the yarn bending angle at the interlacing points is set to 45 ± 5 degrees. Microscopic highlights are calculated using a physically based microsurface model, taking into account the anisotropic reflective properties of the yarn surface. The microsurface normal distribution function uses the GGX distribution, with a roughness parameter set to 0.1 along the yarn direction and 0.3 perpendicular to the yarn direction. The highlight point matrix is ​​discretely sampled within a 1 square millimeter area, with 100 × 100 sampling points. The reflection intensity at each sampling point is calculated at different viewing angles. The Fresnel effect is taken into account in the reflection calculation, with a refractive index of 1.5 for silk, 1.55 for cotton, and 1.6 for wool. The resulting yarn microscopic optical texture has a resolution of 2048 × 2048 pixels. Normal perturbation is used to enhance wrinkle detail in the artifact texture normal data. First, wrinkle regions were extracted from the artifact's deformed geometric mesh data. The extraction criteria were a principal curvature greater than 0.05 mm and a curvature change rate greater than 0.01 mm². A local tangent space coordinate system was constructed for each wrinkle region, with the origin at the wrinkle center, the Z axis along the mean normal vector, and the X axis along the wrinkle direction. Wrinkle texture normals were generated procedurally. The basic texture pattern consisted of high-frequency ripples parallel to the wrinkle direction, with a frequency proportional to the wrinkle curvature. For a wrinkle region with a curvature of 0.1 mm, the texture frequency was set to 5 waves / mm. The normal perturbation intensity was proportional to the wrinkle depth. For a wrinkle with a depth of 2 mm, the normal offset angle was set to ±15 degrees. The perturbation function consisted of a superposition of multiple noise functions with different frequencies, including a fundamental frequency and three harmonics, with an amplitude ratio of 1:0.5:0.25:0.125. The texture coordinates were stretched along the wrinkle direction, with the stretching factor being the ratio of the wrinkle region's length to its width, to ensure that the texture orientation aligned with the wrinkle. Normal superposition uses tangent space blending, and the blending weight is controlled by the wrinkle strength map, with the weight at the center of the wrinkle being 1.0 and smoothly transitioning to 0.0 at the edges. The final result is wrinkle detail normal data with a resolution of 4096×4096 pixels.Surface visual feature mapping is performed by integrating artifact texture normal data, surface light scattering data, yarn microscopic shadow texture, and wrinkle detail normal data. The surface base color texture is extracted from artifact spectral image data, with a resolution of 4096×4096 pixels, a linear RGB color space, and a color depth of 16 bits per channel. A specular-roughness map is generated from surface light scattering data. The reflectance intensity is dependent on the fabric material: for silk, the specular intensity is set to 0.3 and the roughness value to 0.4; for cotton and linen, the reflectance intensity is set to 0.15 and the roughness value to 0.7; and for wool, the reflectance intensity is set to 0.1 and the roughness value to 0.85. The wrinkle detail normal map uses a tangent space normal map format, with the RGB channels storing the offsets of the X, Y, and Z components of the normal, respectively. The resolution is 4096×4096 pixels. The subsurface scattering intensity is positively correlated with the average pore size of the microscopic voids. For an pore size of 0.015 mm, the scattering intensity is set to 0.4, with a scattering distance of 0.5 mm. For an pore size of 0.035 mm, the scattering intensity is set to 0.7, with a scattering distance of 1.2 mm. The yarn microscopic shadow texture is superimposed on the base color using a detail texture blending method with a multiply blending mode. All texture maps are seamlessly stitched together, with the stitching area width being 10% of the texture size. This ultimately generates multidimensional surface visual feature data containing multi-channel texture information for photorealistic rendering.

[0087] Preferably, the calculation of the tiny highlight lattice formed by light reflected on the yarn surface based on the average single yarn diameter, average twist and average interlacing angle in the quantitative data of the microstructure through surface light scattering characteristics data includes:

[0088] Based on the average single yarn diameter, average twist and average interlacing angle in the quantitative data of microstructure, the microscopic geometric fluctuations of the single yarn surface are extracted to obtain the single yarn perturbation geometric data;

[0089] The surface light scattering characteristic data is used to perturb the geometric data of the single yarn to simulate the emission angle when the yarn surface reflects, and the light reflection angle distribution data is generated;

[0090] Determine, based on the light reflection angle distribution data, a tiny area that satisfies a preset specular reflection condition and generates highlights, and obtain highlight area marking data;

[0091] The highlight intensity and range of the yarn interlacing area are calculated based on the highlight area marking data and the average interlacing angle to obtain the highlight intensity range data;

[0092] According to the highlight intensity range data and the highlight area marking data, a yarn microscopic optical texture is generated in which light is reflected on the surface of yarns with different twists to form a highlight dot matrix.

[0093] In an embodiment of the present invention, a precise geometric model of a single yarn is constructed based on the average single yarn diameter, average twist, and average interlacing angle from the quantitative microstructural data. For yarns with an average diameter of 0.15 mm, a twist of 800 twists / m, and an interlacing angle of 45 degrees, yarn surface topology data was acquired using a high-precision confocal laser scanning microscope. The scanning resolution was set to 0.5 microns, and the scanning range was 2 mm x 2 mm. Surface periodic features were extracted using Fourier analysis, with the dominant frequency corresponding to the twist period being 1.25 mm. Surface micro-undulations were modeled using a Gaussian process, with a correlation length set to 20 microns and a standard deviation set to 5 microns. The yarn surface was modeled using a cylindrical helix model, with the helix parameters determined by the twist, and each rotation of the helix corresponding to an axial displacement of 1.25 mm. The yarn is composed of multiple fiber bundles, each containing 15-20 individual fibers with a diameter of 10-15 microns. The surface perturbation geometry is represented using a displacement map with a resolution of 4096×4096 pixels, a height field accuracy of 16 bits, and a physical height range of ±10 microns. The generated single-filament perturbation geometry data includes the yarn centerline, cross-sectional shape parameters, and the surface micro-undulation height field. The surface light scattering characteristics data are combined with the single-filament perturbation geometry data to simulate yarn surface reflection. First, a differential reflection model is constructed, dividing the yarn surface into tiny bins of 5 μm×5 μm. The normal vector and microsurface roughness are extracted for each bin. The normal vector is calculated from the height field of the single-filament perturbation geometry data using the central difference method. The roughness is proportional to the local height variation rate with a coefficient of 0.5. An anisotropic microsurface BRDF is used for the optical model, with the yarn axial roughness α1 set to 0.1 and the circumferential roughness α2 set to 0.3. For silk, the refractive index is set to 1.5 + 0.01i; for cotton fabric, it is 1.55 + 0.03i; and for wool fabric, it is 1.6 + 0.05i. The incident light source uses a high-dynamic-range ambient light map with a resolution of 2048×1024 pixels and covering a 4π solid angle. Ray tracing uses importance sampling, emitting 1000 rays per facet and recording the exit direction and radiant brightness. The exit angles are expressed in spherical coordinates with a polar resolution of 1 degree and an azimuthal resolution of 2 degrees. Statistical clustering is performed on the exit data for all facets to generate light reflection angle distribution data. This data is structured as a three-dimensional tensor with the dimensions [polar angle, azimuthal angle, reflection intensity] and 32-bit floating-point precision. This light reflection angle distribution data is used to identify small areas that generate highlights. The preset specular reflection condition is defined as a highlight area when the exiting light intensity in a certain direction exceeds five times the diffuse reflection intensity, and the deviation from the ideal specular reflection direction in that direction is less than 15 degrees. The ideal specular reflection direction is calculated by reflecting the incident direction about the surface normal. The incident direction is set to the light source direction, which defaults to a normalized vector of (0.577, 0.577, 0.577). The highlight determination process first calculates the angle between each outgoing direction and the ideal specular reflection direction in the spherical coordinate system. This angle is calculated using the spherical distance formula.When the angle is less than 15 degrees and the reflection intensity exceeds a threshold, the corresponding bin is marked as a highlight region. This marking process is performed in a yarn parameterized coordinate system, defined as follows: the u direction is along the yarn axis, the v direction is along the yarn circumference, and the coordinate range is [0, 1] × [0, 1]. The marking results form a two-dimensional Boolean matrix of size 4096 × 4096, where a value of 1 indicates a highlight region, and a value of 0 indicates a non-highlight region. Morphological processing is performed on the marking results, including an opening operation with a radius of 3 pixels to eliminate isolated highlight points and a closing operation with a radius of 5 pixels to fill holes within the highlight region, generating continuous highlight region marking data. The highlight characteristics of light at the yarn intersection are inferred based on the highlight region marking data and the average crossing angle. The geometric model of the yarn intersection is obtained by three-dimensionally modeling the intersection of two yarns at a crossing angle. For a fabric with an average crossing angle of 45 degrees, a double yarn model with a 45-degree intersection is constructed. At the intersection point, the two yarns compress each other by 25% of the yarn diameter, with a radius of 1.5 times the yarn diameter. Ray tracing is used to simulate light propagation in the intersection area. 10,000 rays are emitted at each intersection point, with the incident direction sampled across a hemispherical surface with a resolution of 10 degrees. Light interactions with the yarn surface include direct reflection, single scattering, multiple scattering, and occlusion. Occlusion calculations utilize shadow ray detection. When a ray is blocked by another yarn, the reflected intensity is attenuated by 30%. Enhancement calculations take into account the geometric focusing effect of the intersection area. When the two yarn surface normals form a V-shaped valley, the reflected light is concentrated in a specific direction, with an enhancement factor of up to 2.0. Occlusion and enhancement factors are calculated based on the intersection angle and incident angle, forming a two-dimensional lookup table. The lookup table is applied to the highlight region marker data to modulate the highlight intensity, generating highlight intensity range data. The data format is a 32-bit floating-point image with a range of [0, 2]. The highlight intensity range data and highlight region marker data are used to generate the yarn microscopic optical texture. Texture generation employs a strategy that combines procedural methods with physical simulation. First, the basic fabric structure texture is constructed. The yarn arrangement is determined based on the warp and weft densities. For a fabric with a warp density of 350 yarns / 10 cm and a weft density of 280 yarns / 10 cm, a basic fabric structure with a period of 0.29 mm × 0.36 mm is constructed. Each yarn surface is subdivided into 50 × 200 tiny facets, evenly distributed along the circumferential and axial directions. For yarns with different twists, the twist effect is simulated by adjusting the surface normal distribution. Yarns with high twist (>1000 twists / m) have a high frequency of normal vector changes, while yarns with low twist (<500 twists / m) have a low frequency of normal vector changes. During highlight point generation, highlight intensity range data is applied to each facet to determine highlight brightness. The highlight shape is twist-dependent, with high-twist yarns producing small, discrete highlight points, while low-twist yarns produce more continuous highlight stripes. The texture synthesis resolution is 8192 × 8192 pixels, corresponding to a physical size of 5 cm × 5 cm.The final rendering uses a physically based lighting model, including Fresnel reflections, microsurface scattering, and intra-fiber scattering. The resulting yarn microscopic shadow texture includes three channels: diffuse, specular, and normal maps, accurately representing the microscopic highlights of the yarn surface.

[0094] Preferably, in step S3, performing visual feature rendering on the handicraft deformation geometric grid data based on the multi-dimensional surface visual feature data and performing a three-dimensional dynamic image display of the flexible handicraft includes:

[0095] The deformed geometric mesh data of the craft is used as a geometric entity in the three-dimensional scene, and the position and intensity data of the light source of the virtual scene are configured to obtain the rendering environment configuration data;

[0096] Load multi-dimensional surface visual feature data of geometric entities in the three-dimensional scene, and process surface visual detail attributes based on the rendering environment configuration data to generate surface visual detail data of handicrafts;

[0097] Calculate the reflection and scattering of light on the fabric surface based on the visual detail data of the craft surface, and project the self-shadow and mutual shadow effects to obtain the real-time light and shadow effect data of the craft;

[0098] The real-time light and shadow effect data of the craft is used to analyze the attenuation and color change of light after penetrating the fabric on the geometric entities in the 3D scene, and then the sub-surface scattering effect is presented to generate translucent scattering image data;

[0099] The semi-transparent scattering image data, the visual detail data of the handicraft surface and the rendering environment configuration data are used to display the three-dimensional dynamic image of the flexible handicraft and generate the three-dimensional dynamic image of the handicraft.

[0100] In an embodiment of the present invention, the deformed geometric mesh data of the craft is imported into the 3D rendering engine and set as the main geometric entity of the scene. The mesh accuracy is maintained at the original accuracy without simplification. The virtual scene light source configuration adopts a physical lighting model, including a main light source and three auxiliary light sources. The main light source is set as a surface light source with a size of 1 meter × 1 meter, a position at the scene coordinates (2, 2, 3), a light intensity of 5000 lumens, and a color temperature of 5600K. The three auxiliary light sources are set as point light sources, with positions of (-3, 1, 2), (0, -2, 1), and (2, -1, -2), respectively, with light intensities of 2000 lumens, 1500 lumens, and 1000 lumens, respectively, and color temperatures of 6500K, 4800K, and 3200K, respectively. The ambient lighting uses a high dynamic range environment map with a dynamic range of 0-10000 nits and a resolution of 4096×2048 pixels. The light source attenuation adopts an inverse square attenuation model, and the effective illumination range is set to 10 meters. The camera position is set to (0, -3, 1.5), with a field of view of 60 degrees, a near clipping plane of 0.1 meters, and a far clipping plane of 50 meters. The rendering environment configuration data contains complete light source parameters, material parameters, camera parameters, and ambient lighting information for subsequent real-time rendering calculations. Multidimensional surface visual feature data is loaded for geometric entities in the 3D scene, and surface visual details are processed using a physically based rendering pipeline. This multidimensional surface visual feature data includes four channels: surface base color texture, specular reflection-roughness, wrinkle detail normal, and sub-surface scattering intensity, each mapped to a corresponding attribute in the material system. The surface base color texture is applied to the diffuse channel using texture mapping, with texture filtering using anisotropic 16x sampling and 12 mipmap levels. Specular reflection-roughness data is applied to the specular channel, reflectivity is mapped to the R channel, and roughness is mapped to the G channel. Metalness is fixed at 0.1. Wrinkle detail normal data is applied to the normal channel using tangent space normal mapping, with normal intensity set to 1.0. Subsurface scattering intensity data is mapped to the translucency channel to control light penetration depth, with a maximum penetration depth of 5 mm. Texture mapping uses trilinear interpolation, and texture coordinate stitching error is kept within 1 pixel. Dynamic texture coordinate correction is used for dynamically deforming areas to ensure that textures are stretch-free during deformation. The surface visual processing algorithm is implemented in a GPU shader, consisting of 20 rendering passes with 16-bit floating-point precision per pass. This generates visual detail data for the artifact's surface. This data is used to calculate the interaction between light and the fabric surface. Reflection calculations use a physically based bidirectional reflectance distribution function (BRDF), using the GGX microsurface model to accurately simulate the effect of the fabric's surface microgeometry on light. Diffuse reflections are calculated using the Oren-Nayar model, accounting for retroreflective effects caused by surface micro-undulations. Self-shadowing is calculated using cascaded shadow mapping, with a shadow map resolution of 4096×4096 pixels, four cascade layers, and a coverage range of 0.5 to 10 meters.Soft shadow edges are achieved through Poisson disk sampling, with 32 sampling points and a softening radius of 0.01 times the distance from the object to the light source. Mutual shadow calculations take into account light occlusion between folds and utilize ambient occlusion technology, with a sampling radius of 5 cm and 64 sampling points. High-frequency shadow detail is enhanced through screen-space shading technology, with a resolution of half the full screen. Light bounce calculations utilize screen-space ray tracing, with 2 bounces, 128 steps, and an accuracy of 0.1 mm to achieve indirect lighting effects. All lighting and shadow calculations are updated in real time every frame, generating real-time lighting and shadow effect data for the craft, encompassing four main channels: direct lighting, indirect lighting, shadows, and reflections. This real-time lighting and shadow effect data for the craft is used to analyze the changing characteristics of light after penetrating the fabric. Subsurface scattering calculations utilize segregated subsurface scattering technology, which decomposes the scattering process into two steps: light penetration depth calculation and scattering diffusion calculation. Penetration depth is determined based on the average pore size of microscopic voids. The maximum penetration depth for silk fabrics (pore size 0.015 mm) is 2 mm; for cotton and linen fabrics (pore size 0.025 mm), it is 3.5 mm; and for wool fabrics (pore size 0.035 mm), it is 5 mm. The scattering radius is inversely proportional to the material density: 3 mm for silk, 2 mm for cotton and linen, and 1.5 mm for wool. As light propagates through the fabric, color attenuation is calculated using the Beer-Lambert law. The attenuation coefficients are provided by test data on the illumination characteristics of handicrafts. The attenuation coefficients for the red, green, and blue channels are 3 / mm, 4 / mm, and 5 / mm, respectively. Scattering intensity is calculated using a Gaussian approximation diffusion kernel with a kernel radius twice the scattering radius and 16 sampling points. Transmitted light calculations take into account the effect of backlight illumination on the fabric, achieving a transparent effect in backlighting. The resulting semi-transparent scattering image data has a resolution consistent with the screen resolution and contains three channels: scattered color, scattered intensity, and depth information. The final rendering output is generated by integrating translucent scattering image data, artifact surface visual detail data, and rendering environment configuration data. The rendering pipeline utilizes deferred rendering technology. The G-buffer contains five channels: position, normal, albedo, specular, and roughness, each with 32-bit floating-point precision. Lighting calculations are performed in screen space, supporting up to 128 dynamic light sources simultaneously. Post-processing includes four main steps: depth of field, motion blur, tone mapping, and color grading. The depth of field effect is calculated based on the distance from the artifact to the camera. The focal plane position is 3 meters, the aperture is f / 2.8, and the defocus disk radius is limited to 10 pixels. Motion blur is adaptively adjusted based on the artifact's movement speed, with a maximum blur angle of 30 degrees. Tone mapping uses the ACES scheme, compressing the dynamic range to the range of 0-1. Color grading is achieved using a 3D lookup table with a resolution of 32×32×32. The final composite utilizes temporal anti-aliasing, with a Halton sequence dithering mode and an accumulated frame count of 8.The rendering output resolution is 2560×1440 pixels, the color depth is 10 bits per channel, and the frame rate is stable at 60 frames per second, generating smooth and realistic three-dimensional dynamic images of handicrafts.

[0101] Preferably, the present invention further provides a three-dimensional dynamic display system for handicrafts, which executes the three-dimensional dynamic display method for handicrafts as described above, and the three-dimensional dynamic display system for handicrafts comprises:

[0102] The craft microscopic scanning module is used to perform multi-angle ultra-depth microscopic scanning on flexible craft samples to generate craft spectral image data; extract initial craft point cloud data based on the craft spectral image data, and quantify the microstructure of the flexible craft to generate microstructure quantitative data;

[0103] The flexible deformation simulation module is used to calculate the self-weight deformation gradient generated by stacked draping based on the quantitative data of the microstructure, analyze the macro deformation mode of the draping wrinkles, and generate macro flexible behavior characteristic data. The macro flexible behavior characteristic data is used to perform real-time deformation geometry grid updates on the initial handicraft point cloud data to generate handicraft deformation geometry grid data.

[0104] A three-dimensional visual rendering module is used to map the surface visual features of the handicraft's deformed geometric grid data using the microscopic structural quantitative data to generate multi-dimensional surface visual feature data; based on the multi-dimensional surface visual feature data, the module performs visual feature rendering on the handicraft's deformed geometric grid data and displays a three-dimensional dynamic image of the flexible handicraft to generate a three-dimensional dynamic image of the handicraft;

[0105] The user interaction display module is used to capture the user interaction information input by the user through the terminal device, and generate a three-dimensional interactive dynamic display screen for the three-dimensional dynamic screen of the handicraft, so as to continuously update the screen to form a continuous and smooth three-dimensional interactive dynamic display and obtain an interactive visual performance frame.

[0106] Preferably, the present invention further provides a computer-readable storage medium storing a computer program, wherein when the computer program is executed, the method for three-dimensional dynamic display of handicrafts as described above is implemented.

[0107] The present invention acquires spectral images through multi-angle ultra-depth-of-field microscopy scanning, constructs a point cloud, and extracts fabric microstructural features. It then analyzes the self-weight deformation and wrinkle formation mechanism based on microstructure quantitative data, achieving dynamic updates of physically driven geometric grids. It then performs multi-dimensional visual feature mapping including light scattering, yarn highlights, and wrinkle details. Finally, it achieves real-time dynamic image generation and display through user interaction. Therefore, the present invention's three-dimensional dynamic display method for handicrafts achieves accurate quantification of fabric micropores, yarn diameter, twist, and interlacing angles through physical deformation modeling driven by microstructure quantification, and establishes a mapping relationship between microstructure and macroscopic flexible behavior. It achieves accurate simulation of local wrinkle formation through calculation of stacked draping self-weight deformation gradients and determination of wrinkle activation conditions. It generates yarn microscopic optical textures by combining single-filament perturbation geometry with light reflection angle distribution data, achieving visual detail reproduction at the yarn level. It combines multi-dimensional surface visual feature mapping with dynamic geometric grid updates to achieve synchronous updates of physical real deformation and visual effects of flexible handicrafts under interactive operations.

[0108] The present invention is therefore intended to be illustrative and non-restrictive in all respects, with the scope of the invention being defined by the appended claims rather than the foregoing description, and all changes that come within the meaning and range of equivalents of the application documents are intended to be embraced therein.

[0109] The foregoing description is intended only to provide specific embodiments of the present invention, which will enable those skilled in the art to understand and implement the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not intended to be limited to the embodiments shown herein, but is to be construed in the widest possible manner consistent with the principles and novel features disclosed herein.

Claims

1. A three-dimensional dynamic display method for handicrafts, characterized in that: The following steps are involved: Step S1: Perform multi-angle ultra-depth microscopic scanning on the flexible handicraft sample to generate handicraft spectral image data; Extracting initial handicraft point cloud data based on handicraft spectral image data, and quantifying the microstructure of flexible handicrafts to generate microstructure quantitative data; Step S2: Calculate the self-weight deformation gradient generated by the stacked drape based on the quantitative data of the microstructure, analyze the macroscopic deformation mode of the drape wrinkles, and generate macroscopic flexibility behavior characteristic data; The initial handicraft point cloud data is updated with real-time deformation geometry grid through macro-flexible behavior feature data to generate handicraft deformation geometry grid data; Step S3: performing surface visual feature mapping on the handicraft deformation geometric grid data through the microstructure quantitative data to generate multi-dimensional surface visual feature data; Based on the multi-dimensional surface visual feature data, the deformed geometric grid data of the handicraft is rendered with visual features, and a three-dimensional dynamic picture of the flexible handicraft is displayed to generate a three-dimensional dynamic picture of the handicraft; Step S4: capturing user interaction information input by the user through the terminal device, and generating a three-dimensional interactive dynamic display picture for the three-dimensional dynamic picture of the handicraft to obtain an interactive visual performance frame.

2. The three-dimensional dynamic display method of handicrafts according to claim 1, characterized in that: Step S1 includes the following steps: Constructing an initial three-dimensional geometric point cloud of the flexible craft surface according to the craft spectral image data to obtain initial craft point cloud data; Performing fabric surface texture matching based on the initial handicraft point cloud data and the handicraft spectral image data to identify the fabric structure type and obtain fabric structure type identification data; The initial handicraft point cloud data and fabric structure type identification data are used to analyze the diameter, initial twist, thread density, and microscopic void size and distribution of individual silk threads, generating a microscopic void density map. The microstructure of flexible handicrafts is quantified based on fabric type identification data and micropore density maps to generate microstructure quantitative data. The microstructure quantitative data includes fabric type code, warp and weft yarn density, average single silk thread diameter, average twist, average interlacing angle and average pore size of micro voids.

3. The three-dimensional dynamic display method of handicrafts according to claim 2, characterized in that: Fabric surface texture matching is performed based on the initial handicraft point cloud data and handicraft spectral image data to identify the fabric structure type, including: Gray-level co-occurrence matrix processing is performed on the craft spectral image data. When the contrast value in the gray-level co-occurrence matrix is ​​greater than 0.3 and the entropy value is greater than 4.5, local texture feature data is extracted. Perform pattern matching on local texture feature data using a preset fabric weave pattern library to identify the fabric weave type and obtain preliminary weave type identification data; When the preliminary weave type identification data is plain, twill, or satin regular fabric type, the spacing between adjacent silk thread interlacing points and the silk thread floating and sinking period in the initial handicraft point cloud data are quantified to obtain the regular fabric quantitative parameters; When the preliminary weave type identification data is a jacquard or embroidered complex fabric type, the thread arrangement variation pattern and color interweaving area boundary within each repeating unit in the initial handicraft point cloud data are marked to construct a complex fabric unit code; Based on regular fabric quantization parameters and complex fabric unit coding, fabric structure type identification data including fabric structure type coding, warp and weft yarn density, and average yarn interlacing angle are generated.

4. The three-dimensional dynamic display method of handicrafts according to claim 1, characterized in that: In step S2, the self-weight deformation gradient generated by the stacked drape is calculated based on the quantitative data of the microstructure, and the macroscopic deformation mode of the drape wrinkle is analyzed, including: The initial Young's modulus, Poisson's ratio and shear modulus of the flexible craft are extracted from the microstructure quantitative data through the preset material property association table, and the initial craft point cloud data is parameter mapped to obtain the basic mechanical parameter mapping data; Based on the fabric type code in the microstructure quantitative data and the Young's modulus in the basic mechanical parameter mapping data, dynamic drape characteristics analysis is performed, and the corresponding local area impact factor evaluation of the flexible craft is performed to obtain dynamic drape factor data; Based on the dynamic drape factor data, the initial craft point cloud data is used to calculate the self-weight deformation gradient caused by the stacking drape in the local area to obtain the drape gradient quantitative data; Based on the Young's modulus and Poisson's ratio in the basic mechanical parameter mapping data, the minimum bending angle value for the flexible craft to produce stable dynamic wrinkles when bent is set to obtain the wrinkle critical angle value; wherein, when the Young's modulus is less than 2.0 GPa and the Poisson's ratio is greater than 0.28, the wrinkle critical angle is set to 15°-25°, when the Young's modulus is in the range of 2.0-3.5 GPa and the Poisson's ratio is in the range of 0.20-0.28, the wrinkle critical angle is set to 25°-40°, and when the Young's modulus is greater than 3.5 GPa and the Poisson's ratio is less than 0.20, the wrinkle critical angle is set to 40°-65°, and the wrinkle critical angle value is obtained; The macro deformation pattern of the drape fold is analyzed based on the quantitative data of the drape gradient and the critical angle of the fold to generate the macro deformation pattern data; The macro deformation mode data, basic mechanical parameter mapping data, overhang gradient quantification data and wrinkle critical angle values ​​are associated with the macro deformation properties to generate macro flexible behavior characteristic data.

5. The three-dimensional dynamic display method of handicrafts according to claim 1, characterized in that: In step S2, the real-time deformation geometric mesh update of the initial craft point cloud data using the macro-flexible behavior feature data includes: Constructing a three-dimensional geometric grid structure of the flexible craft based on the initial craft point cloud data; Based on the three-dimensional geometric grid structure of the flexible craft, a virtual gravity field is loaded to obtain the initial deformation grid; Based on the macroscopic flexible behavior characteristic data, the initial stress state of the initial deformation grid is analyzed to obtain the initial stress and strain tensor of the unit; Calculate the initial displacement of each node in the initial deformation grid according to the initial stress and strain tensor of the unit, and generate a set of initial node displacement vectors; Perform iterative solution processing based on the preliminary node displacement vector set until the preset convergence judgment conditions are met to generate static deformation morphological data; The static deformation morphology data is corrected for the vertical displacement of the grid vertices in the stacking area by using the macroscopic flexible behavior characteristic data to obtain the stacking corrected displacement data; Performing local wrinkle displacement superposition processing on static deformation morphology data by stacking corrected displacement data to generate local wrinkle displacement superposition data; Based on the stacked corrected displacement data, the static deformation morphology data is dynamically superimposed on the local wrinkle displacement superposition data, and the real-time deformation geometric grid is updated to generate the handicraft deformation geometric grid data.

6. The three-dimensional dynamic display method of handicrafts according to claim 1, characterized in that: In step S3, surface visual feature mapping of the handicraft deformation geometric grid data using the microstructure quantitative data includes: Surface normals and color textures are extracted based on the deformed geometric mesh data of the craft to obtain texture normal data of the craft; Obtain test data on the lighting characteristics of handicrafts; Based on the test data of the illumination characteristics of handicrafts, the light reflection and scattering characteristics are analyzed by the average pore size of the microscopic voids and the density of the warp and weft yarns in the quantitative data of the microscopic structure to generate the surface light scattering characteristics data; Based on the average single yarn diameter, average twist, and average interlacing angle in the quantitative data of the microscopic structure, the surface light scattering characteristic data is used to calculate the tiny highlight lattice formed by the reflection of light on the yarn surface, and generate the yarn microscopic optical texture; Perform normal perturbation processing on the handicraft texture normal data, and perform wrinkle texture normal superposition on the handicraft deformation geometric mesh data to obtain wrinkle detail normal data; Surface visual feature mapping is performed based on the craft texture normal data, surface light scattering characteristic data, yarn microscopic optical texture and wrinkle detail normal data to generate multi-dimensional surface visual feature data; among which, the surface visual feature mapping includes the surface basic color texture, specular reflection-roughness, wrinkle detail normal and sub-surface scattering intensity.

7. The three-dimensional dynamic display method of handicrafts according to claim 6, characterized in that: Based on the average single yarn diameter, average twist and average interlacing angle in the quantitative data of the microstructure, the surface light scattering characteristics data is used to calculate the tiny highlight points formed by the reflection of light on the yarn surface, including: Based on the average single yarn diameter, average twist and average interlacing angle in the quantitative data of microstructure, the microscopic geometric fluctuations of the single yarn surface are extracted to obtain the single yarn perturbation geometric data; The surface light scattering characteristic data is used to perturb the geometric data of the single yarn to simulate the emission angle when the yarn surface reflects, and the light reflection angle distribution data is generated; Determine, based on the light reflection angle distribution data, a tiny area that satisfies a preset specular reflection condition and generates highlights, and obtain highlight area marking data; The highlight intensity and range of the yarn interlacing area are calculated based on the highlight area marking data and the average interlacing angle to obtain the highlight intensity range data; According to the highlight intensity range data and the highlight area marking data, a yarn microscopic optical texture is generated in which light is reflected on the surface of yarns with different twists to form a highlight dot matrix.

8. The three-dimensional dynamic display method of handicrafts according to claim 1, characterized in that: In step S3, visual feature rendering is performed on the deformed geometric grid data of the handicraft based on the multi-dimensional surface visual feature data, and a three-dimensional dynamic image display of the flexible handicraft is performed, including: The deformed geometric mesh data of the craft is used as a geometric entity in the three-dimensional scene, and the position and intensity data of the light source of the virtual scene are configured to obtain the rendering environment configuration data; Load multi-dimensional surface visual feature data of geometric entities in the three-dimensional scene, and process surface visual detail attributes based on the rendering environment configuration data to generate surface visual detail data of handicrafts; Calculate the reflection and scattering of light on the fabric surface based on the visual detail data of the craft surface, and project self-shadow and mutual shadow effects to obtain real-time light and shadow effect data of the craft; The real-time light and shadow effect data of the craft is used to analyze the attenuation and color change of light after penetrating the fabric on the geometric entities in the 3D scene, and then the sub-surface scattering effect is presented to generate translucent scattering image data; The semi-transparent scattering image data, the visual detail data of the handicraft surface and the rendering environment configuration data are used to display the three-dimensional dynamic image of the flexible handicraft and generate the three-dimensional dynamic image of the handicraft.

9. A three-dimensional dynamic display system for handicrafts, characterized in that: For executing the three-dimensional dynamic display method of handicrafts according to claim 1, the three-dimensional dynamic display system of handicrafts comprises: The craft microscopic scanning module is used to perform multi-angle ultra-depth microscopic scanning on flexible craft samples to generate craft spectral image data; extract initial craft point cloud data based on the craft spectral image data, and quantify the microstructure of the flexible craft to generate microstructure quantitative data; The flexible deformation simulation module is used to calculate the self-weight deformation gradient generated by stacked draping based on the quantitative data of the microstructure, analyze the macro deformation mode of the draping wrinkles, and generate macro flexible behavior characteristic data. The macro flexible behavior characteristic data is used to perform real-time deformation geometry grid updates on the initial handicraft point cloud data to generate handicraft deformation geometry grid data. A three-dimensional visual rendering module is used to map the surface visual features of the handicraft's deformed geometric grid data using the microscopic structural quantitative data to generate multi-dimensional surface visual feature data; based on the multi-dimensional surface visual feature data, the module performs visual feature rendering on the handicraft's deformed geometric grid data and displays a three-dimensional dynamic image of the flexible handicraft to generate a three-dimensional dynamic image of the handicraft; The user interaction display module is used to capture the user interaction information input by the user through the terminal device, and generate a three-dimensional interactive dynamic display screen for the three-dimensional dynamic screen of the handicraft, so as to continuously update the screen to form a continuous and smooth three-dimensional interactive dynamic display and obtain an interactive visual performance frame.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed, the three-dimensional dynamic display method for handicrafts according to any one of claims 1 to 8 is implemented.

Images