National culture digital asset library and art teaching resource generation system

Abstract

The invention relates to the field of cultural heritage digitization, in particular to a national culture digital asset library and art teaching resource generation system. Comprising a curvature-perceived adaptive scanning module, a differential geometry-driven model reconstruction module, a geometric feature-driven semantic mapping module, a teaching resource generation module, an immersive experience module and a quality evaluation module, and the adaptive scanning module dynamically adjusts scanning parameters according to the surface curvature and material of an object to realize high-precision acquisition; the model reconstruction module utilizes a curvature flow theory to repair defects and optimize a three-dimensional model; the semantic mapping module is used for extracting differential invariant features and establishing automatic association between geometric morphology and cultural semantics; the teaching resource generation module generates interactive teaching content and a resource package based on the semantic tag and the three-dimensional model; the immersive experience module constructs an interactive culture learning environment through a VR / AR technology; according to the invention, high-precision digitalization, automatic semantic association and intelligent teaching resource generation of national artware are realized.

Description

Technical Field

[0001] This invention relates to the field of cultural heritage digitization, specifically to a digital asset database of ethnic culture and a system for generating art teaching resources, which is applied to the digital protection of intangible cultural heritage and art education. Background Technology

[0002] With the rapid development of digital technology, the digital preservation of cultural heritage has become an important means of cultural transmission. Traditional ethnic handicrafts, as an important part of intangible cultural heritage, possess rich artistic value and educational significance. However, the digitization of ethnic handicrafts currently faces many technical challenges:

[0003] On the one hand, traditional 3D scanning technology has limited ability to capture the fine structure of complex handicrafts. Ethnic handicrafts often have complex geometric shapes and rich surface details. Conventional uniformly encoded structured light scanning is difficult to accurately capture both the overall shape and local details, resulting in digital models that lack artistic details or contain a lot of noise.

[0004] On the other hand, there is a lack of effective semantic connections between digital assets and educational applications. Existing digital models are mostly stored in the form of geometric data, lacking cultural semantic annotation, making it difficult to support content-based intelligent retrieval and educational applications, thus limiting the value of digital resources in art education.

[0005] In existing technologies, 3D scanning of complex surfaces typically employs methods to increase overall resolution. This not only significantly increases scanning time and data volume but also fails to address scanning challenges posed by material properties. Furthermore, model repair and optimization often utilize globally consistent processing strategies, making it difficult to balance the conflict between preserving artistic details and optimizing data. Moreover, the integration of digital models with educational applications relies heavily on manual annotation and processing, which is inefficient and lacks systematic methodological support. Summary of the Invention

[0006] The purpose of this invention is to solve the above problems and provide a digital asset library of ethnic culture and a system for generating art teaching resources based on differential geometry theory. This system can accurately capture the complex geometric features of ethnic handicrafts, preserve artistic details, realize the automatic association between geometric features and cultural semantics, and support the generation of resources for teaching applications.

[0007] This invention proposes a digital asset database of ethnic culture and a system for generating art teaching resources, including:

[0008] The curvature-aware adaptive scanning module is used for:

[0009] Collect surface curvature distribution data of the target object;

[0010] The structured light projection parameters are dynamically adjusted based on the surface curvature distribution data.

[0011] Based on the adjusted structured light projection parameters, the target object is scanned in a detailed regional manner to generate point cloud data.

[0012] A differential geometry-driven model reconstruction module, connected to the curvature-aware adaptive scanning module, is used for:

[0013] Receive the point cloud data;

[0014] Based on the point cloud data, an initial triangular mesh model is constructed;

[0015] The defective regions in the initial triangular mesh model are adaptively repaired using the curvature flow evolution equation;

[0016] Perform multi-scale geometric processing to generate an optimized 3D model;

[0017] A geometric feature-driven semantic mapping module, connected to the differential geometry-driven model reconstruction module, is used for:

[0018] Extract the differential invariant features of the optimized 3D model;

[0019] Based on the differential invariant features, the ethnic cultural style features of the optimized 3D model are identified;

[0020] Establish the mapping relationship between the geometric features of the optimized 3D model and cultural semantic tags;

[0021] The teaching resource generation module, connected to the geometric feature-driven semantic mapping module, is used for:

[0022] Based on the cultural semantic tags and the optimized 3D model, art teaching resources are generated.

[0023] The art teaching resources are categorized and stored in a resource library;

[0024] It provides a multi-dimensional search interface, supporting resource retrieval based on geometric features and semantic tags.

[0025] Preferably, the curvature-sensing adaptive scanning module includes:

[0026] The curvature pre-analysis unit is used to perform a rapid pre-scan of the target object, calculate the surface curvature distribution map, and divide the surface of the target object into high, medium, and low curvature regions.

[0027] The material property analysis unit is used to identify the surface material type of a target object through multispectral reflectance analysis and to construct a material-reflectance characteristic parameter table.

[0028] The microprism array control unit is used to dynamically adjust the density and arrangement pattern of the microprism array according to the curvature distribution diagram and the material-reflection characteristic parameter table.

[0029] The composite light source control unit is used to adjust the intensity, wavelength, and projection angle of the projected light source to adapt to different curvature areas and material properties.

[0030] Preferably, the differential geometry-driven model reconstruction module includes:

[0031] A geometric defect identification unit is used to analyze the initial triangular mesh model, identify holes, non-manifold structures and singularities, and generate a defect classification map;

[0032] The curvature flow computation unit is used to construct the logical framework of differential equations describing the evolution of the surface and to control the geometric changes of the defect region.

[0033] The feature preservation unit is used to extract artistic feature lines and regions of the target object, generate a feature preservation mapping map, and preserve the artistic feature lines and regions during the restoration process;

[0034] A multi-scale processing unit is used to decompose the initial triangular mesh model into different frequency components, optimize each frequency component separately, and re-synthesize to generate the optimized three-dimensional model.

[0035] Preferably, the multi-scale processing unit includes:

[0036] Hierarchical geometric decomposition sub-units are used to construct a multi-resolution representation of the initial triangular mesh model, decomposing geometric information into low-frequency, mid-frequency, and high-frequency components;

[0037] The frequency-selective processing subunit is used to apply a global smoothing strategy to low-frequency geometric components, structural optimization processing to mid-frequency components, and detail-preserving algorithms to high-frequency components.

[0038] The multi-scale reconstruction sub-unit is used to recombine the processed frequency components and ensure a smooth transition between different frequency components.

[0039] Preferably, the geometric feature-driven semantic mapping module includes:

[0040] The differential invariant calculation unit is used to calculate the differential invariants such as the principal curvature, Gaussian curvature, and shape index of the optimized three-dimensional model surface;

[0041] A cultural feature recognition unit is used to identify typical geometric features and artistic style features of ethnic handicrafts based on the differential invariants.

[0042] The semantic ontology mapping unit is used to establish association rules between geometric features and cultural semantic concepts, and to calculate the association strength.

[0043] The knowledge base management unit is used to maintain the model library and the association rule library, and to optimize mapping rules based on newly added models.

[0044] Preferably, the teaching resource generation module includes:

[0045] The resource template library is used to store resource templates for different teaching objectives and learning stages;

[0046] The content customization unit is used to select an appropriate resource template based on the cultural semantic tags and teaching objectives;

[0047] A three-dimensional interactive processing unit is used to generate interactive three-dimensional teaching content based on the optimized three-dimensional model.

[0048] The teaching scenario construction unit is used to integrate the interactive 3D teaching content with the teaching script to form a complete teaching resource.

[0049] As a preferred option, it also includes:

[0050] The immersive experience module, connected to the teaching resource generation module, is used for:

[0051] Receive the optimized 3D model and the art teaching resources;

[0052] Construct interactive cultural experience environments based on virtual reality or augmented reality technologies;

[0053] It supports users to engage in multi-dimensional interaction and learning within the interactive cultural experience environment.

[0054] Preferably, the immersive experience module includes:

[0055] A virtual scene construction unit is used to create a cultural background scene based on the optimized 3D model.

[0056] An interactive control unit is used to recognize user action commands and convert them into interactive operations in the virtual environment;

[0057] The cultural explanation unit is used to provide explanations of cultural knowledge related to the 3D model based on the cultural semantic tags.

[0058] The learning feedback unit is used to record user learning behavior data and generate learning effectiveness evaluation reports.

[0059] Preferably, the structured light projection parameters in the curvature-aware adaptive scanning module include:

[0060] The density of the microprism array ranges from 400 to 900 microprisms per square centimeter in the high curvature region and from 50 to 200 microprisms per square centimeter in the low curvature region.

[0061] The intensity of the projected light is adjusted within the range of 20 to 200 lumens, depending on the reflectivity of the target object's surface.

[0062] The scanning resolution is 0.01–0.05 mm in the high curvature region and 0.1–0.5 mm in the low curvature region.

[0063] As a preferred option, it also includes:

[0064] The quality assessment module, connected to the differential geometry-driven model reconstruction module, is used for:

[0065] Evaluate the geometric integrity and artistic feature fidelity of the optimized 3D model;

[0066] Calculate the morphological deviation index between the optimized 3D model and the original object;

[0067] A quality assessment report is generated, and a rescan instruction is sent to the curvature-aware adaptive scanning module when the deviation index exceeds a preset threshold.

[0068] This invention introduces differential geometry theory to guide the 3D scanning and reconstruction process, achieving the following beneficial effects:

[0069] 1. Improved the digitization accuracy of complex ethnic handicrafts. The adaptive scanning strategy based on curvature analysis can dynamically adjust scanning parameters for different curvature regions and material characteristics, significantly improving the ability to capture artistic details while reducing scanning time and data redundancy.

[0070] 2. Enhanced the intelligence level of model repair and optimization. The adaptive repair mechanism based on curvature flow theory can intelligently adjust the processing strategy according to geometric characteristics, effectively filling in data gaps and removing noise while preserving artistic features, thus improving the quality and usability of digital models.

[0071] 3. A bridge has been established between geometric features and cultural semantics. Through differential invariant feature extraction and mapping mechanisms, the automatic association between geometric forms and cultural semantics has been achieved, supporting content-based intelligent retrieval and teaching applications, and enhancing the educational value of digital resources.

[0072] 4. A complete technology chain has been formed, from the digitization of cultural relics to teaching applications. The system integrates a full-process solution from 3D scanning, model reconstruction, semantic mapping to the generation of teaching resources, supporting diverse teaching scenarios and application needs. Attached Figure Description

[0073] Figure 1 This is a system architecture diagram according to an embodiment of the present invention;

[0074] Figure 2 This is a schematic diagram of the curvature-sensing adaptive scanning module in an embodiment of the present invention;

[0075] Figure 3 This is a schematic diagram of the structure of the differential geometry-driven model reconstruction module in an embodiment of the present invention;

[0076] Figure 4 This is a schematic diagram of the structure of the geometric feature-driven semantic mapping module in an embodiment of the present invention;

[0077] Figure 5 This is a schematic diagram of the teaching resource generation module in an embodiment of the present invention;

[0078] Figure 6 This is a schematic diagram of the immersive experience module in an embodiment of the present invention;

[0079] Figure 7 This is a system workflow diagram according to an embodiment of the present invention. Detailed Implementation

[0080] Please refer to the attached document. Figure 1-7 The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. Those skilled in the art should understand that these embodiments are for illustrative purposes only and are not intended to limit the scope of the invention.

[0081] like Figure 1 As shown, the ethnic cultural digital asset database and art teaching resource generation system provided by the present invention includes a curvature-aware adaptive scanning module 1, a differential geometry-driven model reconstruction module 2, a geometric feature-driven semantic mapping module 3, a teaching resource generation module 4, an immersive experience module 5, and a quality assessment module 6.

[0082] like Figure 2 As shown, the curvature-aware adaptive scanning module 1 is used to collect surface curvature distribution data of the target object, dynamically adjust the structured light projection parameters according to the surface curvature distribution data, and perform fine regional scanning of the target object based on the adjusted structured light projection parameters to generate point cloud data.

[0083] Preferably, the curvature-sensing adaptive scanning module 1 includes a curvature pre-analysis unit 11, a material property analysis unit 12, a microprism array control unit 13, and a composite light source control unit 14.

[0084] The curvature pre-analysis unit 11 performs a rapid pre-scan of the target object, calculates the surface curvature distribution map, and divides the target object surface into high, medium, and low curvature regions. Specifically, the curvature pre-analysis unit 11 first performs a rapid scan of the target object using a low-resolution scanning mode (approximately 0.5 mm resolution) to obtain the overall shape; then, based on the obtained initial point, it calculates the surface normal vector and estimates the surface curvature using the rate of change of adjacent normal vectors. The curvature calculation can be performed as follows:

[0085] For each point in the point cloud Select the set of points in its neighborhood. Calculate the normal vector of this point. Then, the principal curvature of the point is estimated by analyzing the changes in the normal vector within the neighborhood. and

[0086] ,

[0087] in: For the current point in the point cloud, For point The neighborhood set of points is typically selected from points whose Euclidean distance is less than a preset threshold (e.g., 2 mm). For point The unit normal vector at that location, For neighborhood points The unit normal vector, The weighting coefficient is usually set to 1. ,in The smoothing parameter is typically 1 / 3 of the neighborhood radius; eigenvalues ​​represent the eigenvalues ​​of the computed matrix. and These are the maximum and minimum principal curvatures, respectively, indicated by superscript. This indicates the matrix transpose.

[0088] Based on the calculated principal curvature values, the curvature pre-analysis unit 11 divides the surface of the target object into regions with different curvatures. In one embodiment of the present invention, the following classification criteria are used:

[0089] High curvature region: The maximum absolute value of the principal curvature is greater than 0.5 mm⁻¹;

[0090] Medium curvature region: The maximum absolute value of the principal curvature is between 0.1 and 0.5 mm⁻¹;

[0091] Low curvature region: The maximum absolute value of the principal curvature is less than 0.1 mm⁻¹.

[0092] The material characteristic analysis unit 12 identifies the surface material type of the target object through multispectral reflectance analysis and constructs a material-reflectance characteristic parameter table. Specifically, the material characteristic analysis unit 12 first illuminates the target object with light sources of different wavelengths (including visible light and near-infrared) under normal light mode and collects reflectance spectrum data; then, it matches the collected reflectance spectrum with a preset material characteristic database to identify the surface material type of the target object; finally, based on the identification results, it extracts the corresponding reflectance characteristic parameters from the preset parameter library. In an embodiment of the present invention, typical examples of ethnic handicraft materials and their reflectance characteristic parameters are as follows:

[0093] Ceramic: Diffuse reflectance coefficient 0.7-0.9, specular reflectance coefficient 0.1-0.3, optimal luminous intensity range 80-120 lumens;

[0094] Wood: Diffuse reflectance coefficient 0.5-0.8, specular reflectance coefficient 0.05-0.2, optimal light intensity range 60-100 lumens;

[0095] Metals: Diffuse reflectance 0.1-0.4, specular reflectance 0.6-0.9, optimal luminous intensity range 40-80 lumens;

[0096] Fabric: Diffuse reflectance coefficient 0.8-0.95, specular reflectance coefficient 0.05-0.15, optimal light intensity range 100-150 lumens.

[0097] The microprism array control unit 13 dynamically adjusts the density and arrangement of the microprism array based on the curvature distribution map and the material-reflection characteristic parameter table. Specifically, the microprism array control unit 13 receives the curvature distribution map output by the curvature pre-analysis unit 11 and the material-reflection characteristic parameter table output by the material characteristic analysis unit 12, generates microprism array configuration instructions based on this information, and then controls the programmable microprism array to perform corresponding adjustments. In an embodiment of the present invention, the correspondence between the density of the microprism array and the curvature region is as follows:

[0098] High curvature region: 600-900 microprisms per square centimeter (corresponding to a resolution of approximately 0.01-0.03 mm);

[0099] Medium curvature region: 200-400 microprisms per square centimeter (corresponding to a resolution of approximately 0.05-0.1 mm);

[0100] Low curvature region: 50-150 microprisms per square centimeter (corresponding to a resolution of approximately 0.1-0.5 mm).

[0101] The composite light source control unit 14 adjusts the intensity, wavelength, and projection angle of the projected light source to adapt to different curvature regions and material properties. Specifically, the composite light source control unit 14 selects the optimal light source parameters according to a material-reflection characteristic parameter table; simultaneously, it adjusts the projection angle according to a curvature distribution diagram to ensure that the light can effectively illuminate high curvature regions and concave regions. In embodiments of the present invention, the composite light source control unit 14 supports the following parameter adjustment ranges:

[0102] Luminous intensity: 20-200 lumens, automatically adjusted according to the reflectivity of the material;

[0103] Wavelength: Blue light (450-495nm), green light (495-570nm), or red light (620-750nm) can be selected, and the optimal wavelength can be chosen according to the material characteristics;

[0104] Projection angle: ±30° horizontally, ±20° vertically, automatically adjusted according to curvature distribution.

[0105] Based on the collaborative work of the aforementioned units, the curvature-sensing adaptive scanning module 1 can automatically optimize the scanning strategy for the complex shapes and diverse materials of ethnic handicrafts, thereby improving scanning efficiency and accuracy. For example, for wooden handicrafts with finely carved textures, the system will use a high-density microprism array and moderate light intensity in the carved areas, while using a low-density array and higher light intensity in smooth areas, thus improving scanning efficiency while ensuring detail capture.

[0106] like Figure 3 As shown, the differential geometry-driven model reconstruction module 2 is used to receive point cloud data, construct an initial triangular mesh model based on the point cloud data, adaptively repair the defective areas in the initial triangular mesh model using the curvature flow evolution equation, perform multi-scale geometric processing, and generate an optimized three-dimensional model.

[0107] Preferably, the differential geometry-driven model reconstruction module 2 includes a geometric defect identification unit 21, a curvature flow calculation unit 22, a feature preservation unit 23, and a multi-scale processing unit 24.

[0108] The geometric defect identification unit 21 analyzes the initial triangular mesh model, identifies holes, non-manifold structures, and singularities, and generates a defect classification map. Specifically, the geometric defect identification unit 21 first checks the topological structure of the mesh, including edge connectivity and vertex adjacency; then it identifies various defects in the mesh, such as holes (closed regions enclosed by boundary edges), non-manifold structures (edges connecting more than two faces), and singularities (abnormal vertex link relationships); finally, it maps the identification results onto the model to generate a defect classification map. In embodiments of the present invention, the defect identification adopts the following criteria:

[0109] A hole: There are one or more closed boundary loops, and there are no mesh surfaces within the boundary loops.

[0110] Non-manifold edge: An edge shared by more than two triangles.

[0111] Non-manifold vertices: Adjacent faces of a vertex cannot form a single connected region.

[0112] Singularity: A vertex whose connectivity (the number of edges connected to that vertex) is abnormal, typically much higher or much lower than the average.

[0113] The curvature flow calculation unit 22 constructs a logical framework of differential equations describing surface evolution, controlling the geometric changes in the defect region. Curvature flow is an important tool in differential geometry for describing surface evolution, and this invention innovatively applies it to 3D model repair. Specifically, the curvature flow calculation unit 22 selects an appropriate curvature flow model according to the defect type; then it constructs the corresponding differential equations to define the rules of surface evolution; finally, it obtains the repair result of the defect region through numerical solution.

[0114] In an embodiment of the present invention, the Mean Curvature Flow model is used for hole repair, and its evolution equation is as follows:

[0115] ,

[0116] in: A three-dimensional position vector of a point on the surface , This is an evolution time parameter, and its value range is typically [range missing]. ,in This represents the total evolution time, dynamically adjusted based on the defect size, with a typical value of 0.1-1.0 seconds. The mean curvature, i.e. , This is the unit normal vector at that point, pointing outwards from the surface.

[0117] For handling noise and small defects, a biharmonic flow model is adopted, and its evolution equation is:

[0118] ,

[0119] in: For the bilaplace operator, denotes the Laplace operator. Secondary applications, namely The Laplace operator on a discrete mesh is defined as follows: ,in Represents vertices The set of adjacent vertices, These are the weighting coefficients, typically set as cotangent weights. ,in and For the edge The interior angles of two adjacent triangles. Biharmonic flow can better preserve shape characteristics while smoothing curved surfaces.

[0120] Feature preservation unit 23 extracts artistic feature lines and regions of the target object, generates a feature preservation map, and preserves the artistic feature lines and regions during the restoration process. Specifically, feature preservation unit 23 first identifies artistic features based on differential geometric features, such as edges (high curvature change lines), texture boundaries, etc.; then it constructs a feature preservation map, assigning feature importance weights to each vertex; finally, during curvature flow evolution, it adjusts the evolution speed according to the feature weights to ensure that important features are preserved. In embodiments of the present invention, feature recognition is mainly based on the following geometric indices:

[0121] Abstract line: A line whose principal curvature direction is consistent with the curvature gradient direction;

[0122] Edges: Regions with high maximum principal curvature values ​​and large rates of change;

[0123] Wrinkle lines: The rate of change of curvature exceeds a threshold (typically 0.5 mm). -1 Lines in mm ( / mm);

[0124] Artistic texture: A region where the ratio of texture frequency to amplitude falls within a specific range;

[0125] The multi-scale processing unit 24 decomposes the initial triangular mesh model into different frequency components, optimizes each frequency component separately, and resynthesizes them to generate the optimized 3D model. Specifically, the multi-scale processing unit 24 includes a hierarchical geometric decomposition sub-unit 241, a frequency-selective processing sub-unit 242, and a multi-scale reconstruction sub-unit 243.

[0126] Hierarchical geometric decomposition sub-units 241 construct a multi-resolution representation of the initial triangular mesh model, decomposing the geometric information into low-frequency, mid-frequency, and high-frequency components. This decomposition process can employ the Laplace spectral decomposition method, expanding the mesh vertex coordinate functions on the Laplace eigenfunction basis.

[0127] ,

[0128] in: The function for the coordinates of the grid vertex is expressed as: A dimensional matrix, where The number of grid vertices, with each row containing one vertex. coordinate, For the Laplace operator There are 1 characteristic functions, denoted as: A dimensional column vector is one that satisfies The function, where For the corresponding eigenvalues, For the corresponding coefficients, is a 3D row vector, representing the coordinate function in the feature function. The projection on the surface.

[0129] Low-frequency components correspond to characteristic functions with smaller eigenvalues, reflecting the overall shape; high-frequency components correspond to characteristic functions with larger eigenvalues, reflecting local details. In embodiments of the present invention, the frequency component division criteria are as follows:

[0130] Low-frequency components: eigenvalues ​​less than 10% of the total eigenvalue range;

[0131] Mid-frequency components: Eigenvalues ​​range from 10% to 50% of the total eigenvalues;

[0132] High-frequency components: eigenvalues ​​greater than 50% of the total eigenvalue range.

[0133] The frequency-selective processing subunit 242 applies a global smoothing strategy to low-frequency geometric components, structural optimization processing to mid-frequency components, and detail-preserving algorithms to high-frequency components. Specifically, for low-frequency components, Laplacian smoothing is applied to eliminate large-scale deformations; for mid-frequency components, a feature enhancement filter is used to strengthen key structural features; and for high-frequency components, bilateral filtering and anisotropic filtering are combined to remove noise while preserving detailed textures. In embodiments of the present invention, examples of processing parameters for each frequency component are as follows:

[0134] Low-frequency components: Laplace smoothing intensity 0.3-0.5, number of iterations 3-5;

[0135] Mid-frequency components: Feature enhancement coefficient 1.2-1.5, threshold 0.1-0.3;

[0136] High-frequency components: bilateral filter radius 0.5-2mm, intensity 0.6-0.8.

[0137] The multi-scale reconstruction subunit 243 recombines the processed frequency components and ensures a smooth transition between different frequency components. Specifically, the multi-scale reconstruction subunit 243 recombines the optimized frequency components through inverse transformation:

[0138] ,

[0139] in: The reconstructed vertex coordinate function is expressed as: A dimensional matrix, where The number of grid vertices, with each row containing one vertex. coordinate, For the Laplace operator The eigenfunctions are the same as those used in the decomposition. The coefficients after processing are... 3D row vectors.

[0140] To ensure a smooth transition between different frequency components, a transition weighting function is introduced:

[0141] ,

[0142] in: The coefficients after processing low-frequency, mid-frequency, and high-frequency components are respectively, and are all... 3D row vectors , , The corresponding weight function is about the feature function index. The function has a range of values. And satisfy Typical weighting functions are smooth transition functions, such as variations of the sigmoid function.

[0143] Through the processing of the differential geometry-driven model reconstruction module 2, the system can repair and optimize the initial model acquired from the scan with high quality, removing noise and filling in missing parts while preserving artistic features. For example, for a wooden mask with intricate carvings, the system can accurately identify and preserve the geometric features of the carving texture, while effectively repairing data loss caused by scanning blind spots or material characteristics, generating a complete and faithful digital model.

[0144] like Figure 4 As shown, the geometric feature-driven semantic mapping module 3 is used to extract the differential invariant features of the optimized 3D model, identify the ethnic cultural style features of the optimized 3D model based on the differential invariant features, and establish the mapping relationship between the geometric features of the optimized 3D model and the cultural semantic labels.

[0145] Preferably, the geometric feature-driven semantic mapping module 3 includes a differential invariant calculation unit 31, a cultural feature recognition unit 32, a semantic ontology mapping unit 33, and a knowledge base management unit 34.

[0146] The differential invariant calculation unit 31 calculates the principal curvature, Gaussian curvature, and shape index of the optimized 3D model surface, among other differential invariants. Differential invariants are important indicators describing the geometric properties of a surface, are unaffected by coordinate transformations, and are suitable as feature descriptors. Specifically, the differential invariant calculation unit 31 first calculates the local differential properties of each vertex of the model surface; then, based on these properties, it derives various differential invariants; and finally, it constructs the model's feature descriptor vector. In embodiments of the present invention, the main differential invariants calculated include:

[0147] Principal curvatures κ1 and κ2: Curvature values ​​of the surface at a certain point along the principal direction, obtained through eigenvalue decomposition.

[0148] Gaussian curvature The product of principal curvatures reflects the intrinsic geometric properties of the surface.

[0149] ,

[0150] Where: K is the Gaussian curvature, in mm. -2 κ1 and κ2 are the maximum and minimum principal curvatures, respectively, in mm. -1 The sign of Gaussian curvature indicates the local shape of the surface: a positive value represents an ellipsoid (such as a mountain peak or valley), a zero value represents a cylinder or plane, and a negative value represents a hyperboloid (such as a saddle surface).

[0151] Mean curvature The average value of the principal curvature reflects the intrinsic geometric properties of the surface.

[0152] ,

[0153] Where: H is the mean curvature, in mm. -1 κ1 and κ2 are the maximum and minimum principal curvatures, respectively, in mm. -1 The sign of the mean curvature indicates the local concavity or convexity of the surface: a positive value indicates convexity in the direction of the normal vector, and a negative value indicates concavity in the direction of the normal vector.

[0154] Shape index S: A dimensionless parameter describing the type of local shape.

[0155] ,

[0156] Where: S is the shape index, dimensionless, ranging from [-1, 1], and κ1 and κ2 are the maximum and minimum principal curvatures, respectively, in mm. -1 , where arctan is the arctangent function. Different ranges of the shape index correspond to different local shapes: concave sphere (-1.0 to -0.75), concave cylinder (-0.75 to -0.5), concave saddle (-0.5 to -0.25), saddle (-0.25 to 0.25), convex saddle (0.25 to 0.5), convex cylinder (0.5 to 0.75), and convex sphere (0.75 to 1.0).

[0157] Rate of change of curvature and The rate of curvature change along the surface is used to identify feature lines and boundaries. Represents Gaussian curvature Directional derivative along the surface, Mean curvature Directional derivative along the surface, This represents the arc length parameter on the surface (Note: the arc length parameter here). Lowercase, similar to the aforementioned shape index (Note: The two are uppercase letters and are different mathematical quantities; the unit is mm.) The typical range is 0.01-1.0 mm. -3 , The typical range is 0.01-0.5 mm. -2 .

[0158] The cultural feature recognition unit 32 identifies typical geometric and artistic style features of ethnic handicrafts based on differential invariants. Specifically, the cultural feature recognition unit 32 first analyzes the spatial distribution patterns of differential invariants; then matches these patterns with preset cultural feature templates; and finally identifies the cultural style features of the model. In embodiments of the present invention, the main cultural features identified include:

[0159] Pattern types: By analyzing the spatial arrangement patterns of high-frequency curvature changes, common ethnic patterns such as meander, cloud, and geometric patterns can be identified.

[0160] Manufacturing process: By analyzing the differential properties of the curved surface, the manufacturing process can be deduced, such as carving, weaving, forging, etc.

[0161] Ethnic Style: Identifying the artistic style of a specific ethnic group by comprehensively analyzing the distribution of shape indices, decorative features, and production techniques.

[0162] The semantic ontology mapping unit 33 establishes association rules between geometric features and cultural semantic concepts, and calculates the association strength. Specifically, the semantic ontology mapping unit 33 first constructs a cultural semantic ontology model, defining cultural concepts and their relationships; then it establishes mapping rules between geometric features and semantic concepts; finally, it calculates the association strength of the mapping and evaluates the credibility of the mapping results. In embodiments of the present invention, the semantic ontology includes the following hierarchical structure:

[0163] Primary concept: Ethnic classification (e.g., Han, Tibetan, Mongolian, etc.);

[0164] Secondary concept: Era style (e.g., Ming and Qing Dynasties, contemporary times, etc.);

[0165] Level 3 concept: Craft type (such as wood carving, ceramics, brocade, etc.);

[0166] Level 4 concept: decorative themes (such as animals, plants, geometric shapes, etc.);

[0167] Level 5 concept: artistic expression (such as realism, abstraction, decoration, etc.).

[0168] The mapping rules are constructed based on statistical learning methods. By analyzing the correspondence between geometric features and semantic labels in labeled samples, a feature-semantic association model is generated. The association strength is calculated using the following formula:

[0169] ,

[0170] in: Features With semantic concepts The correlation strength, with a value range of . A larger value indicates a stronger association. The feature vector of the current model, For the sample The feature vectors of both have the same dimension. For the target semantic concept, For the sample semantic tags, Features With sample feature The similarity is calculated as follows: ,in This is a similarity scaling parameter, typically ranging from 0.5 to 2 times the standard deviation of the feature space. For semantic concept matching function, when and A value of 1 indicates a match, and a value of 0 indicates a non-match. Sample weights reflect the sample weights. The importance of a sample is usually determined based on its time recentity and the reliability of its annotations. The total number of samples, This is the Laplace smoothing parameter, used to avoid zero-probability problems; its typical value is 0.1-1.0. The normalization smoothing parameter takes a value of ,in This represents the total number of semantic concept categories.

[0171] The knowledge base management unit 34 maintains the model library and the association rule library, and optimizes the mapping rules based on newly added models. Specifically, the knowledge base management unit 34 is responsible for the storage, retrieval, and management of digital models; simultaneously, with the addition of new models, it continuously optimizes and updates the feature-semantic mapping rules to improve mapping accuracy. In embodiments of the present invention, the knowledge base management unit 34 adopts an incremental learning strategy, periodically updating the mapping model based on newly added data, typically every 50 new samples.

[0172] Through the processing of the geometric feature-driven semantic mapping module 3, the system can automatically identify and label the cultural features and semantic attributes of digital models, establishing a connection between geometric form and cultural connotation. For example, for a Yi ethnic wooden mask, the system can automatically identify its ethnic affiliation, production process, and decorative themes based on its geometric features, and associate them with relevant cultural background knowledge, providing semantic support for subsequent teaching applications.

[0173] like Figure 5 As shown, the teaching resource generation module 4 is used to generate art teaching resources based on cultural semantic tags and optimized 3D models, classify and store the art teaching resources in the resource library, provide multi-dimensional search interfaces, and support resource retrieval based on geometric features and semantic tags.

[0174] Preferably, the teaching resource generation module 4 includes a resource template library 41, a content customization unit 42, a three-dimensional interactive processing unit 43, and a teaching scenario construction unit 44.

[0175] The resource template library 41 stores resource templates for different teaching objectives and learning stages. Specifically, the resource template library 41 includes various preset teaching resource templates, such as courseware templates, practical activity templates, and assessment templates; these templates are categorized and organized according to teaching objectives and learning stages for easy selection and application. In embodiments of the present invention, the resource templates mainly include the following types:

[0176] Knowledge explanation category: Used to introduce theoretical knowledge such as cultural background and artistic characteristics;

[0177] Technique demonstration category: Used to demonstrate practical skills such as production processes and creative techniques;

[0178] Creative Practice Category: Used to guide learners in creative practice and artistic expression;

[0179] Comprehensive evaluation category: used to assess learning outcomes and artistic understanding.

[0180] Each type of template is further subdivided according to the learning stage (elementary school, junior high school, senior high school, university) and difficulty level (basic, intermediate, advanced), forming a complete template system.

[0181] The content customization unit 42 selects an appropriate resource template based on cultural semantic tags and teaching objectives. Specifically, the content customization unit 42 receives the cultural semantic tags output by the geometric feature-driven semantic mapping module 3; simultaneously obtains the teaching objectives and learning stage specified by the user; and then selects the most matching resource template based on this information. In this embodiment of the invention, the template selection employs a weighted matching algorithm, mainly considering the following factors:

[0182] The matching degree between semantic tags and template themes (weight 0.4);

[0183] The degree of matching between teaching objectives and template functions (weight 0.3);

[0184] The degree of matching between the learning stage and the template difficulty (weight 0.3).

[0185] The 3D interactive processing unit 43 generates interactive 3D teaching content based on the optimized 3D model. Specifically, the 3D interactive processing unit 43 first performs teaching adaptation processing on the optimized 3D model, such as simplification, decomposition, and annotation; then adds interactive functions, such as rotation, scaling, decomposition, and highlighting; and finally integrates it into the teaching content. In embodiments of the present invention, the 3D interactive functions include:

[0186] Multi-angle viewing: 360° rotation and zoom at any angle;

[0187] Structural breakdown: Displays the internal structure by component;

[0188] Feature annotation: Automatically annotates artistic features and cultural elements;

[0189] Comparative analysis: Comparing with similar or dissimilar models;

[0190] Process demonstration: Animation shows the production process.

[0191] The teaching scenario construction unit 44 integrates interactive 3D teaching content with teaching scripts to form a complete teaching resource. Specifically, based on a selected resource template, the teaching scenario construction unit 44 integrates elements such as 3D content, text descriptions, and image data; it also adds teaching guidance and assessment content to form a complete teaching resource package. In embodiments of the present invention, a typical teaching resource includes the following components:

[0192] Introduction: Cultural background introduction and explanation of learning objectives;

[0193] Main body: Interactive display of 3D models and explanation of key knowledge points;

[0194] Extended Section: Analysis of related works and inspiration for creative work;

[0195] Practical component: Creative guidance and technique practice;

[0196] Assessment section: Learning outcome assessment and self-evaluation.

[0197] Through the processing of module 4, the system can transform digital models into art education resources with educational value, supporting diverse teaching applications. For example, for a Yi ethnic wooden mask, the system can automatically generate a complete teaching resource package that includes an introduction to the cultural background, analysis of its stylistic features, demonstration of its production process, and guidance on creative practice, based on the needs of the high school art course "Ethnic and Folk Art".

[0198] like Figure 6 As shown, the immersive experience module 5 is connected to the teaching resource generation module 4. It is used to receive optimized 3D models and art teaching resources, and to build an interactive cultural experience environment based on virtual reality or augmented reality technology, supporting users to conduct multi-dimensional interaction and learning in the interactive cultural experience environment.

[0199] Preferably, the immersive experience module 5 includes a virtual scene construction unit 51, an interactive control unit 52, a cultural explanation unit 53, and a learning feedback unit 54.

[0200] The virtual scene construction unit 51 creates a cultural background scene based on the optimized 3D model. Specifically, the virtual scene construction unit 51 selects or generates a matching cultural background environment according to the cultural semantic tags of the model; then, it places the optimized 3D model into this environment to construct a complete virtual scene. In embodiments of the present invention, the virtual scene types include:

[0201] Museum-style: Simulates the environment of a museum exhibition hall, emphasizing academic rigor and systematic approach;

[0202] Original style: Recreates the original use scenario of handicrafts, emphasizing cultural authenticity;

[0203] Studio-style: Simulates an art creation environment, emphasizing skill learning and practice.

[0204] Thematic: A contextual environment customized according to the teaching theme, emphasizing storytelling and fun.

[0205] The interaction control unit 52 recognizes user action commands and converts them into interactive operations in the virtual environment. Specifically, the interaction control unit 52 receives various commands input by the user through the VR / AR device; then converts these commands into operations in the virtual environment; and finally controls the virtual objects to react accordingly. In embodiments of the present invention, supported interactive operations include:

[0206] Basic operations: grab, rotate, scale, move, etc.

[0207] Inspection operations: disassembly, sectioning, perspective, measurement, etc.;

[0208] Creative processes: drawing, carving, assembling, decorating, etc.;

[0209] Social interactions: sharing, collaboration, discussion, evaluation, etc.

[0210] The cultural explanation unit 53 provides explanations of cultural knowledge related to the 3D model based on cultural semantic tags. Specifically, the cultural explanation unit 53 extracts relevant cultural knowledge from the knowledge base based on the model's semantic tags; then, it dynamically adjusts the explanation content and method according to the user's interaction behavior and learning progress; finally, it presents the explanation to the user in the form of voice, text, images, etc. In the embodiments of the present invention, the explanation content mainly includes:

[0211] Cultural background: ethnic history, regional characteristics, social functions, etc.;

[0212] Artistic features: style of design, symbolic patterns, aesthetic concepts, etc.

[0213] Production process: material selection, techniques, tool usage, etc.;

[0214] Creative concepts: artistic expression, emotional expression, cultural significance, etc.

[0215] The learning feedback unit 54 records user learning behavior data and generates a learning effectiveness evaluation report. Specifically, the learning feedback unit 54 records the user's learning behavior and interactive operations throughout the virtual environment; then analyzes this data based on preset evaluation criteria; finally, it generates a learning effectiveness evaluation report and provides targeted learning suggestions. In embodiments of the present invention, the evaluation dimensions include:

[0216] Knowledge comprehension: The degree and depth of mastery and understanding of cultural knowledge;

[0217] Aesthetic perception: the ability to perceive artistic features and make aesthetic judgments;

[0218] Skill mastery: The ability to imitate and apply relevant skills;

[0219] Creative expression: Innovative thinking and expression based on traditional elements.

[0220] Through Module 5 of the immersive experience, the system provides an immersive learning experience that transcends traditional teaching, enabling learners to intuitively feel the charm of ethnic culture and art and deeply understand its connotations and characteristics. For example, in teaching Yi ethnic wood carving masks, learners can use VR devices to enter a simulated Yi village environment, observe the details of the masks up close, understand how they are used in sacrificial rituals, and even try to virtually create a simplified version of the mask, achieving a comprehensive and multi-sensory learning experience.

[0221] like Figure 1As shown, the quality assessment module 6 is connected to the differential geometry-driven model reconstruction module 2. It is used to assess the geometric integrity and artistic feature fidelity of the optimized 3D model, calculate the morphological deviation index between the optimized 3D model and the original object, generate a quality assessment report, and send a rescan command to the curvature-aware adaptive scanning module 1 when the deviation index exceeds a preset threshold.

[0222] Specifically, the quality assessment module 6 first checks the geometric integrity of the model, including mesh topology and surface continuity; then it assesses the fidelity of artistic features, analyzing whether the model retains the key artistic features of the original object; next, it calculates the morphological deviation index to quantify the difference between the model and the original reference data; finally, it generates a quality assessment report and decides whether a rescan or optimization is needed based on the assessment results.

[0223] In embodiments of the present invention, the main quality assessment indicators include:

[0224] Geometric integrity: Mesh quality index not less than 0.85, non-manifold structure proportion not exceeding 0.1%;

[0225] Artistic feature fidelity: Feature retention rate is not less than 90%, and feature distortion does not exceed 5%;

[0226] Morphological deviation: The average distance deviation from the reference data shall not exceed 0.1 mm, and the maximum deviation shall not exceed 0.5 mm;

[0227] When any indicator exceeds the preset threshold, the quality assessment module 6 will generate targeted improvement suggestions and, if necessary, send a rescan instruction to the curvature-aware adaptive scanning module 1, specifying the areas and parameters that need to be improved.

[0228] like Figure 7 As shown, the workflow of the ethnic cultural digital asset database and art teaching resource generation system provided by this invention mainly includes the following steps:

[0229] S1, the curvature-aware adaptive scanning module 1 pre-scans the target object, analyzes the surface curvature distribution and material properties, dynamically adjusts the scanning parameters based on the analysis results, performs fine scanning, and generates point cloud data;

[0230] S2, the differential geometry-driven model reconstruction module 2, receives point cloud data, constructs an initial triangular mesh model, identifies geometric defects, applies curvature flow theory for adaptive repair, performs multi-scale geometric processing, and generates an optimized 3D model.

[0231] S3, Quality Assessment Module 6 evaluates the quality of the optimized 3D model. If it does not meet the requirements, it returns to S1 for rescanning or optimization; if it meets the requirements, it continues to the next step.

[0232] S4, the geometric feature-driven semantic mapping module 3 extracts the differential invariant features of the optimized 3D model, identifies ethnic cultural style features, and establishes the mapping relationship between geometric features and cultural semantic labels.

[0233] S5, Teaching Resource Generation Module 4, selects appropriate resource templates based on cultural semantic tags and optimized 3D models to generate interactive 3D teaching content, constructs complete teaching resources, and stores them in the resource library in categories;

[0234] S6, Immersive Experience Module 5 receives the optimized 3D model and art teaching resources, constructs an interactive cultural experience environment based on VR / AR technology, supports users to conduct multi-dimensional interaction and learning, records learning behavior data, and generates learning effect evaluation reports.

[0235] Through the above process, this invention realizes the entire chain of processing from high-precision digital acquisition of ethnic handicrafts to generation of art teaching resources, providing a systematic solution for the digital protection and educational inheritance of ethnic culture.

[0236] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any form or substance. Those skilled in the art will make changes to the specific implementation methods and application scope based on the technical solutions and inventive concepts of the present invention, and these changes should also be considered within the scope of protection of the present invention.

Claims

1. A system for generating digital assets library of national culture and art teaching resources, characterized in that, The application relates to a teaching resource generation method based on a curvature-aware adaptive scanning module, a differential geometry-driven model reconstruction module, a geometry-feature-driven semantic mapping module and a teaching resource generation module. The curvature-aware adaptive scanning module comprises a curvature pre-analysis unit, a material characteristic analysis unit, a micro-prism array control unit and a composite light source control unit. The curvature pre-analysis unit is used for performing rapid pre-scanning of a target object, calculating a surface curvature distribution map and dividing a target object surface into high-curvature, medium-curvature and low-curvature regions. The material characteristic analysis unit is used for identifying a target object surface material type through multi-spectrum reflectivity analysis and constructing a material-reflection characteristic parameter table. The micro-prism array control unit is used for dynamically adjusting the density and arrangement mode of a micro-prism array according to the curvature distribution map and the material-reflection characteristic parameter table. The composite light source control unit is used for adjusting the intensity, wavelength and projection angle of a projection light source to adapt to different curvature regions and material characteristics. The differential geometry-driven model reconstruction module comprises a geometric defect identification unit, a curvature flow calculation unit, a feature preservation unit and a multi-scale processing unit. The geometric defect identification unit is used for analyzing the initial triangular mesh model, identifying holes, non-manifold structures and singular points and generating a defect classification map. The curvature flow calculation unit is used for constructing a differential equation logical framework describing surface evolution and controlling the geometric change of a defect region. The feature preservation unit is used for extracting artistic feature lines and regions of a target object, generating a feature protection mapping map and maintaining the artistic feature lines and regions in a repair process. The multi-scale processing unit is used for decomposing the initial triangular mesh model into different frequency components, respectively optimizing the frequency components and re-synthesizing the optimized three-dimensional model. The multi-scale processing unit comprises a hierarchical geometric decomposition subunit and a multi-scale processing subunit. The hierarchical geometric decomposition subunit is used for constructing a multi-resolution representation of the initial triangular mesh model and decomposing geometric information into low-frequency, medium-frequency and high-frequency components. The multi-scale processing subunit is used for performing multi-scale processing on the low-frequency, medium-frequency and high-frequency components and re-synthesizing the optimized three-dimensional model. The geometry-feature-driven semantic mapping module comprises a differential invariant feature extraction unit, a cultural style feature recognition unit and a geometric feature-cultural semantic label mapping unit. The differential invariant feature extraction unit is used for extracting differential invariant features of the optimized three-dimensional model. The cultural style feature recognition unit is used for recognizing the cultural style features of the optimized three-dimensional model based on the differential invariant features. The geometric feature-cultural semantic label mapping unit is used for establishing a mapping relationship between the geometric features of the optimized three-dimensional model and cultural semantic labels.

2. The system of claim 1, wherein, The teaching resource generation module comprises a teaching resource generation unit, a resource storage unit and a multi-dimensional retrieval interface. The teaching resource generation unit is used for generating art teaching resources according to the cultural semantic labels and the optimized three-dimensional model. The resource storage unit is used for classifying and storing the art teaching resources into a resource library. The multi-dimensional retrieval interface is used for supporting resource retrieval based on geometric features and semantic labels. ​ 3. The system of claim 1, wherein, ​ ​ ​ ​ ​ 4. The system of claim 3, wherein, ​ ​ The frequency selective processing subunit is configured to apply a global smoothing strategy to low-frequency geometric components, a structure optimization process to medium-frequency components, and a detail preservation algorithm to high-frequency components; The multi-scale reconstruction subunit is configured to recombine the processed frequency components and ensure smooth transitions between different frequency components.

5. The system of claim 1, wherein, The geometry feature driven semantic mapping module comprises: A differential invariant calculation unit configured to calculate differential invariants of the optimized three-dimensional model surface, such as principal curvatures, Gaussian curvatures, and shape indices; A cultural feature recognition unit configured to recognize typical geometric features and artistic style features of ethnic handicrafts based on the differential invariants; A semantic ontology mapping unit configured to establish association rules between geometric features and cultural semantic concepts and calculate association strengths; A knowledge base management unit configured to maintain a model library and an association rule library and optimize mapping rules based on newly added models.

6. The system of claim 1, wherein, The teaching resource generation module comprises: A resource template library configured to store resource templates for different teaching objectives and learning stages; A content customization unit configured to select appropriate resource templates according to the cultural semantic tags and teaching objectives; A three-dimensional interactive processing unit configured to generate interactive three-dimensional teaching content based on the optimized three-dimensional model; A teaching scene construction unit configured to integrate the interactive three-dimensional teaching content with a teaching script to form complete teaching resources.

7. The system of claim 1, wherein, Further comprising: An immersive experience module connected to the teaching resource generation module and configured to: Receive the optimized three-dimensional model and the art teaching resources; Construct an interactive cultural experience environment based on virtual reality or augmented reality technology; Support multi-dimensional interaction and learning of users in the interactive cultural experience environment.

8. The system of claim 7, wherein, The immersive experience module comprises: A virtual scene construction unit configured to create a cultural background scene based on the optimized three-dimensional model; An interactive control unit configured to recognize user action instructions and convert them into interactive operations in the virtual environment; A cultural explanation unit configured to provide cultural knowledge explanations related to the three-dimensional model according to the cultural semantic tags; A learning feedback unit configured to record user learning behavior data and generate a learning effect evaluation report.

9. The system of claim 1, wherein, The structure light projection parameters in the curvature-aware adaptive scanning module comprise: Micro-prism array density, with a density range of 400-900 micro-prisms per square centimeter in high-curvature areas and a density range of 50-200 micro-prisms per square centimeter in low-curvature areas; Projection light intensity, adjusted within a range of 20-200 lumens according to the reflectivity of the target object surface; Scanning resolution, with a resolution of 0.01-0.05 millimeters in high-curvature areas and a resolution of 0.1-0.5 millimeters in low-curvature areas.

10. The system of claim 1, wherein, Further comprising: A quality evaluation module connected to the differential geometry driven model reconstruction module and configured to: Evaluate the geometric integrity and artistic feature fidelity of the optimized three-dimensional model; Calculate a morphological deviation index of the optimized three-dimensional model from the original object; Generate a quality evaluation report and send a re-scanning instruction to the curvature-aware adaptive scanning module when the deviation index exceeds a preset threshold.

Images