Artificial Intelligence-Assisted Digital Reproduction System for Cultural Relics

Abstract

The present invention relates to the field of cultural heritage protection and restoration technology, and specifically to an artificial intelligence-assisted cultural relic digital reproduction system, including a data acquisition module, a data processing module, a data storage module, an artificial intelligence analysis module, an interactive display module, and a user interaction module, wherein the data acquisition module acquires the geometric shape, color, material, and surface texture data of the cultural relic, the data processing module generates a three-dimensional model of the cultural relic, the data storage module stores the three-dimensional model data, the artificial intelligence analysis module extracts the historical information and cultural background of the cultural relic, the interactive display module combines virtual reality technology to realize the interactive display of the cultural relic, and the user interaction module dynamically adjusts the display content of the cultural relic according to the user's input information. The artificial intelligence analysis module of the present invention analyzes the three-dimensional model data of the cultural relic, extracts the historical information and cultural background of the cultural relic, and while viewing the cultural relic, understands the historical and cultural stories behind it, thereby increasing the educational significance of the cultural relic display.

Description

Technical Field

[0001] The present invention relates to the technical field of cultural heritage protection and restoration, and in particular to an artificial intelligence-assisted cultural relic digital reproduction system. Background Art

[0002] With the continuous development of cultural heritage protection and digital technology, the digital reconstruction and display of cultural relics has become an important task for museums and cultural heritage institutions. Through the digital reconstruction and display of cultural relics, people can watch the cultural relics up close and display every detail of them in detail. Compared with the traditional viewing method through glass, this method provides a more in-depth and interactive viewing experience.

[0003] However, there are still some technical problems in the digital display of cultural relics with existing technologies. First, the accuracy and integrity of data collection need to be improved, especially when dealing with cultural relics with complex shapes and multiple materials, data loss and errors are prone to occur. Secondly, the interactivity of cultural relics display is insufficient, and traditional display methods cannot meet the needs of modern users for immersive and interactive experience. In addition, it is impossible to fully utilize the three-dimensional model data of cultural relics to extract valuable historical information and cultural background. Summary of the invention

[0004] Based on the above objectives, the present invention provides an artificial intelligence-assisted digital reproduction system for cultural relics.

[0005] The artificial intelligence-assisted cultural relic digital reproduction system includes a data acquisition module, a data processing module, a data storage module, an artificial intelligence analysis module, an interactive display module, and a user interaction module, among which;

[0006] The data acquisition module acquires the geometric shape, color, material and surface texture data of the cultural relics through three-dimensional scanners, photogrammetry equipment and sensors;

[0007] The data processing module receives and processes the data acquired by the data acquisition module to generate a three-dimensional model of the cultural relics;

[0008] The data storage module stores the cultural relics three-dimensional model data generated by the data processing module;

[0009] The artificial intelligence analysis module analyzes the three-dimensional model data of the cultural relics, extracts the historical information and cultural background of the cultural relics, and generates corresponding explanation texts;

[0010] The interactive display module realizes interactive display of cultural relics based on the interpretation text generated by the artificial intelligence analysis module and combined with virtual reality technology;

[0011] The user interaction module receives user input information and dynamically adjusts the display content of the cultural relics according to the user input information.

[0012] Furthermore, the data acquisition module specifically includes:

[0013] 3D scanner: 3D scanner uses laser or structured light technology to scan the surface of cultural relics and generate point cloud data and surface texture information;

[0014] Photogrammetry equipment: Use cameras to take photos of cultural relics from multiple angles to obtain image data of cultural relics, including information on their color and material;

[0015] Multispectral and environmental sensors: Use multispectral sensors to scan cultural relics at different wavelengths to obtain multispectral data, including the material and surface texture characteristics of the cultural relics. Environmental sensors detect the environmental data of the cultural relics, including temperature, humidity and light intensity.

[0016] Furthermore, the data processing module specifically includes:

[0017] Data reception and preprocessing: receiving point cloud data acquired by 3D scanners, cultural relic image data acquired by photogrammetry equipment, multispectral and environmental sensor data, removing noise and abnormal points in the point cloud data, filtering the point cloud data, denoising, color correction and geometric correction of the image to ensure image quality, performing spectral analysis on multispectral data, and normalizing environmental data to ensure data consistency and accuracy;

[0018] Point cloud data processing: Use the ICP (Iterative Closest Point) algorithm to align point cloud data from different perspectives, fuse the aligned point cloud data, and generate a 3D point cloud model;

[0019] 3D mesh generation: Use triangulation algorithm to convert point cloud data into 3D mesh data and generate 3D mesh model;

[0020] Texture mapping: Use the SIFT algorithm to detect common feature points in point cloud data and image data, match the feature points, ensure the spatial alignment of point cloud data and image data, assign UV coordinates to each vertex of the 3D mesh model. UV coordinates are coordinates in the 2D texture space and are used to map the image to the surface of the 3D model. Based on the feature point matching results, the image data is projected onto the surface of the 3D mesh model, and the multi-view image data is combined to generate a complete texture mapping.

[0021] Sensor data fusion: Fuse multispectral data into the texture mapping process to enhance texture details and material performance. Use texture information at different wavelengths in multispectral data to enhance the details of the model surface.

[0022] Generate a three-dimensional model: Combine the processed geometric shape data and texture data to generate a three-dimensional model of the cultural relic. The geometric shape data includes point cloud data and three-dimensional mesh data, and the texture data includes image data and multi-spectral data.

[0023] Furthermore, the data storage module specifically includes:

[0024] Data storage format selection: point cloud data uses PLY (Polygon File Format) file format, 3D mesh data uses OBJ (Wavefront Object) file format, image data uses JPEG file format, and multi-spectral data uses multi-layer TIFF format;

[0025] Data reception and classification: Receive the geometric shape data and texture data generated by the data processing module, and classify and store the data into corresponding folders according to the data type;

[0026] Data access and management: Use NoSQL database to store data indexes, provide fast retrieval functions, and regularly back up stored data to ensure data security and recoverability.

[0027] Furthermore, the artificial intelligence analysis module specifically includes:

[0028] Data input and preprocessing: input geometric shape data and texture data, perform denoising and filtering, and extract features from the geometric shape data and texture data, including shape, texture, and color;

[0029] Cultural relic feature analysis: Use shape descriptors to extract the geometric features of the 3D model, match the extracted geometric features with known shapes in the preset cultural relic database, identify the type and age of the cultural relic, use the LBP texture analysis algorithm to extract the texture features of the cultural relic surface, analyze the material composition of the cultural relic through multispectral data, and identify the composition and characteristics of different materials;

[0030] Extraction of historical information and cultural background of cultural relics: Use geometric features, texture features, and material features to search in the preset cultural relics database to find matching cultural relics information, perform text analysis on the matched cultural relics information, and extract key information, including the historical background, cultural significance, and usage scenarios of the cultural relics;

[0031] Generate explanatory text: Use the pre-trained language model to generate explanatory text, integrate the extracted historical information and cultural background into the explanatory text, add background information and detailed descriptions, and improve the richness and readability of the text.

[0032] Furthermore, the use of shape descriptors to extract geometric features of the three-dimensional model specifically includes:

[0033] The ORB (Oriented FAST and Rotated BRIEF) extraction algorithm is used to extract feature points and descriptors from the three-dimensional model, and the point cloud data is projected onto a two-dimensional plane. The projection formula is expressed as:

[0034] Where (x, y, z) are the coordinates of the three-dimensional point, f is the focal length, and (x′, y′) are the coordinates projected onto the two-dimensional plane;

[0035] ORB combines FAST key point detection and Hartis corner point measurement for feature point detection. FAST key point detection includes the following steps:

[0036] Select a pixel p and its surrounding 16 pixels and compare the grayscale value I of pixel p p and the grayscale values ​​of the surrounding pixels. If there are 12 or more consecutive neighboring pixels, their grayscale values ​​are greater than I p +t is greater than or equal to I p -t is small, then p is considered to be a corner point, where t is the threshold;

[0037] Harris corner measurement: Calculate the Harris response value of each FAST key point, expressed as:

[0038] R = der(M) - k (trace(M)) 2 , where M is the autocorrelation matrix, defined as:

[0039] Among them, I x and I y is the gradient of the image, w(u, v) is the window function;

[0040] Select key points with high Harris response values;

[0041] ORB direction calculation: Calculate the direction of each key point to obtain rotation invariance. The grayscale centroid method calculates the direction as follows: Where P is the pixel neighborhood around the key point p, and I(x, y) is the grayscale value of the pixel;

[0042] ORB descriptor extraction: ORB uses the rotated BRIEF descriptor, including:

[0043] Generate descriptors: For each key point, define a sampling pattern containing n point pairs, compare the grayscale values ​​of each pair of points, generate a binary descriptor, and rotate the sampling pattern of BRIEF to the direction of the key point θ.

[0044] Furthermore, the material composition of the cultural relics is analyzed by multispectral data to identify the components and characteristics of different materials, specifically including:

[0045] Multispectral data acquisition and preprocessing: Use multispectral sensors to photograph cultural relics at multiple wavelengths (such as ultraviolet, visible light, near infrared, etc.) to generate multi-band images, with each band corresponding to one wavelength. Perform radiometric and geometric calibration on the multispectral data to ensure consistency between different bands, and apply mean filtering to reduce image noise;

[0046] Spectral feature extraction: Calculate the reflectance of each pixel at different wavelengths. The reflectance R(λ) is determined by the sensor's observed value I(λ) and the reflectance value I(λ) of the reference white plate. ref (λ), we can calculate: The spectral feature vector S is extracted for each pixel, which contains the reflectivity of the pixel at different wavelengths and is expressed as: S = [R(λ1), R(λ2), ..., R(λ n )];

[0047] Material feature analysis: The spectral angle mapping algorithm is used to match the extracted spectral feature vector with the preset spectral library. The spectral library contains standard reflectance spectra of various materials at different wavelengths. The spectral angle mapping algorithm is expressed as: Where S is the spectral feature vector of the material to be tested, S ref is the spectral vector of the reference material in the spectral library, and θ is the spectral angle between the two. The smaller the angle, the more similar the materials are.

[0048] Classification and clustering: Use spectral feature vectors to classify and cluster different areas of the artifact surface to identify different material components;

[0049] Material component identification: Match the extracted spectral feature vector with the features of known materials in the spectral library to identify the components of different materials.

[0050] Furthermore, the interactive display module specifically includes:

[0051] Input data: 3D model of cultural relics (geometric shape data and texture data), interpretation text generated by the artificial intelligence analysis module;

[0052] Virtual reality technology: Hardware includes VR head-mounted display devices (such as Oculus Rift, HTC Vive), positioning sensors, and handheld controllers; software includes VR content production tools (such as Unity, Unreal Engine) and VR content display platforms;

[0053] 3D model loading and scene construction: Import 3D models into VR content production tools, create a virtual display environment, place the 3D model in the virtual environment, and set lighting, background and interactive elements.

[0054] Furthermore, the user interaction module receives the user's input information, specifically including:

[0055] Input type: receiving user touch operations through a touch screen device and receiving user voice commands through a microphone;

[0056] Input processing: Process touch events of touch screen devices, obtain touch position and action type, and use speech recognition technology (such as Google Speech-to-Text, Amazon Transcribe) to convert voice commands into text;

[0057] Dynamic adjustment strategy: When the user clicks on the preset part of the cultural relic, the cultural relic information window pops up, displaying the explanatory text generated by the artificial intelligence analysis module. The user slides or clicks on the preset area to trigger the decomposition, combination or dynamic demonstration of the cultural relic. The user controls the display of the cultural relic through voice commands, including zooming in, out and rotation.

[0058] The present invention uses advanced 3D scanners, photogrammetry equipment and multispectral sensors to comprehensively and efficiently obtain the geometric shape, color, material and surface texture data of cultural relics. This greatly reduces the data loss and errors that may occur when processing complex shapes and multi-material cultural relics, and ensures the high accuracy and integrity of the digital reconstruction of cultural relics.

[0059] The present invention combines virtual reality technology to realize interactive display of cultural relics. Users can interact with cultural relics through touch operation and voice commands, dynamically adjust the display content, and provide a richer and more immersive viewing experience. Compared with traditional display methods, the present invention significantly improves user participation and experience, and meets the needs of modern users for interactive and immersive experience.

[0060] The artificial intelligence analysis module of the present invention conducts in-depth analysis on the three-dimensional model data of cultural relics, extracts the historical information and cultural background of the cultural relics, and generates detailed explanatory texts. In this way, users can learn about the historical and cultural stories behind the cultural relics while viewing them, which increases the educational significance and cultural value of the cultural relics display and provides a more comprehensive and in-depth viewing experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0062] Figure 1 A schematic diagram of a system flow of an embodiment of the present invention;

[0063] Figure 2 The figure is a flowchart of a data processing module according to an embodiment of the present invention. DETAILED DESCRIPTION

[0064] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.

[0065] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0066] like Figure 1-Figure 2 As shown, the artificial intelligence-assisted cultural relics digital reproduction system includes a data acquisition module, a data processing module, a data storage module, an artificial intelligence analysis module, an interactive display module and a user interaction module, wherein;

[0067] The data acquisition module acquires the geometric shape, color, material and surface texture data of cultural relics through 3D scanners, photogrammetry equipment and sensors;

[0068] The data processing module receives and processes the data acquired by the data acquisition module to generate a three-dimensional model of the cultural relics;

[0069] The data storage module stores the cultural relics three-dimensional model data generated by the data processing module;

[0070] The artificial intelligence analysis module analyzes the three-dimensional model data of cultural relics, extracts the historical information and cultural background of the cultural relics, and generates corresponding explanatory texts;

[0071] The interactive display module uses the interpretation text generated by the artificial intelligence analysis module and combines it with virtual reality technology to realize the interactive display of cultural relics;

[0072] The user interaction module receives the user's input information and dynamically adjusts the display content of the cultural relics according to the user's input information.

[0073] The data acquisition module specifically includes:

[0074] 3D scanner: 3D scanner uses laser or structured light technology to scan the surface of cultural relics and generate point cloud data and surface texture information;

[0075] Photogrammetry equipment: Use high-definition cameras to take photos of cultural relics from multiple angles to obtain image data of cultural relics, including information on their color and material;

[0076] Multispectral and environmental sensors: Use multispectral sensors to scan cultural relics at different wavelengths to obtain multispectral data, including the material and surface texture characteristics of the cultural relics. Environmental sensors detect the environmental data of the cultural relics, including temperature, humidity and light intensity.

[0077] The data processing module specifically includes:

[0078] Data reception and preprocessing: receiving point cloud data acquired by 3D scanners, cultural relic image data acquired by photogrammetry equipment, multispectral and environmental sensor data, removing noise and abnormal points in the point cloud data, filtering the point cloud data, denoising, color correction and geometric correction of the image to ensure image quality, performing spectral analysis on multispectral data, and normalizing environmental data to ensure data consistency and accuracy;

[0079] Point cloud data processing: Use ICP (Iterative Cl0sest Point) algorithm to align point cloud data from different perspectives, fuse the aligned point cloud data, and generate a 3D point cloud model;

[0080] The ICP algorithm specifically includes:

[0081] Goal: Align two sets of point cloud data P and Q so that their weights The parts are as consistent as possible;

[0082] Input: point cloud P = {p i} and Q = {q j};

[0083] Output: transformation matrix T (including rotation matrix R and translation vector t);

[0084] The ICP algorithm is calculated as:

[0085] Through iterative optimization, a rotation matrix R and translation vector t are found so that point p in point cloud P i After transformation T, the point q closest to the point cloud Q nearest(i) The sum of squared distances between them is minimal;

[0086] 3D mesh generation: Use triangulation algorithm to convert point cloud data into 3D mesh data and generate 3D mesh model;

[0087] The triangulation algorithm consists of the following steps:

[0088] Objective: Convert point cloud data into a 3D mesh model;

[0089] Input: point cloud P = {p i};

[0090] Output: triangle mesh;

[0091] The triangulation satisfies the following conditions: for each triangle in the triangular mesh, its circumscribed circle does not contain any other points;

[0092] Texture mapping: Use the SIFT algorithm to detect common feature points in point cloud data and image data, match the feature points, ensure the spatial alignment of point cloud data and image data, assign UV coordinates to each vertex of the 3D mesh model. UV coordinates are coordinates in the 2D texture space and are used to map the image to the surface of the 3D model. Based on the feature point matching results, the image data is projected onto the surface of the 3D mesh model, and the multi-view image data is combined to generate a complete texture mapping.

[0093] The SIFT (Scale Invariant Feature Transform) algorithm includes the following steps:

[0094] Goal: Detect and match feature points in an image;

[0095] Input: image I;

[0096] Output: Feature point set F = {f i} and the descriptor set D = {d i};

[0097] Key point detection: D(x, y, σ) = (G(x, y, kσ) - G(x, y, σ)) * I(x, y);

[0098] Direction assignment:

[0099] Perform Gaussian blur processing on image I to obtain differential images D(x, y, σ) at different scales, detect extreme points as feature points, assign directions to feature points, and ensure that feature points are invariant in scale and rotation;

[0100] The UV coordinates are calculated as: u and v represent the UV coordinates, x and y are the coordinates of the vertex in the texture image, and W and H are the width and height of the texture image;

[0101] Texture projection formula: I(x, y) = T(U, V), maps the UV coordinates of the texture image T to the coordinates I(x, y) of the 3D model to achieve texture mapping;

[0102] Sensor data fusion: Integrate multispectral data into the texture mapping process to enhance texture details and material performance. Use texture information at different wavelengths in multispectral data to enhance the details of the model surface. For example, ultraviolet and infrared light scanning can reveal fine cracks and repair marks on the surface of cultural relics.

[0103] Multispectral data fusion is expressed as: Among them, R final Represents the final reflectivity, R i represents the reflectivity at the i-th wavelength, w i is the weight coefficient, which realizes material recognition and texture enhancement by weighted averaging the reflectivity at different wavelengths;

[0104] Generate 3D model: Combine the processed geometric shape data and texture data to generate a 3D model of the cultural relic. The geometric shape data includes point cloud data and 3D mesh data, and the texture data includes image data and multispectral data.

[0105] Through the above steps, the data processing module can efficiently and accurately process and fuse the geometric shape, color, material and surface texture data obtained by the data acquisition module to generate a high-precision three-dimensional model of the cultural relics, providing a solid technical foundation for the digital reproduction of the cultural relics.

[0106] The data storage module specifically includes:

[0107] Data storage format selection: point cloud data uses PLY (Polygon File Format) file format, 3D mesh data uses OBJ (Wavefront Object) file format, image data uses JPEG file format, and multi-spectral data uses multi-layer TIFF format;

[0108] Data reception and classification: Receive the geometry data and texture data generated by the data processing module, and classify and store the data into corresponding folders according to the data type (the geometry data folder stores point cloud data and three-dimensional mesh data, and the texture data folder stores image data and multispectral data);

[0109] Data access and management: Use NoSQL database to store data indexes, provide fast retrieval functions, and regularly back up stored data to ensure data security and recoverability.

[0110] The AI ​​analysis module specifically includes:

[0111] Data input and preprocessing: input geometric shape data and texture data, perform denoising and filtering, and extract features from the geometric shape data and texture data, including shape, texture, and color;

[0112] Cultural relic feature analysis: Use shape descriptors to extract the geometric features of the 3D model, match the extracted geometric features with known shapes in the preset cultural relic database, identify the type and age of the cultural relic, use the LBP texture analysis algorithm to extract the texture features of the cultural relic surface, analyze the material composition of the cultural relic through multispectral data, and identify the composition and characteristics of different materials;

[0113] Extraction of historical information and cultural background of cultural relics: Use geometric features, texture features, and material features to search in the preset cultural relics database to find matching cultural relics information, perform text analysis on the matched cultural relics information, and extract key information, including the historical background, cultural significance, and usage scenarios of the cultural relics;

[0114] Generate explanatory text: Use the pre-trained language model to generate explanatory text, integrate the extracted historical information and cultural background into the explanatory text, add background information and detailed descriptions, and improve the richness and readability of the text;

[0115] The LBP (Local Binary Patterns) texture analysis algorithm is a descriptor used for texture feature extraction. It describes the texture features of the local area of ​​the image, including:

[0116] Calculate the LBP value of the central pixel: For each pixel in the image, select its 3×3 neighborhood, take the gray value of the central pixel as the threshold, and compare it with the gray values ​​of the surrounding 8 pixels. If the gray value of the neighborhood pixel is greater than or equal to the gray value of the central pixel, the pixel position is 1, otherwise it is 0. The 8 comparison results constitute an 8-bit binary number, namely the LBP value, which is specifically expressed as:

[0117] Among them, I c is the gray value of the center pixel, Ik is the gray value of the neighborhood pixel, and the function s(x) is defined as:

[0118] Construct LBP feature histogram: Calculate the LBP values ​​of all pixels in the entire image, count the frequency of each LBP pattern, and construct a 256-dimensional histogram as the texture feature of the image.

[0119] The geometric features of the 3D model extracted using shape descriptors include:

[0120] The ORB (Oriented FAST and Rotated BRIEF) extraction algorithm is used to extract feature points and descriptors from the three-dimensional model, and the point cloud data is projected onto a two-dimensional plane. The projection formula is expressed as:

[0121] Where (x, y, z) are the coordinates of the three-dimensional point, f is the focal length, and (x′, y′) are the coordinates projected onto the two-dimensional plane;

[0122] ORB combines FAST key point detection and Harris corner point measurement for feature point detection. FAST key point detection includes the following steps:

[0123] Select a pixel p and its surrounding 16 pixels and compare the grayscale value I of pixel p p and the grayscale values ​​of the surrounding pixels. If there are 12 or more consecutive neighboring pixels, their grayscale values ​​are greater than I p +t is greater than or equal to I p -t is small, then p is considered to be a corner point, where t is the threshold;

[0124] Harris corner measurement: Calculate the Harris response value of each FAST key point, expressed as:

[0125] R = det(M) - k (trace(M)) 2 , where M is the autocorrelation matrix, defined as:

[0126] Among them, I x and I y is the gradient of the image, w(u, v) is the window function;

[0127] Select key points with high Harris response values;

[0128] ORB direction calculation: Calculate the direction of each key point to obtain rotation invariance. The grayscale centroid method calculates the direction as follows: Where P is the pixel neighborhood around the key point p, and I(x, y) is the grayscale value of the pixel;

[0129] ORB descriptor extraction: ORB uses the rotated BRIEF descriptor, including:

[0130] Generate descriptors: For each key point, define a sampling pattern containing n point pairs, compare the grayscale values ​​of each pair of points, generate a binary descriptor, and rotate the sampling pattern of BRIEF to the direction θ of the key point;

[0131] Sampling mode rotation:

[0132] Binary descriptor:

[0133] Analyze the material composition of cultural relics through multispectral data and identify the components and characteristics of different materials, including:

[0134] Multispectral data acquisition and preprocessing: Use multispectral sensors to photograph cultural relics at multiple wavelengths (such as ultraviolet, visible light, near infrared, etc.) to generate multi-band images, with each band corresponding to one wavelength. Perform radiometric and geometric calibration on the multispectral data to ensure consistency between different bands, and apply mean filtering to reduce image noise;

[0135] Spectral feature extraction: Calculate the reflectance of each pixel at different wavelengths. The reflectance R(λ) is determined by the sensor's observed value I(λ) and the reflectance value I(λ) of the reference white plate. ref (λ), we can calculate: The spectral feature vector S is extracted for each pixel, which contains the reflectivity of the pixel at different wavelengths and is expressed as: S = [R(λ1), R(λ2), ..., R(λ n )];

[0136] Material feature analysis: The spectral angle mapping algorithm is used to match the extracted spectral feature vector with the preset spectral library. The spectral library contains standard reflectance spectra of various materials at different wavelengths. The spectral angle mapping algorithm is expressed as: Where S is the spectral feature vector of the material to be tested, S ref is the spectral vector of the reference material in the spectral library, and θ is the spectral angle between the two. The smaller the angle, the more similar the materials are.

[0137] Classification and clustering: Use spectral feature vectors to classify and cluster different areas of the artifact surface to identify different material components, using the K-means clustering method, including:

[0138] Initially set K cluster centers μ k ;

[0139] Iteratively update clustering: assign each pixel to the nearest cluster center, expressed as:

[0140] c i =argmin k S i -μ k ;

[0141] Update cluster centers:

[0142] Material component identification: Match the extracted spectral feature vector with the features of known materials in the spectral library to identify the components of different materials.

[0143] The interactive display module specifically includes:

[0144] Input data: 3D model of cultural relics (geometric shape data and texture data), interpretation text generated by the artificial intelligence analysis module;

[0145] Virtual reality technology: Hardware includes VR head-mounted display devices (such as Oculus Rift, HTC Vive), positioning sensors, and handheld controllers; software includes VR content production tools (such as Unity, Unreal Engine) and VR content display platforms;

[0146] 3D model loading and scene construction: Import 3D models into VR content production tools, create a virtual display environment, place the 3D model in the virtual environment, and set lighting, background and interactive elements.

[0147] The user interaction module receives the user's input information, including:

[0148] Input type: receiving user touch operations through a touch screen device and receiving user voice commands through a microphone;

[0149] Input processing: Process touch events of touch screen devices, obtain touch position and action type, and use speech recognition technology (such as Google Speech-to-Text, Amazon Transcribe) to convert voice commands into text;

[0150] Dynamic adjustment strategy: When the user clicks on the preset part of the cultural relic, the cultural relic information window pops up, displaying the explanatory text generated by the artificial intelligence analysis module. The user slides or clicks on the preset area to trigger the decomposition, combination or dynamic demonstration of the cultural relic. The user controls the display of the cultural relic through voice commands, including zooming in, out and rotation.

[0151] Those skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of the present invention is limited to these examples. Under the concept of the present invention, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of the different aspects of the present invention as described above, which are not provided in detail for the sake of simplicity.

[0152] The present invention is intended to cover all such substitutions, modifications and variations that fall within the broad scope of the claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. The artificial intelligence-assisted digital reproduction system of cultural relics is characterized by: It includes data acquisition module, data processing module, data storage module, artificial intelligence analysis module, interactive display module and user interaction module, among which; The data acquisition module obtains the geometric shape, color, material and surface texture data of the cultural relics through three-dimensional scanners, photogrammetry equipment and sensors, specifically including: 3D scanner: 3D scanner uses laser or structured light technology to scan the surface of cultural relics and generate point cloud data and surface texture information; Photogrammetry equipment: Use cameras to take photos of cultural relics from multiple angles to obtain image data of cultural relics, including information on their color and material; Multispectral and environmental sensors: Use multispectral sensors to scan cultural relics at different wavelengths to obtain multispectral data, including the material and surface texture characteristics of the cultural relics. Environmental sensors detect the environmental data of the cultural relics, including temperature, humidity and light intensity. The data processing module receives and processes the data acquired by the data acquisition module to generate a three-dimensional model of the cultural relics, specifically including: Data reception and preprocessing: receiving point cloud data acquired by 3D scanners, cultural relic image data acquired by photogrammetry equipment, multispectral and environmental sensor data, removing noise and abnormal points in the point cloud data, filtering the point cloud data, denoising, color correction and geometric correction of the image, performing spectral analysis on the multispectral data, and normalizing the environmental data to ensure data consistency and accuracy; Point cloud data processing: Use the ICP algorithm to align point cloud data from different perspectives, fuse the aligned point cloud data, and generate a 3D point cloud model; 3D mesh generation: Use triangulation algorithm to convert point cloud data into 3D mesh data and generate 3D mesh model; Texture mapping: Use the SIFT algorithm to detect common feature points in point cloud data and image data, match the feature points to ensure spatial alignment of point cloud data and image data, assign UV coordinates to each vertex of the 3D mesh model, project the image data onto the surface of the 3D mesh model based on the feature point matching results, and generate a complete texture mapping by combining multi-view image data; Sensor data fusion: Fuse multispectral data into the texture mapping process to enhance texture details and material performance. Use texture information at different wavelengths in multispectral data to enhance the details of the model surface. Generate 3D model: Combine the processed geometric shape data and texture data to generate a 3D model of the cultural relic. The geometric shape data includes point cloud data and 3D mesh data, and the texture data includes image data and multispectral data. The data storage module stores the cultural relics three-dimensional model data generated by the data processing module; The artificial intelligence analysis module analyzes the three-dimensional model data of the cultural relics, extracts the historical information and cultural background of the cultural relics, and generates corresponding interpretation texts, including: Data input and preprocessing: input geometric shape data and texture data, perform denoising and filtering, and extract features from the geometric shape data and texture data, including shape, texture, and color; Cultural relic feature analysis: Use shape descriptors to extract the geometric features of the 3D model, match the extracted geometric features with known shapes in the preset cultural relic database, identify the type and age of the cultural relic, use the LBP texture analysis algorithm to extract the texture features of the cultural relic surface, analyze the material composition of the cultural relic through multispectral data, and identify the composition and characteristics of different materials; Extraction of historical information and cultural background of cultural relics: Use geometric features, texture features, and material features to search in the preset cultural relics database to find matching cultural relics information, perform text analysis on the matched cultural relics information, and extract key information, including the historical background, cultural significance, and usage scenarios of the cultural relics; Generate explanatory text: Use the pre-trained language model to generate explanatory text, integrate the extracted historical information and cultural background into the explanatory text, add background information and detailed descriptions, and improve the richness and readability of the text; The method of extracting the geometric features of the three-dimensional model using the shape descriptor specifically includes: The ORB extraction algorithm is used to extract feature points and descriptors from the three-dimensional model, and the point cloud data is projected onto a two-dimensional plane. The projection formula is expressed as: ,in are the coordinates of the three-dimensional point, is the focal length, are the coordinates projected onto the two-dimensional plane; ORB combines FAST key point detection and Harris corner point measurement for feature point detection. FAST key point detection includes the following steps: Select a pixel and the 16 pixels around it, comparing the pixels Gray value and the grayscale values ​​of the surrounding pixels. If there are 12 or more consecutive neighboring pixels, their grayscale values ​​are Bigger or bigger Small, then is considered to be a corner point, where is the threshold value; Harris corner measurement: Calculate the Harris response value of each FAST key point, expressed as: ,in, is the autocorrelation matrix, defined as: ,in, and is the gradient of the image, is a window function; Select key points with high Harris response values; ORB direction calculation: Calculate the direction of each key point to obtain rotation invariance. The grayscale centroid method calculates the direction as follows: ,in, It is the key point The surrounding pixel neighborhood, is the gray value of the pixel; ORB descriptor extraction: ORB uses the rotated BRIEF descriptor, including: Generate descriptors: For each key point, define a The sampling mode of the point pairs is used to compare the grayscale values ​​of each pair of points and generate binary descriptors according to the direction of the key points. , rotate the sampling mode of BRIEF to this direction; The interactive display module realizes interactive display of cultural relics based on the interpretation text generated by the artificial intelligence analysis module and combined with virtual reality technology; The user interaction module receives user input information and dynamically adjusts the display content of the cultural relics according to the user input information.

2. The artificial intelligence-assisted cultural relics digital reproduction system according to claim 1, characterized in that: The data storage module specifically includes: Data storage format selection: point cloud data uses PLY file format, 3D mesh data uses OBJ file format, image data uses JPEG file format, and multi-spectral data uses multi-layer TIFF format; Data reception and classification: Receive the geometric shape data and texture data generated by the data processing module, and classify and store the data into corresponding folders according to the data type; Data access and management: Use NoSQL database to store data indexes, provide fast retrieval functions, and regularly back up stored data to ensure data security and recoverability.

3. The artificial intelligence-assisted cultural relics digital reproduction system according to claim 1, characterized in that: The method of analyzing the material composition of cultural relics through multispectral data and identifying the components and characteristics of different materials specifically includes: Multispectral data acquisition and preprocessing: Use multispectral sensors to photograph cultural relics at multiple wavelengths to generate multi-band images, with each band corresponding to one wavelength. Perform radiometric and geometric calibration on the multispectral data to ensure consistency between different bands, and apply mean filtering to reduce image noise. Spectral feature extraction: Calculate the reflectance of each pixel at different wavelengths, reflectance Observations from sensors and the reflectance value of the reference white plate , we calculate: , extract the spectral feature vector S for each pixel, which contains the reflectivity of the pixel at different wavelengths, expressed as: ; Material feature analysis: The spectral angle mapping algorithm is used to match the extracted spectral feature vector with the preset spectral library. The spectral angle mapping algorithm is expressed as: ,in is the spectral feature vector of the material to be tested, is the spectral vector of the reference material in the spectral library, is the spectral angle between the two; Classification and clustering: Use spectral feature vectors to classify and cluster different areas of the artifact surface to identify different material components; Material component identification: Match the extracted spectral feature vector with the features of known materials in the spectral library to identify the components of different materials.

4. The artificial intelligence-assisted cultural relics digital reproduction system according to claim 1, characterized in that: The interactive display module specifically includes: Input data: 3D model of cultural relics, explanatory text generated by artificial intelligence analysis module; Virtual reality technology: Hardware includes VR head-mounted display devices, positioning sensors, and handheld controllers; software includes VR content production tools and VR content display platforms; 3D model loading and scene construction: Import 3D models into VR content production tools, create a virtual display environment, place the 3D model in the virtual environment, and set lighting, background and interactive elements.

5. The artificial intelligence-assisted cultural relics digital reproduction system according to claim 1, characterized in that: The user interaction module receives the user's input information, specifically including: Input type: receiving user touch operations through a touch screen device and receiving user voice commands through a microphone; Input processing: Process touch events of touch screen devices, obtain touch position and action type, and use speech recognition technology to convert voice commands into text; Dynamic adjustment strategy: When the user clicks on the preset part of the cultural relic, the cultural relic information window pops up, displaying the explanatory text generated by the artificial intelligence analysis module. The user slides or clicks on the preset area to trigger the decomposition, combination or dynamic demonstration of the cultural relic. The user controls the display of the cultural relic through voice commands, including zooming in, out and rotation.

Images