Digital protection system and restoration method for cultural ruins and images

Abstract

The invention discloses a cultural ruin and image digital protection system and restoration method, and relates to the technical field of cultural heritage protection, and the system comprises a digital collection module, an intelligent restoration module, a diversified transmission module, and a safety protection and copyright protection module. And the digital acquisition module is used for carrying out all-directional and multi-angle digital acquisition on the cultural ruins by adopting high-precision three-dimensional scanning, unmanned aerial vehicle aerial photography and high-definition camera shooting technologies. According to the cultural ruin and image digital protection system and restoration method provided by the invention, through high-precision three-dimensional scanning, unmanned aerial vehicle aerial photography and high-definition camera shooting technologies, the cultural ruin can be digitally acquired in an omnibearing and multi-angle manner, and accurate three-dimensional models, high-definition texture images and video data are acquired; the public participation sense and interactivity are enhanced, the situation that resources are illegally obtained and used is effectively prevented, and powerful legal and technical guarantee is provided for digital protection of cultural heritage.

Description

Technical Field

[0001] The present invention relates to the technical field of cultural heritage protection, and in particular to a cultural heritage site and image digital protection system and restoration method. Background Art

[0002] With the rapid development of science and technology, digital technology has become an important means of cultural heritage protection. However, there are still many problems in the digital protection and dissemination of cultural relics and images. Specifically, although digital technology has been widely used in the field of cultural heritage protection, in actual operation, the digitization accuracy is often insufficient, and it is difficult to completely restore the original details and characteristics of cultural relics and their related images. At the same time, due to various historical reasons and natural factors, many cultural relics and images are damaged, scratched and faded, and the current restoration technology is not ideal, and it is difficult to effectively restore their original appearance. In addition, the dissemination channels of digital resources are relatively limited, which leads to the limitation of the dissemination scope and influence of cultural heritage. In this regard, we propose a digital protection system and restoration method for cultural relics and images. Summary of the invention

[0003] In order to solve the above technical problems, a cultural relic and image digital protection system and restoration method are provided. This technical solution solves the above problems.

[0004] In order to achieve the above purpose, the technical solution adopted by the present invention is:

[0005] Cultural relics and image digital protection system, including: digital acquisition module, intelligent restoration module, diversified communication module and security protection and copyright protection module;

[0006] The digital acquisition module is used to use high-precision 3D scanning, drone aerial photography and high-definition camera technology to conduct all-round and multi-angle digital acquisition of cultural relics, obtain accurate 3D models, high-definition texture images and video data of the relics, and perform high-definition digital processing of images, including image denoising, color correction and resolution enhancement;

[0007] The intelligent restoration module is used to develop intelligent restoration technology for cultural relics and images based on deep learning algorithms. By training the neural network model, it can automatically identify and repair damage, scratches and fading in the image, restore the original appearance of the image, and use virtual reality technology to virtually reconstruct the missing parts in the three-dimensional model of the ruins to ensure the integrity and ornamental value of the ruins.

[0008] The diversified communication module is used to build a digital resource library of cultural relics and images based on cloud computing and big data technology, realize centralized storage and efficient retrieval of resources, and develop diversified communication channels to enable the public to browse and learn characteristic culture anytime and anywhere. The communication channels include: mobile applications, online exhibition platforms and virtual reality experiences;

[0009] The security protection and copyright protection module is used to use encryption technology and digital watermark technology to provide security protection and copyright protection for digital resources, ensuring the legal use and dissemination of resources.

[0010] Preferably, the digital acquisition module includes: a three-dimensional scanning unit, a drone aerial photography unit, a high-definition camera unit and an image processing unit;

[0011] The 3D scanning unit is used to digitally collect cultural relics using high-precision 3D scanning technology to obtain accurate 3D models of the relics;

[0012] The drone aerial photography unit is used to shoot cultural relics and surrounding environments using drone aerial photography technology to obtain high-definition images and video materials;

[0013] The high-definition camera unit is used to photograph cultural relics and obtain high-definition texture images and video data;

[0014] The image processing unit is used to perform high-definition digital processing on images, including image denoising, color correction, and resolution enhancement.

[0015] Preferably, the intelligent restoration module includes: a deep learning algorithm unit, an image restoration unit and a virtual reality reconstruction unit;

[0016] The deep learning algorithm unit is used to automatically identify damage, scratches and fading issues in images by training a neural network model;

[0017] The image restoration unit is used to automatically repair damage, scratches and fading problems in the image based on the recognition results of the deep learning algorithm unit, and restore the original appearance of the image;

[0018] The virtual reality reconstruction unit is used to virtually reconstruct the missing parts in the three-dimensional model of the site using virtual reality technology to ensure the integrity and viewing experience of the site.

[0019] Preferably, the diversified communication module includes: a resource library construction unit, a mobile application development unit, an online exhibition platform development unit and a virtual reality experience development unit;

[0020] The resource library construction unit is used to build a digital resource library of cultural relics and images based on cloud computing and big data technology, realizing centralized storage and efficient retrieval of resources;

[0021] The mobile application development unit is used to develop mobile applications, enabling the public to browse and learn about special culture through mobile devices anytime and anywhere, and providing interactive functions;

[0022] The online exhibition platform development unit is used to develop an online exhibition platform to provide the public with a channel to browse and learn about special culture online;

[0023] The Virtual Reality Experience Development Unit is used to develop virtual reality experiences to allow the public to experience the charm of cultural heritage sites more immersively.

[0024] Preferably, the security protection and copyright protection module includes: an encryption technology unit and a digital watermark technology unit;

[0025] The encryption technology unit is used to use encryption technology to protect digital resources and prevent them from being illegally obtained and used;

[0026] The digital watermark technology unit is used to use digital watermark technology to protect the copyright of digital resources and ensure the legal use and dissemination of resources.

[0027] Preferably, the automatic identification of damage, scratches and fading problems in images by training a neural network model specifically includes:

[0028] Collect a large number of images of cultural relics with different degrees of damage, scratches and fading problems, as well as normal images as controls;

[0029] Preprocessing the image, including normalization and resizing;

[0030] The input layer receives the preprocessed image data, and the image size is n×n×c, where n is the side length of the image and c is the number of channels of the image;

[0031] The convolution layer uses multiple convolution kernels to perform convolution operations on the image to extract the features of the image. The convolution operation is expressed as:

[0032]

[0033] In the formula, represents the value of the output of the lth layer at the coordinate (i, j), f represents the activation function, Represents the value of the convolution kernel of the lth layer at the coordinate (m,n), represents the value of the l-1th layer at the coordinate (i+m,j+n), b l represents the bias term of the lth layer;

[0034] The pooling layer downsamples the output of the convolutional layer, reducing the number of parameters and computation while retaining the main features;

[0035] The fully connected layer flattens the feature map processed by the convolutional layer and the pooling layer into a one-dimensional vector, connects it to the fully connected layer, and performs a comprehensive analysis of the image features to determine whether there are damage, scratches, and fading problems in the image;

[0036] The output layer outputs a probability value, indicating that there are damage, scratches, and fading problems in the image. The sigmoid activation function is used to map the output value to the [0,1] input interval, where the expression of the sigmoid activation function is:

[0037]

[0038] In the formula, u represents the output of the neuron, v represents the input of the neuron, and e represents a natural constant;

[0039] The cross entropy loss function is used to measure the difference between the model prediction and the true label, where the cross entropy loss function is expressed as:

[0040]

[0041] In the formula, A represents the total number of samples, y i represents the true label, Represents the probability value predicted by the model;

[0042] Use stochastic gradient descent to continuously adjust the model parameters to minimize the loss function. The stochastic gradient descent update formula is:

[0043]

[0044] In the formula, C represents the parameters of the model, α represents the learning rate, Represents the gradient of the loss function with respect to the parameters;

[0045] The preprocessed image data is divided into a training set and a validation set. The model is trained using the training set, and the performance of the model is regularly evaluated using the validation set. During the training process, the model parameters are continuously adjusted until the model's performance on the validation set is optimal, and the trained model is evaluated using the test set.

[0046] Input the image of the cultural heritage site to be identified into the trained model, and the model will automatically output the probability value of the presence of damage, scratches and fading problems in the image.

[0047] Preferably, according to the recognition result of the deep learning algorithm unit, automatically repairing the damage, scratches and fading problems in the image and restoring the original appearance of the image specifically includes:

[0048] Receiving image region information of damage, scratches and fading problems identified by the deep learning algorithm unit, and preparing a set of image samples similar to the image to be repaired that are not damaged, scratched or faded, for feature extraction and reference repair;

[0049] Based on the acquired image to be repaired and the undamaged image, extract the features of the image, the features including: texture, color, edge, and establish a feature database;

[0050] For the damaged area, the boundary is determined by detecting the edge of the damaged area and the surrounding normal area. Based on the partial differential equation method, the damaged area is gradually filled from the boundary to the inside, so that the pixel value of the damaged area gradually tends to be consistent with the pixel value of the surrounding normal area.

[0051] For scratches, the Canny edge detection algorithm is used to detect the edges in the image, and the edges of the scratches are repaired by interpolation;

[0052] For the fading problem, the color features of the faded area are determined, similar color information is extracted from the surrounding normal area for filling, the color histogram of the faded area is matched with the color histogram of the surrounding normal area using the color histogram matching method, and the color of the faded area is adjusted according to the matching result;

[0053] The restored image is smoothed to remove unnatural transitions and noise.

[0054] Preferably, the partial differential equation method specifically includes:

[0055] Assume that the pixel value I(x,y) inside the damaged area needs to be filled with the surrounding known pixel values;

[0056] By solving the heat diffusion equation, the pixel values ​​of the damaged area gradually become consistent with the pixel values ​​of the surrounding normal area. The expression of the heat diffusion equation is:

[0057]

[0058] in, represents the Laplace operator, that is, the second-order derivative of the image, In a two-dimensional image it is represented as:

[0059]

[0060] Preferably, repairing by interpolation method specifically includes:

[0061] Assume that the point to be interpolated is between two known points (x1, y1) and (x2, y2), and its value is f(x, y), then the linear interpolation formula is:

[0062]

[0063] In scratch repair, linear interpolation is used to infer the pixel value at the scratch based on the pixel values ​​on both sides of the scratch.

[0064] The method for digital restoration of cultural relics and images is characterized in that it is used to implement the cultural relics and image digital protection system, and comprises:

[0065] Use a high-precision 3D scanner to scan the site in all directions and obtain 3D point cloud data of the site;

[0066] Use drones to take aerial photos and obtain high-definition images and video materials of the site and its surroundings;

[0067] Perform high-definition digital processing on the collected images and video materials to remove noise, correct colors, and improve resolution;

[0068] The processed images and video data are input into the intelligent restoration system and automatically restored using deep learning algorithms;

[0069] Upload the restored digital resources to the cloud storage platform to build a digital resource library of cultural relics and images;

[0070] Develop mobile applications and online exhibition platforms to open digital resources to the public and provide browsing, learning and interactive functions.

[0071] Compared with the prior art, the present invention has the following beneficial effects:

[0072] The cultural relics and image digital protection system and restoration method proposed in the present invention can digitally collect cultural relics in all directions and from multiple angles through high-precision three-dimensional scanning, drone aerial photography and high-definition camera technology, and obtain accurate three-dimensional models, high-definition texture images and video materials, which greatly improves the digitization accuracy, so that the original details and features of the cultural relics and their related images can be more completely preserved and presented, and can automatically identify and repair damage, scratches and fading problems in the images, improve the restoration efficiency, and more accurately restore the original appearance of the images, so that the ornamental value and integrity of the cultural relics are significantly improved, the dissemination scope and influence of cultural heritage are expanded, and the public's sense of participation and interactivity are enhanced, ensuring the legal use and dissemination of resources, effectively preventing the illegal acquisition and use of resources, and providing strong legal and technical guarantees for the digital protection of cultural heritage. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] Figure 1 The figure is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0074] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are only examples, and those skilled in the art may think of other obvious variations.

[0075] Reference Figure 1 As shown, the cultural relics and image digital restoration methods include:

[0076] Use 3D scanners to conduct all-round, high-precision scanning of cultural relics to ensure the most detailed 3D point cloud data of the relics. At the same time, dispatch multiple drones of different models to shoot the relics and surrounding environment from multiple angles to obtain high-resolution images and video data. These drones are equipped with professional high-definition cameras that can capture every detail of the relics, including the texture, color and surrounding natural landscape. Combined with ground shooting equipment, professional photographers will take close-up photos of the relics to supplement the parts that may be missed by drones and 3D scanners. Archaeologists and historians can be invited to guide the shooting on site to ensure that the content of the shooting has historical value and academic significance.

[0077] Perform preliminary screening on the collected images and videos to remove blurry, repeated and poor quality parts, and then use professional image processing software to perform color correction, contrast adjustment and sharpening on the images to make them clearer and more vivid. For videos, perform denoising and stabilization to improve the quality of the videos;

[0078] We use advanced image stitching technology to stitch multiple images into a complete panoramic view of the site. We edit and synthesize videos to produce a wonderful introduction video of the site. At the same time, we use 3D modeling software to convert 3D point cloud data into realistic 3D models to better display the form and structure of the site. In order to ensure the accuracy and professionalism of digital processing, we invite professional image and video processing experts to participate in the processing process. According to the characteristics and historical background of the site, we develop the most appropriate processing plan to make the processed data more consistent with historical reality.

[0079] The processed data is input into the intelligent restoration module. This module uses artificial intelligence technology to automatically identify the damaged parts of the site and repair them, restoring the color of the site to its original color. For parts that have faded due to time and natural factors, the system can infer and repair them based on historical data and the colors of similar cultural relics, making the color of the site more realistic.

[0080] During the automatic restoration process, the system will continuously learn and optimize the restoration algorithm to improve the accuracy and efficiency of the restoration. At the same time, professional restoration personnel will review and adjust the restoration results to ensure that the restored site conforms to historical authenticity.

[0081] Categorize and organize the restored resources according to the different parts of the site, the types of cultural relics and the historical periods, so that users can find and use them better. At the same time, add detailed descriptions and labels to each resource, including the name of the site, location, historical background, and restoration methods;

[0082] Using advanced database technology, a digital resource library is constructed, which can store a large number of images, videos, and 3D models, and provide fast retrieval and browsing functions. Users can access the resource library anytime and anywhere through mobile applications and online platforms to understand the historical and cultural value of cultural relics.

[0083] In order to ensure the security and stability of the resource library, multiple backup and encryption technologies are used to prevent resource loss and tampering. At the same time, the resource library is regularly maintained and updated, and new repair resources and historical data are added to make the resource library richer and more complete;

[0084] Develop powerful mobile applications and online platforms to open the digital resources of cultural relics to the public. Mobile applications can provide convenient access methods, and users can browse images, videos and 3D models of relics anytime and anywhere through their mobile phones. Online platforms can provide richer functions, including virtual tours, interactive communication, academic research, etc.

[0085] On mobile applications and online platforms, various display methods are set up, including panoramic display, virtual reality display, and augmented reality display, allowing users to feel the charm of cultural relics more intuitively. At the same time, multiple language versions are provided for the convenience of domestic and foreign users. In order to increase public participation and interactivity, interactive communication sections are set up on mobile applications and online platforms, where users can post comments, ask questions and share their insights. At the same time, online activities are held to stimulate the public's interest and love for cultural relics.

[0086] The cultural heritage and image digital protection system includes: digital acquisition module, intelligent restoration module, diversified communication module and security and copyright protection module.

[0087] By training the neural network model, the damage, scratches and fading problems in the image are automatically identified, including:

[0088] Collect a large number of images of cultural relics with different degrees of damage, scratches and fading problems, as well as normal images as controls;

[0089] Preprocess the image, which includes normalization and resizing. Normalization is one of the important steps in image preprocessing. A standardized normalization method is used to adjust the pixel value of the image to a specific range, usually [0, 1] or [-1, 1], so that the pixel values ​​of different images are comparable and the influence of illumination and contrast factors on model training is reduced. Resizing is also an indispensable step. All images are adjusted to a uniform size so that they can be input into the neural network model for processing. When adjusting the size, attention should be paid to maintaining the proportion and clarity of the image to avoid deformation or blur. An interpolation algorithm is used to ensure the quality of the adjusted image. At the same time, considering the influence of images of different sizes on the performance of the model, multiple sizes are tried for experimentation to select the most suitable size.

[0090] The input layer receives the preprocessed image data, and the image size is n×n×c, where n is the side length of the image and c is the number of channels of the image;

[0091] The convolution layer uses multiple convolution kernels to perform convolution operations on the image to extract the features of the image. The convolution operation is expressed as:

[0092]

[0093] In the formula, represents the value of the output of the lth layer at the coordinate (i, j), f represents the activation function, Represents the value of the convolution kernel of the lth layer at the coordinate (m,n), represents the value of the l-1th layer at the coordinate (i+m,j+n), b l represents the bias term of the lth layer;

[0094] The pooling layer downsamples the output of the convolutional layer, reducing the number of parameters and computation while retaining the main features;

[0095] The fully connected layer flattens the feature map processed by the convolutional layer and the pooling layer into a one-dimensional vector, connects it to the fully connected layer, and performs a comprehensive analysis of the image features to determine whether there are damage, scratches, and fading problems in the image;

[0096] The output layer outputs a probability value, indicating that there are damage, scratches, and fading problems in the image. The sigmoid activation function is used to map the output value to the [0,1] input interval, where the expression of the sigmoid activation function is:

[0097]

[0098] In the formula, u represents the output of the neuron, v represents the input of the neuron, and e represents a natural constant;

[0099] The cross entropy loss function is used to measure the difference between the model prediction and the true label. The cross entropy loss function is a commonly used loss function, which is suitable for binary classification problems. The cross entropy loss function is expressed as:

[0100]

[0101] In the formula, A represents the total number of samples, y i represents the true label, Represents the probability value predicted by the model. The smaller the value of the cross entropy loss function, the closer the model's prediction is to the true label.

[0102] Use stochastic gradient descent to continuously adjust the model parameters to minimize the loss function. The stochastic gradient descent update formula is:

[0103]

[0104] In the formula, C represents the parameters of the model, α represents the learning rate, Represents the gradient of the loss function with respect to the parameters;

[0105] The preprocessed image data is divided into a training set and a validation set. The model is trained using the training set, and the performance of the model is regularly evaluated using the validation set. During the training process, the model parameters are continuously adjusted until the model's performance on the validation set is optimal, and the trained model is evaluated using the test set.

[0106] The image of the cultural relic to be identified is input into the trained model, and the model will automatically output the probability value of the presence of damage, scratches and fading problems in the image. Based on the size of the probability value, it can determine whether there is a problem in the image and the severity of the problem.

[0107] Based on the recognition results of the deep learning algorithm unit, damage, scratches and fading problems in the image are automatically repaired to restore the original appearance of the image. Specifically, the following are included:

[0108] Receive information about image areas with damage, scratches and fading problems identified by the deep learning algorithm unit, and prepare a set of undamaged, unscratched and unfaded image samples similar to the image to be repaired for feature extraction and reference repair. This information will serve as an important basis for the subsequent repair process to ensure that the repair work can be carried out in a targeted manner. These samples can be obtained from multiple channels, such as professional image databases, high-quality images in historical archives, and intact images of cultural relics of the same type. In the collection process, attention should be paid to the diversity and representativeness of the samples, covering different shooting angles, lighting conditions and color styles, so as to provide rich references for the repair work;

[0109] Based on the acquired image to be repaired and the undamaged image, extract the features of the image, the features including: texture, color, edge, and establish a feature database;

[0110] For the damaged area, the detection algorithm is first used to determine the edge between the damaged area and the surrounding normal area. The edge detection algorithm is combined with the morphological processing method to clearly outline the boundary of the damaged area. During the detection process, different types of damage, such as cracks, gaps, and missing parts, must be taken into account. Different edge detection strategies are adopted for different types of damage.

[0111] Based on the partial differential equation method, the damaged area is gradually filled from the boundary. This method can gradually deduce the pixel value inside the damaged area by solving the partial differential equation according to the pixel value information of the surrounding normal area. In the filling process, attention should be paid to maintaining the continuity and naturalness of the filling, avoiding obvious transition marks and discordant color changes. A multi-scale filling strategy can be adopted to gradually fill the damaged area from coarse to fine, so as to improve the filling accuracy and quality;

[0112] For scratches, the Canny edge detection algorithm is used. The Canny edge detection algorithm has high accuracy and stability, and can effectively detect various edge information in the image. It detects the edges in the image and repairs the edges of scratches through interpolation.

[0113] For the fading problem, determine the color features of the faded area, extract similar color information from the surrounding normal area for filling, use the color histogram matching method to match the color histogram of the faded area with the color histogram of the surrounding normal area, and adjust the color of the faded area according to the matching result;

[0114] The restored image is smoothed to remove unnatural transitions and noise.

[0115] Partial differential equation methods specifically include:

[0116] Assume that the pixel value I(x,y) inside the damaged area needs to be filled with the surrounding known pixel values;

[0117] By solving the heat diffusion equation, the pixel values ​​of the damaged area gradually become consistent with the pixel values ​​of the surrounding normal area. The expression of the heat diffusion equation is:

[0118]

[0119] in, represents the Laplace operator, that is, the second-order derivative of the image, In a two-dimensional image it is represented as:

[0120]

[0121] The repair by interpolation method specifically includes:

[0122] Assume that the point to be interpolated is between two known points (x1, y1) and (x2, y2), and its value is f(x, y), then the linear interpolation formula is:

[0123]

[0124] In scratch repair, linear interpolation is used to infer the pixel value at the scratch based on the pixel values ​​on both sides of the scratch.

[0125] The usage process of the present invention is as follows: use a high-precision 3D scanner to perform an all-round scan of the site to obtain 3D point cloud data of the site; use a drone to take aerial photos to obtain high-definition images and video data of the site and its surrounding environment; perform high-definition digital processing on the collected images and video data to remove noise, correct color, and improve resolution; input the processed images and video data into an intelligent restoration system, and use a deep learning algorithm to perform automatic restoration; upload the restored digital resources to a cloud storage platform to build a digital resource library of cultural relics and images; develop mobile applications and online exhibition platforms, open digital resources to the public, and provide browsing, learning, and interactive functions.

[0126] In summary, the advantages of the present invention are: improving accuracy by collecting data through multiple technologies, automatically repairing images to improve viewing and integrity, expanding the scope of dissemination to enhance public participation, ensuring the legal use of resources, and providing legal and technical guarantees for the digital protection of cultural heritage.

[0127] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions only describe the principles of the present invention. The present invention may be subject to various changes and improvements without departing from the spirit and scope of the present invention. These changes and improvements fall within the scope of the present invention. The scope of protection claimed by the present invention is defined by the attached claims and their equivalents.

Claims

1. Cultural relics and image digital protection system, characterized by: include: Digital acquisition module, intelligent restoration module, diversified dissemination module, and security and copyright protection module; The digital acquisition module is used to use high-precision 3D scanning, drone aerial photography and high-definition camera technology to conduct all-round and multi-angle digital acquisition of cultural relics, obtain accurate 3D models, high-definition texture images and video data of the relics, and perform high-definition digital processing of images, including image denoising, color correction and resolution enhancement; The intelligent restoration module is used to develop intelligent restoration technology for cultural relics and images based on deep learning algorithms. By training the neural network model, it can automatically identify and repair damage, scratches and fading in the image, restore the original appearance of the image, and use virtual reality technology to virtually reconstruct the missing parts in the three-dimensional model of the ruins to ensure the integrity and ornamental value of the ruins. The diversified communication module is used to build a digital resource library of cultural relics and images based on cloud computing and big data technology, realize centralized storage and efficient retrieval of resources, and develop diversified communication channels to enable the public to browse and learn characteristic culture anytime and anywhere. The communication channels include: mobile applications, online exhibition platforms and virtual reality experiences; The security protection and copyright protection module is used to use encryption technology and digital watermark technology to provide security protection and copyright protection for digital resources, ensuring the legal use and dissemination of resources.

2. The cultural heritage and image digital protection system according to claim 1, characterized in that: The digital acquisition module includes: 3D scanning unit, drone aerial photography unit, high-definition camera unit and image processing unit; The 3D scanning unit is used to digitally collect cultural relics using high-precision 3D scanning technology to obtain accurate 3D models of the relics; The drone aerial photography unit is used to shoot cultural relics and surrounding environments using drone aerial photography technology to obtain high-definition images and video materials; The high-definition camera unit is used to photograph cultural relics and obtain high-definition texture images and video data; The image processing unit is used to perform high-definition digital processing on images, including image denoising, color correction, and resolution enhancement.

3. The cultural heritage and image digital protection system according to claim 1, characterized in that: The intelligent restoration module includes: deep learning algorithm unit, image restoration unit and virtual reality reconstruction unit; The deep learning algorithm unit is used to automatically identify damage, scratches and fading issues in images by training a neural network model; The image restoration unit is used to automatically repair damage, scratches and fading problems in the image based on the recognition results of the deep learning algorithm unit, and restore the original appearance of the image; The virtual reality reconstruction unit is used to virtually reconstruct the missing parts in the three-dimensional model of the site using virtual reality technology to ensure the integrity and viewing experience of the site.

4. The cultural heritage and image digital protection system according to claim 1, characterized in that: The diversified communication modules include: resource library construction unit, mobile application development unit, online exhibition platform development unit and virtual reality experience development unit; The resource library construction unit is used to build a digital resource library of cultural relics and images based on cloud computing and big data technology, realizing centralized storage and efficient retrieval of resources; The mobile application development unit is used to develop mobile applications, enabling the public to browse and learn about special culture through mobile devices anytime and anywhere, and providing interactive functions; The online exhibition platform development unit is used to develop an online exhibition platform to provide the public with a channel to browse and learn about special culture online; The Virtual Reality Experience Development Unit is used to develop virtual reality experiences to allow the public to experience the charm of cultural heritage sites more immersively.

5. The cultural heritage and image digital protection system according to claim 1, characterized in that: The security protection and copyright protection module includes: encryption technology unit and digital watermark technology unit; The encryption technology unit is used to use encryption technology to protect digital resources and prevent them from being illegally obtained and used; The digital watermark technology unit is used to use digital watermark technology to protect the copyright of digital resources and ensure the legal use and dissemination of resources.

6. The cultural heritage site and image digital protection system according to claim 3, characterized in that: By training the neural network model, the damage, scratches and fading problems in the image are automatically identified, including: Collect a large number of images of cultural relics with different degrees of damage, scratches and fading problems, as well as normal images as controls; Preprocessing the image, including normalization and resizing; The input layer receives the preprocessed image data, and the image size is n×n×c, where n is the side length of the image and c is the number of channels of the image; The convolution layer uses multiple convolution kernels to perform convolution operations on the image to extract the features of the image. The convolution operation is expressed as: In the formula, represents the value of the output of the lth layer at the coordinate (i, j), f represents the activation function, Represents the value of the convolution kernel of the lth layer at the coordinate (m,n), represents the value of the l-1th layer at the coordinate (i+m,j+n), b l represents the bias term of the lth layer; The pooling layer downsamples the output of the convolutional layer, reducing the number of parameters and computation while retaining the main features; The fully connected layer flattens the feature map processed by the convolutional layer and the pooling layer into a one-dimensional vector, connects it to the fully connected layer, and performs a comprehensive analysis of the image features to determine whether there are damage, scratches, and fading problems in the image; The output layer outputs a probability value, indicating that there are damage, scratches, and fading problems in the image. The sigmoid activation function is used to map the output value to the [0,1] input interval, where the expression of the sigmoid activation function is: In the formula, u represents the output of the neuron, v represents the input of the neuron, and e represents a natural constant; The cross entropy loss function is used to measure the difference between the model prediction and the true label, where the cross entropy loss function is expressed as: In the formula, A represents the total number of samples, y i represents the true label, Represents the probability value predicted by the model; Use stochastic gradient descent to continuously adjust the model parameters to minimize the loss function. The stochastic gradient descent update formula is: In the formula, C represents the parameters of the model, α represents the learning rate, Represents the gradient of the loss function with respect to the parameters; The preprocessed image data is divided into a training set and a validation set. The training set is used to train the model. The validation set is used to regularly evaluate the performance of the model. During the training process, the model parameters are continuously adjusted until the model performs best on the validation set. The trained model is evaluated using the test set. Input the image of the cultural heritage site to be identified into the trained model, and the model will automatically output the probability value of the presence of damage, scratches and fading problems in the image.

7. The cultural heritage and image digital protection system according to claim 3 is characterized in that: Based on the recognition results of the deep learning algorithm unit, damage, scratches and fading problems in the image are automatically repaired to restore the original appearance of the image. Specifically, the following are included: Receiving image area information of damage, scratches and fading problems identified by the deep learning algorithm unit, and preparing a set of image samples similar to the image to be repaired that are not damaged, scratched or faded, for feature extraction and reference repair; Based on the acquired image to be repaired and the undamaged image, extract the features of the image, the features including: texture, color, edge, and establish a feature database; For the damaged area, the boundary is determined by detecting the edge of the damaged area and the surrounding normal area. Based on the partial differential equation method, the damaged area is gradually filled from the boundary to the inside, so that the pixel value of the damaged area gradually tends to be consistent with the pixel value of the surrounding normal area. For scratches, the Canny edge detection algorithm is used to detect the edges in the image, and the edges of the scratches are repaired by interpolation; For the fading problem, determine the color features of the faded area, extract similar color information from the surrounding normal area for filling, use the color histogram matching method to match the color histogram of the faded area with the color histogram of the surrounding normal area, and adjust the color of the faded area according to the matching result; The restored image is smoothed to remove unnatural transitions and noise.

8. The cultural heritage and image digital protection system according to claim 7, characterized in that: Partial differential equation methods specifically include: Assume that the pixel value I(x,y) inside the damaged area needs to be filled with the surrounding known pixel values; By solving the heat diffusion equation, the pixel values ​​of the damaged area gradually become consistent with the pixel values ​​of the surrounding normal area. The expression of the heat diffusion equation is: in, represents the Laplace operator, that is, the second-order derivative of the image, In a two-dimensional image it is represented as:

9. The cultural heritage site and image digital protection system according to claim 7, characterized in that: The repair by interpolation method specifically includes: Assume that the point to be interpolated is between two known points (x1, y1) and (x2, y2), and its value is f(x, y), then the linear interpolation formula is: In scratch repair, linear interpolation is used to infer the pixel value at the scratch based on the pixel values ​​on both sides of the scratch.

10. A method for digital restoration of cultural relics and images, characterized in that: A system for digitalizing and protecting cultural relics and images as claimed in any one of claims 1 to 9, comprising: Use a high-precision 3D scanner to scan the site in all directions and obtain 3D point cloud data of the site; Use drones to take aerial photos and obtain high-definition images and video materials of the site and its surroundings; Perform high-definition digital processing on the collected images and video materials to remove noise, correct colors, and improve resolution; The processed images and video data are input into the intelligent restoration system and automatically restored using deep learning algorithms; Upload the restored digital resources to the cloud storage platform to build a digital resource library of cultural relics and images; Develop mobile applications and online exhibition platforms to open digital resources to the public and provide browsing, learning and interactive functions.

Images