A movable cultural relic registration method based on AID authentication technology
Through AID forensic technology and deep learning, the automated positioning and feature recognition of cultural relics are achieved, the error and inconsistency problems in traditional cultural relics registration methods are solved, the accuracy and management convenience of cultural relics are improved, and the efficiency and accuracy of cultural relics are achieved. Register and enter.
Patent Information
- Application Number
- CN202410681683.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-05-29
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2044-05-29
AI Technical Summary
The traditional cultural relics registration and entry methods lack automation and intelligence, which leads to errors and inconsistencies in logging and recording, making it difficult to fully capture the micro-character and fiber feature information of cultural relics, and is inconvenient to manage and cannot be traced.
Using a movable cultural relics login method based on AID forensic technology, the scanning system automatically locates and tracks cultural relics, combines deep learning and machine learning technology to identify and analyze microscopic features and fiber features, establish a cultural relics feature database, and realize automated classification and entry. .
It improves the accuracy and authority of cultural relics information, reduces the burden of manual identification, improves the accuracy and efficiency of identification, and realizes traceability and convenient management of cultural relics.
Smart Images

Figure CN118644706B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of artwork supervision, and in particular to a movable cultural relic registration method based on AID authentication technology. Background Art
[0002] As an important part of the cultural and economic fields, cultural relics and artworks play a vital role in international trade. However, due to the existence of problems such as counterfeiting and smuggling, the identification and supervision of artworks have become particularly important.
[0003] For example, the patent with publication number CN113240437A discloses a blockchain-based art supervision platform and management method, wherein the supervision platform includes: supervision nodes, operating nodes, authentication nodes, evidence nodes and purchasing nodes, wherein: the operating node broadcasts corresponding information materials to the authentication node and discloses the smart contract interface; the authentication node, after confirming that the rights and interests of the artwork are valid, forms the artwork's non-homogeneous digital assets, equity proof and digital asset safe, and stores them in the blockchain evidence node, and broadcasts them publicly to the blockchain supervision platform; the purchasing node completes the docking with the smart contract, participates in the art transaction according to the agreement of the smart contract, and adopts the above-mentioned technical solution to cover the entire process of the creation (or acquisition), authentication, transaction and equity transfer of artworks. The information on the chain is difficult to tamper with, thereby realizing information transparency of the art industry chain. At the same time, the regulator has the only highest authority and can conduct effective supervision at any time.
[0004] Although the above scheme has the above advantages, the traditional cultural relics registration and entry methods lack automation and intelligence, and most operations require manual intervention and processing, which are prone to errors and inconsistencies in registration and entry. At the same time, the traditional manual registration and entry method is limited by manpower and time, and it is difficult to fully capture the microscopic characteristics and fiber characteristic information of cultural relics. There are also problems such as inconvenient and non-traceable cultural relics management. Therefore, a movable cultural relics registration and entry method based on AID authentication technology is urgently needed to solve such problems. Summary of the invention
[0005] In view of the shortcomings of the prior art, the present invention provides a movable cultural relic registration method based on AID authentication technology, which solves the problems that most operations in the prior art require manual intervention and processing, are prone to errors and inconsistencies in registration and entry, are difficult to fully capture the microscopic features and fiber feature information of the cultural relics, and have the problems of inconvenient cultural relic management and non-traceability.
[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions:
[0007] The present invention provides a movable cultural relic registration method based on AID authentication technology, comprising:
[0008] Step 1. AID authentication and cultural relic scanning: Place the cultural relic to be registered in the scanning area and scan it through the scanning system; the scanning system includes a camera, a sensor, and a processing chip; the cultural relic scanning uses target detection and real-time tracking technology to automatically locate and track the cultural relic, and collect images and related data of the cultural relic; and establish an artwork ID card, a database and query system for the certification and filing of cultural relics and cultural works in the customs bonded area, and monitor the tax payment status of the works in the bonded area;
[0009] Step 2. Micro-feature recognition: Based on deep learning technology, a micro-feature recognition model is trained to identify the micro-features of cultural relics. The scanning system analyzes the scanned image data and extracts the micro-features of cultural relics, including texture and particles.
[0010] Step 3. Multi-point fiber feature detection, deploying the fiber feature detection method of regional extraction and multi-scale analysis, detecting the fiber features at different positions of the cultural relics, and recording the fiber feature information of each part of the cultural relics;
[0011] Step 4. Feature data storage: associate the microscopic feature recognition results and fiber feature detection results with the cultural relics and store them in the cultural relic feature database, where each cultural relic is given a unique identifier;
[0012] Step 5. Automated classification and entry of cultural relics: Automatically classify and enter cultural relics based on machine learning algorithms, automatically identify the categories and attributes of cultural relics based on microscopic features and fiber feature information, and enter them into the database.
[0013] The present invention is further configured as follows: in the AID authentication cultural relic scanning step, firstly, the area for cultural relic scanning is determined, and the placement position and angle of the scanning device are set, and the scanning system observes and senses the cultural relic through the camera and the sensor, and automatically locates and tracks the cultural relic;
[0014] The scanning system uses target detection and real-time tracking technology to automatically locate and track cultural relics, identify their location and movement trajectory, and obtain images and related data;
[0015] In the process of locating and tracking cultural relics, images and related data of cultural relics are collected in real time;
[0016] The present invention is further configured as follows: in the AID authentication cultural relic scanning step, in the bonded area, images of cultural relics and artworks entering the bonded area are collected from the artwork source, and AID authentication data of the artworks, including microscopic features and fiber features, are extracted;
[0017] The extracted data is stored for record, and unique identity information for each artwork is established;
[0018] At the same time, an artwork authentication and filing database is established to record the identity information and authentication data of each artwork;
[0019] Establish a query system for customs and relevant departments to query and verify the authentication information of artworks. Customs and relevant departments monitor the authentication and filing of works in the bonded area and the status of their exit from the warehouse;
[0020] The present invention is further configured as follows: in step 2, collecting cultural relic microscopic feature image data, including cultural relics of different types and states and microscopic features of the cultural relics;
[0021] Convolutional neural network (CNN) is used to identify microscopic features;
[0022] Use annotated micro-feature image datasets to train deep learning models to enable them to identify micro-features of artifacts, including textures and particles;
[0023] Define the loss function to measure the gap between the model prediction results and the true label;
[0024] After training is completed, the model is evaluated using the validation set to check the performance and accuracy of the model, and the model parameters are adjusted based on the evaluation results;
[0025] The present invention is further configured as follows: the specific steps of performing microscopic feature recognition include:
[0026] Prepare an annotated microscopic feature image dataset, including cultural relic images and corresponding labels, where the labels represent the microscopic features, textures, and particles of the cultural relic;
[0027] Deploy the convolutional neural network (CNN) structure, including convolutional layer, pooling layer, and fully connected layer; use the convolutional layer to extract image features, the pooling layer to reduce feature dimensions, and the fully connected layer for classification;
[0028] Then the model is trained and the cross entropy loss function is defined to measure the gap between the model prediction results and the true labels: in, is the model's predicted probability of the cth category for the ith sample, y ic is the true label of the cth category of the ith sample, N is the number of samples, and C is the number of categories;
[0029] Use the annotated micro-feature image dataset to train the CNN model: Among them, θ represents the model parameters;
[0030] During the training process, the model is evaluated using the validation set to check the performance and accuracy of the model, and the model parameters are adjusted based on the evaluation results;
[0031] The present invention is further configured as follows: the multi-point fiber feature detection method includes:
[0032] Use edge detection to extract regions from cultural relic images and divide them into different regions;
[0033] Each region is pre-processed, and multi-scale analysis techniques are applied to each region to detect fiber characteristics;
[0034] Gaussian filters of different scales are used to detect fiber features of different scales in the image;
[0035] For each scale, the morphological operation is used to enhance the edges of fiber features;
[0036] For each region, the position, shape, and size information of the detected fiber features are recorded;
[0037] The present invention is further configured as follows: the multi-point fiber feature detection step comprises:
[0038] Use edge detection algorithms to process cultural relic images, extract edge information of cultural relics, extract and segment regions of cultural relic images, and pre-process each region, including denoising and smoothing;
[0039] Multiscale analysis is applied to each region to detect fiber features at different scales, and the image is convolved using Gaussian filters of different scales to obtain smoothed versions of the image at different scales: Where (x, y) is the pixel coordinate, σ is the standard deviation of the Gaussian function, and 2πσ 2 represents the normalized coefficient of the Gaussian function, and e is the base of the exponential function;
[0040] For each scale, morphological operations are used to enhance the edge features of the image;
[0041] On the enhanced image, Hough transform is used to detect fiber features, and the position, shape, and size information of the detected fiber features are recorded;
[0042] The present invention is further configured as follows: in the feature data storage step, the cultural relics feature database includes a cultural relics information table and a feature information table:
[0043] The cultural relic information table is used to store the basic information of each cultural relic, including the cultural relic number, name, category, and description;
[0044] The feature information table is used to store the microscopic feature recognition results and fiber feature detection results of each cultural relic, as well as other information associated with the cultural relic, including image path and feature vector;
[0045] Associating the microscopic feature recognition results and the fiber feature detection results with the cultural relics, and storing the feature information in a feature information table according to the cultural relic number;
[0046] The present invention is further configured as follows: the automated cultural relics classification and input method is:
[0047] The microscopic features and fiber feature information extracted from step 2 and step 3 are used as input data;
[0048] Prepare the labeled cultural relics categories and attributes as training data;
[0049] Perform feature engineering on microscopic features and fiber feature information, including feature selection and feature transformation processing, to extract the most discriminative features;
[0050] Build a learning model based on random forest and use the training data to train the machine learning model;
[0051] Divide the dataset into training and testing sets, and perform cross-validation to evaluate model performance;
[0052] Tune the model based on the results of cross-validation, including adjusting hyperparameters and feature engineering;
[0053] Use the test set to evaluate the tuned model, apply the trained model to unknown cultural relic data, and automatically identify the categories and attributes of cultural relics;
[0054] Enter the identification results into the database and store them in association with the cultural relics information;
[0055] The present invention is further configured as follows: the automated cultural relics classification and entry step comprises:
[0056] The microscopic features and fiber feature information extracted from step 2 and step 3 are used as input data, denoted as X, and the annotated cultural relics categories and attributes are prepared as training data, denoted as Y. Feature engineering is performed on the microscopic features and fiber feature information, including feature selection and feature transformation processing, to extract the most discriminative features;
[0057] Suppose K features are selected in the feature engineering process, denoted as X', X'∈R N×K , which is the feature data matrix after feature engineering, where N is the number of samples;
[0058] A learning model is constructed based on the random forest algorithm. Let the number of trees in the random forest model be T, represented by RF T ;
[0059] Use training data to train machine learning models and optimize model parameters;
[0060] Divide the data set into a training set and a validation set, perform K-fold cross-validation to evaluate the model performance, and use the cross-validation results to tune the model. Suppose the number of cross-validation folds used is K. cv ; Use the test set to evaluate the tuned model and use the trained model RF T Applied to unknown cultural relic data to automatically identify the categories and attributes of cultural relics;
[0061] The identification results are entered into the database and stored in association with the cultural relics information.
[0062] Compared with the prior art, the present invention has the following beneficial effects:
[0063] The present invention uses AID authentication technology to automatically locate and track cultural relics and collect relevant data through a scanning system to establish unique identity information for each cultural relic, which not only improves the accuracy and authority of cultural relic information, but also provides a basis for the tracing and management of cultural relics. It uses deep learning technology to identify the microscopic features of cultural relics, and through training models and optimizing parameters, it realizes automatic extraction and classification of microscopic features, effectively reducing the burden of manual identification, while improving the accuracy and efficiency of identification.
[0064] The present invention analyzes the fiber characteristics of cultural relics through multi-point fiber characteristic detection, and effectively captures the fiber characteristic information of different positions of cultural relics through edge detection and multi-scale analysis;
[0065] The present invention automatically classifies and enters cultural relics based on a machine learning algorithm, thereby improving the efficiency and accuracy of registration and entry; and realizes automatic recognition and entry of cultural relic categories and attributes through feature engineering and model training. BRIEF DESCRIPTION OF THE DRAWINGS
[0066] Figure 1 The present invention is a flow chart of the movable cultural relics registration and entry method based on AID authentication technology. DETAILED DESCRIPTION
[0067] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.
[0068] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.
[0069] The present invention is further described in detail below in conjunction with the accompanying drawings:
[0070] Example
[0071] See also Figure 1 The present invention provides a movable cultural relic registration method based on AID authentication technology, comprising:
[0072] Step 1. AID authentication and cultural relic scanning: Place the cultural relic to be registered in the scanning area and scan it through the scanning system; the scanning system includes a camera, a sensor, and a processing chip; the cultural relic scanning uses target detection and real-time tracking technology to automatically locate and track the cultural relic, and collect images and related data of the cultural relic; and establish an artwork ID card, a database and query system for the certification and filing of cultural relics and cultural works in the customs bonded area, and monitor the tax payment status of the works in the bonded area;
[0073] In the AID authentication cultural relic scanning step, the area to be scanned is first determined, and the placement and angle of the scanning equipment are set. The scanning system observes and senses the cultural relic through cameras and sensors, and automatically locates and tracks the cultural relic.
[0074] The scanning system uses target detection and real-time tracking technology to automatically locate and track cultural relics, identify their location and movement trajectory, and obtain images and related data;
[0075] In the process of locating and tracking cultural relics, images and related data of cultural relics are collected in real time;
[0076] In the AID authentication of cultural relics scanning step, in the bonded area, images of cultural relics and artworks entering the bonded area are collected from the source of the artworks, and the AID authentication data of the artworks, including microscopic features and fiber features, are extracted;
[0077] The extracted data is stored for record, and unique identity information for each artwork is established;
[0078] At the same time, an artwork authentication and filing database is established to record the identity information and authentication data of each artwork;
[0079] Establish a query system for customs and relevant departments to query and verify the authentication information of artworks. Customs and relevant departments monitor the authentication and filing of works in the bonded area and the status of their exit from the warehouse;
[0080] Through monitoring, customs can track the tax payment status of works and avoid tax evasion in bonded areas;
[0081] Step 2. Micro-feature recognition: Based on deep learning technology, a micro-feature recognition model is trained to identify the micro-features of cultural relics. The scanning system analyzes the scanned image data and extracts the micro-features of cultural relics, including texture and particles.
[0082] In step 2, the image data of microscopic features of cultural relics are collected, including cultural relics of different types and states and microscopic features of cultural relics;
[0083] Convolutional neural network (CNN) is used to identify microscopic features;
[0084] Use annotated micro-feature image datasets to train deep learning models to enable them to identify micro-features of artifacts, including textures and particles;
[0085] Define the loss function to measure the gap between the model prediction results and the true label;
[0086] After training is completed, the model is evaluated using the validation set to check the performance and accuracy of the model, and the model parameters are adjusted based on the evaluation results;
[0087] The specific steps of microscopic feature recognition include: preparing annotated microscopic feature image datasets, including cultural relic images and corresponding labels, where the labels represent the microscopic features, textures and particles of the cultural relic;
[0088] Deploy the convolutional neural network (CNN) structure, including convolutional layer, pooling layer, and fully connected layer; use the convolutional layer to extract image features, the pooling layer to reduce feature dimensions, and the fully connected layer for classification;
[0089] Then the model is trained and the cross entropy loss function is defined to measure the gap between the model prediction results and the true labels: in, is the model's predicted probability of the cth category for the ith sample, y ic is the true label of the cth category of the ith sample, N is the number of samples, and C is the number of categories;
[0090] Use the annotated micro-feature image dataset to train the CNN model: Among them, θ represents the model parameters;
[0091] During the training process, the model is evaluated using the validation set to check the performance and accuracy of the model, and the model parameters are adjusted based on the evaluation results;
[0092] By performing microscopic feature recognition, a convolutional neural network model with high accuracy and performance is trained;
[0093] Step 3. Multi-point fiber feature detection, deploying the fiber feature detection method of regional extraction and multi-scale analysis, detecting the fiber features at different positions of the cultural relics, and recording the fiber feature information of each part of the cultural relics;
[0094] Multi-point fiber feature detection methods include:
[0095] Use edge detection to extract regions from cultural relic images and divide them into different regions;
[0096] Each region is pre-processed, and multi-scale analysis techniques are applied to each region to detect fiber characteristics;
[0097] Gaussian filters of different scales are used to detect fiber features of different scales in the image;
[0098] For each scale, the morphological operation is used to enhance the edges of fiber features;
[0099] For each region, the position, shape, and size information of the detected fiber features are recorded;
[0100] The multi-point fiber feature detection steps include:
[0101] Use edge detection algorithms to process cultural relic images, extract edge information of cultural relics, extract and segment regions of cultural relic images, and pre-process each region, including denoising and smoothing;
[0102] Multiscale analysis is applied to each region to detect fiber features at different scales, and the image is convolved using Gaussian filters of different scales to obtain smoothed versions of the image at different scales: Where (x, y) is the pixel coordinate, σ is the standard deviation of the Gaussian function, and 2πσ 2 represents the normalized coefficient of the Gaussian function, and e is the base of the exponential function;
[0103] For each scale, morphological operations are used to enhance the edge features of the image;
[0104] On the enhanced image, Hough transform is used to detect fiber features, and the position, shape, and size information of the detected fiber features are recorded;
[0105] Through multi-point fiber feature detection of cultural relic images, the location and information of fiber features are extracted;
[0106] Step 4. Feature data storage: associate the microscopic feature recognition results and fiber feature detection results with the cultural relics and store them in the cultural relic feature database. Each cultural relic is given a unique identifier for subsequent management and retrieval;
[0107] In the feature data storage step, the cultural relic feature database includes a cultural relic information table and a feature information table:
[0108] The cultural relic information table is used to store the basic information of each cultural relic, including the cultural relic number, name, category, and description;
[0109] The feature information table is used to store the microscopic feature recognition results and fiber feature detection results of each cultural relic, as well as other information associated with the cultural relic, including image path and feature vector;
[0110] Associating the microscopic feature recognition results and the fiber feature detection results with the cultural relics, and storing the feature information in a feature information table according to the cultural relic number;
[0111] Step 5. Automated classification and entry of cultural relics: Automatically classify and enter cultural relics based on machine learning algorithms, automatically identify the categories and attributes of cultural relics based on microscopic features and fiber feature information, and enter them into the database;
[0112] The automated cultural relics classification and entry method is as follows:
[0113] The microscopic features and fiber feature information extracted from step 2 and step 3 are used as input data;
[0114] Prepare the labeled cultural relics categories and attributes as training data;
[0115] Perform feature engineering on microscopic features and fiber feature information, including feature selection and feature transformation processing, to extract the most discriminative features;
[0116] Build a learning model based on random forest and use the training data to train the machine learning model;
[0117] Divide the dataset into training and testing sets, and perform cross-validation to evaluate model performance;
[0118] Tune the model based on the results of cross-validation, including adjusting hyperparameters and feature engineering;
[0119] Use the test set to evaluate the tuned model, apply the trained model to unknown cultural relic data, and automatically identify the categories and attributes of cultural relics;
[0120] Enter the identification results into the database and store them in association with the cultural relics information;
[0121] The steps for automated cultural relics classification and entry include:
[0122] The microscopic features and fiber feature information extracted from step 2 and step 3 are used as input data, denoted as X, and the annotated cultural relics categories and attributes are prepared as training data, denoted as Y. Feature engineering is performed on the microscopic features and fiber feature information, including feature selection and feature transformation processing, to extract the most discriminative features;
[0123] Suppose K features are selected in the feature engineering process, denoted as X', X'∈R N×K , which is the feature data matrix after feature engineering, where N is the number of samples;
[0124] A learning model is constructed based on the random forest algorithm. Let the number of trees in the random forest model be T, represented by RF T ;
[0125] Use training data to train machine learning models and optimize model parameters;
[0126] Divide the data set into a training set and a validation set, perform K-fold cross-validation to evaluate the model performance, and use the cross-validation results to tune the model. Suppose the number of cross-validation folds used is K. cv ; Use the test set to evaluate the tuned model and use the trained model RF T Applied to unknown cultural relic data to automatically identify the categories and attributes of cultural relics;
[0127] Enter the identification results into the database and store them in association with the cultural relics information;
[0128] The movable cultural relics registration and entry method based on AID authentication technology can accurately and efficiently complete the registration and entry of cultural relics, and provide convenient data management and query functions. Micro-feature recognition and multi-point fiber feature detection can further improve the accuracy and intelligence level of registration and entry.
[0129] The present invention uses AID authentication technology to automatically locate and track cultural relics and collect relevant data through a scanning system to establish unique identity information for each cultural relic, which not only improves the accuracy and authority of cultural relic information, but also provides a basis for the tracing and management of cultural relics. It uses deep learning technology to identify the microscopic features of cultural relics, and through training models and optimizing parameters, it realizes automatic extraction and classification of microscopic features, effectively reducing the burden of manual identification, while improving the accuracy and efficiency of identification.
[0130] The present invention analyzes the fiber characteristics of cultural relics through multi-point fiber characteristic detection, and effectively captures the fiber characteristic information of different positions of cultural relics through edge detection and multi-scale analysis;
[0131] The present invention automatically classifies and enters cultural relics based on machine learning algorithms, thereby improving the efficiency and accuracy of registration and entry; and automatically identifies and enters cultural relics categories and attributes through feature engineering and model training;
[0132] The AID authentication technology used in this application plays a core role in ensuring the "original entry and exit" of cultural relics and artworks. It establishes a unique and authoritative identity card for artworks. Starting from the source of the artworks in the bonded area, high-definition image collection, AID authentication data extraction, and data storage and filing are carried out for the cultural relics and artworks entering the bonded area. A unique and authoritative identity card information is established for each artwork to ensure the traceability of each work entering the bonded area; at the same time, a customs bonded area cultural relics and art certification filing database and query system are established to ensure the orderly circulation of cultural relics and artworks in the bonded area. The relevant customs departments can monitor the tax payment status of the works in the bonded area by certifying and filing the works in the bonded area and verifying the outbound status, and try to avoid tax evasion in the bonded area as much as possible.
[0133] The application of AID authentication technology ensures the authenticity and utilization value of artworks, increases the trust and reliability of exchanges, and attracts more international participants in art exhibitions and exchange activities, promotes the exchange and integration of cultural diversity, and enhances the exchange and market competitiveness of artworks in the comprehensive bonded zone. By using AID authentication technology, the transparency of the artwork transaction process is achieved, injecting new vitality into the artwork market.
[0134] In summary, it can be seen that the present invention introduces automation technology, utilizes scanning systems and deep learning models to realize automatic positioning, tracking and feature extraction of cultural relics, significantly improves the efficiency and accuracy of registration and entry, and adopts a multi-point fiber feature detection method, combined with edge detection and multi-scale analysis technology, which can more comprehensively analyze the characteristics of cultural relics and improve the comprehensiveness and accuracy of information acquisition. At the same time, a cultural relic feature database and identity card information are established to achieve unique identification and authoritative authentication of each cultural relic, improve the traceability of cultural relic information and the convenience of management, and have higher accuracy, comprehensiveness and convenience.
[0135] The above contents are only for explaining the technical idea of the present invention and cannot be used to limit the protection scope of the present invention. Any changes made on the basis of the technical solution in accordance with the technical idea proposed by the present invention shall fall within the protection scope of the claims of the present invention.
Claims
1. A movable cultural relic registration method based on AID authentication technology, characterized in that: include: Step 1. AID authentication cultural relic scanning: Place the cultural relic to be registered in the scanning area and scan it through the scanning system; the scanning system includes a camera, a sensor, and a processing chip; the cultural relic scanning uses target detection and real-time tracking technology to automatically locate and track the cultural relic, and collect images and related data of the cultural relic; We will also establish an artwork ID card, a certification and filing database and query system for cultural relics and art works in the customs bonded area, and monitor the tax payment status of works in the bonded area; Step 2. Microscopic feature recognition: Based on deep learning technology, a microscopic feature recognition model is trained to recognize the microscopic features of cultural relics. The scanning system analyzes the scanned image data and extracts the microscopic features of cultural relics, including texture and particles. Step 3. Multi-point fiber feature detection, deploying the fiber feature detection method of regional extraction and multi-scale analysis, detecting the fiber features at different positions of the cultural relics, and recording the fiber feature information of each part of the cultural relics; Step 4. Feature data storage: associate the microscopic feature recognition results and fiber feature detection results with the cultural relics and store them in the cultural relic feature database, where each cultural relic is given a unique identifier; Step 5. Automated classification and entry of cultural relics: Automatically classify and enter cultural relics based on machine learning algorithms, automatically identify the categories and attributes of cultural relics based on microscopic features and fiber feature information, and enter them into the database; Multi-point fiber feature detection methods include: Use edge detection to extract regions from cultural relic images and divide them into different regions; Each region is pre-processed, and multi-scale analysis techniques are applied to each region to detect fiber characteristics; Gaussian filters of different scales are used to detect fiber features of different scales in the image; For each scale, the morphological operation is used to enhance the edges of fiber features; For each region, the position, shape, and size information of the detected fiber features are recorded. In the feature data storage step, the cultural relic feature database includes a cultural relic information table and a feature information table: The cultural relic information table is used to store the basic information of each cultural relic, including the cultural relic number, name, category, and description; The feature information table is used to store the microscopic feature recognition results and fiber feature detection results of each cultural relic, as well as other information associated with the cultural relic, including image path and feature vector; Associating the microscopic feature recognition results and the fiber feature detection results with the cultural relics, and storing the feature information in a feature information table according to the cultural relic number; The automated cultural relics classification and entry method is as follows: The microscopic features and fiber feature information extracted from step 2 and step 3 are used as input data; Prepare the labeled cultural relics categories and attributes as training data; Perform feature engineering on microscopic features and fiber feature information, including feature selection and feature transformation processing, to extract the most discriminative features; Build a learning model based on random forest and use the training data to train the machine learning model; Divide the dataset into training and testing sets, and perform cross-validation to evaluate model performance; Tune the model based on the results of cross-validation, including adjusting hyperparameters and feature engineering; Use the test set to evaluate the tuned model, apply the trained model to unknown cultural relic data, and automatically identify the categories and attributes of cultural relics; The identification results are entered into the database and stored in association with the cultural relics information.
2. According to claim 1, a movable cultural relic registration method based on AID authentication technology is characterized in that: In the AID authentication cultural relic scanning step, the area to be scanned is first determined, and the placement and angle of the scanning equipment are set. The scanning system observes and senses the cultural relic through cameras and sensors, and automatically locates and tracks the cultural relic. The scanning system uses target detection and real-time tracking technology to automatically locate and track cultural relics, identify their location and movement trajectory, and obtain images and related data; During the process of locating and tracking cultural relics, images and related data of the cultural relics are collected in real time.
3. According to claim 2, a movable cultural relic registration method based on AID authentication technology is characterized in that: In the AID authentication of cultural relics scanning step, in the bonded area, images of cultural relics and artworks entering the bonded area are collected from the source of the artworks, and the AID authentication data of the artworks, including microscopic features and fiber features, are extracted; The extracted data is stored for record, and unique identity information for each artwork is established; At the same time, an artwork authentication and filing database is established to record the identity information and authentication data of each artwork; A query system is set up for customs and relevant departments to query and verify the authentication information of artworks. The relevant customs departments monitor the certification filing and verification of the works in the bonded area.
4. According to claim 3, a movable cultural relic registration method based on AID authentication technology is characterized in that: In step 2, the image data of microscopic features of cultural relics are collected, including cultural relics of different types and states and microscopic features of cultural relics; Convolutional neural network (CNN) is used to identify microscopic features; Use annotated micro-feature image datasets to train deep learning models to enable them to identify micro-features of artifacts, including textures and particles; Define the loss function to measure the gap between the model prediction results and the true label; After training is completed, the model is evaluated using the validation set to check the performance and accuracy of the model, and the model parameters are adjusted based on the evaluation results.
5. The movable cultural relic registration method based on AID authentication technology according to claim 4 is characterized in that: The specific steps for microscopic feature recognition include: Prepare an annotated microscopic feature image dataset, including cultural relic images and corresponding labels, where the labels represent the microscopic features, textures, and particles of the cultural relic; Deploy the convolutional neural network (CNN) structure, including convolutional layer, pooling layer, and fully connected layer; use the convolutional layer to extract image features, the pooling layer to reduce feature dimensions, and the fully connected layer for classification; Then the model is trained and the cross entropy loss function is defined to measure the gap between the model prediction results and the true labels: in, is the model's predicted probability of the cth category for the ith sample, y ic is the true label of the cth category of the ith sample, N is the number of samples, and C is the number of categories; Use the annotated micro-feature image dataset to train the CNN model: Among them, θ represents the model parameters; During the training process, the validation set is used to evaluate the model, check the performance and accuracy of the model, and adjust the model parameters based on the evaluation results.
6. The movable cultural relic registration method based on AID authentication technology according to claim 5 is characterized in that: The multi-point fiber feature detection steps include: Use edge detection algorithms to process cultural relic images, extract edge information of cultural relics, extract and segment regions of cultural relic images, and pre-process each region, including denoising and smoothing; Multiscale analysis is applied to each region to detect fiber features at different scales, and the image is convolved using Gaussian filters of different scales to obtain smoothed versions of the image at different scales: Where (x, y) is the pixel coordinate, σ is the standard deviation of the Gaussian function, and 2πσ 2 represents the normalized coefficient of the Gaussian function, and e is the base of the exponential function; For each scale, morphological operations are used to enhance the edge features of the image; On the enhanced image, Hough transform is used to detect fiber features, and the position, shape, and size information of the detected fiber features are recorded.
7. The movable cultural relic registration method based on AID authentication technology according to claim 6 is characterized in that: The steps for automated cultural relics classification and entry include: The microscopic features and fiber feature information extracted from step 2 and step 3 are used as input data, denoted as X, and the annotated cultural relics categories and attributes are prepared as training data, denoted as Y. Feature engineering is performed on the microscopic features and fiber feature information, including feature selection and feature transformation processing, to extract the most discriminative features; Suppose K features are selected in the feature engineering process, denoted as X', X'∈R N×K , which is the feature data matrix after feature engineering, where N is the number of samples; A learning model is constructed based on the random forest algorithm. Let the number of trees in the random forest model be T, represented by RF T ; Use training data to train machine learning models and optimize model parameters; Divide the data set into a training set and a validation set, perform K-fold cross-validation to evaluate the model performance, and use the cross-validation results to tune the model. Suppose the number of cross-validation folds used is K. cv ; Use the test set to evaluate the tuned model and use the trained model RF T Applied to unknown cultural relic data to automatically identify the categories and attributes of cultural relics; The identification results are entered into the database and stored in association with the cultural relics information.
