Method for constructing and interacting with digital twin of qin dynasty bamboo slips based on multi-modal data

Abstract

The application discloses a kind of Qin Jian digital twin based on multi-modal data construction and interaction method, comprising the following steps: S1, obtain the multi-modal data of Qin Jian and carry out standardization preprocessing, obtain the standardization multi-modal data set that can be fused;S2, based on the double-state twin modeling method driven by degradation inversion, the micro-region division and degradation parameter construction are carried out to standardization multi-modal data, inversion Qin Jian morphological evolution path, construct present situation twin layer and original appearance inversion layer, form the double-state digital twin basic model of single Qin Jian, the present application relates to cultural heritage digitization protection technical field.This kind of Qin Jian digital twin based on multi-modal data construction and interaction method, through the standardization preprocessing and fusion of multi-modal data, can accurately reconstruct the present situation twin layer and original appearance inversion layer of Qin Jian, realize the scientific deduction to Qin Jian morphological evolution path, ensure the high accuracy and authenticity of digital twin.

Description

Technical Field

[0001] This invention relates to the field of digital preservation of cultural heritage, and more specifically, to a method for constructing and interacting with a digital twin of Qin bamboo slips based on multimodal data. Background Technology

[0002] With the continuous development of digital technology, digital twin technology is increasingly being used in the field of cultural relic preservation to preserve and restore ancient artifacts. Traditional methods of cultural relic restoration mainly rely on manual operation, which is subject to certain errors and irreversibility. Therefore, there is an urgent need for a more scientific, precise, and sustainable restoration technology. Qin bamboo slips, as an important legacy of ancient Chinese civilization, possess extremely high historical, cultural, and artistic value. However, after thousands of years of weathering the elements, Qin bamboo slips often face serious damage problems, such as fading of the characters, surface corrosion, and the expansion of cracks. These problems make the digital restoration and preservation of Qin bamboo slips an important research direction in the field of cultural relic preservation.

[0003] Against this backdrop, the construction of digital twins of Qin bamboo slips based on multimodal data has gradually become a research hotspot. By integrating multiple technologies such as high-resolution scanning, laser point cloud, microscopic imaging, and spectral analysis, various physical features of Qin bamboo slips can be accurately captured, and their original appearance can be reconstructed through digital modeling, thus providing new solutions for cultural relic restoration and digital preservation. However, most existing technologies focus on static reconstruction and do not deeply consider user interaction and dispute resolution mechanisms, making it difficult to meet the needs of the cultural relic preservation field for high precision, dynamic feedback, and sustainable updates in practical applications. Summary of the Invention

[0004] The purpose of this invention is to provide a method for constructing and interacting with a digital twin of Qin bamboo slips based on multimodal data. This method addresses the problem that most existing technologies focus on static reconstruction and do not deeply consider user interaction and dispute resolution mechanisms, making it difficult to meet the needs of cultural relic protection for high precision, dynamic feedback, and sustainable updates in practical applications.

[0005] This invention achieves the above objective through the following technical solution: a method for constructing and interacting with a Qin Dynasty bamboo slip digital twin based on multimodal data, comprising the following steps:

[0006] S1. Obtain multimodal data from Qin bamboo slips and perform standardized preprocessing to obtain a set of standardized multimodal data that can be fused.

[0007] S2. Based on the degradation inversion-driven dual-state twin modeling method, the standardized multimodal data is divided into micro-regions and degradation parameters are constructed to invert the morphological evolution path of the Qin bamboo slips, construct the current twin layer and the original appearance inversion layer, and form a dual-state digital twin basic model of a single Qin bamboo slip.

[0008] S3. Adopting a self-evident splicing modeling method based on three-chain coupling, integrating the collaborative calculation results of three types of evidence chains, establishing splicing relationships between multiple Qin bamboo slips, writing splicing-related evidence information into the dual-state digital twin basic model, and forming a combined digital twin of Qin bamboo slips;

[0009] S4. Based on the dispute focus-driven local evidence reconstruction interaction mechanism, the system identifies the dispute type and dispute area of ​​the Qin bamboo slip digital twin, extracts the core evidence combination to perform targeted reconstruction and local re-rendering of the dispute area, outputs a multi-dimensional discrimination view, and dynamically updates the evidence weight and relationship confidence of the Qin bamboo slip digital twin according to the user's interaction operation, forming a closed-loop operation mechanism.

[0010] Furthermore, the acquisition of multimodal data from the Qin bamboo slips in step S1 includes the following steps:

[0011] One or more of the following methods—high-resolution scanning, 3D laser point cloud acquisition, microscopic imaging, and spectral analysis—were used to collect Qin bamboo slips in all dimensions.

[0012] The multimodal data covers at least one of the following: the handwriting, surface morphology, material texture, cracks, edge contours, and contamination / occlusion information of the Qin bamboo slips.

[0013] Furthermore, the standardization preprocessing described in step S1 includes the following steps:

[0014] Targeted processing operations are performed on the features and noise characteristics of different types of multimodal data, including one or more of the following processing methods for handwriting information: deblurring, grayscale enhancement, and stroke outline extraction.

[0015] Perform one or more of the following processing on the surface morphology and material texture information: point cloud denoising, coordinate registration, and texture mapping;

[0016] One or more of the following processes are applied to the crack and edge contour information: feature point extraction, contour fitting, and fracture surface feature quantization.

[0017] The pollution-occluded information is processed by one or more of the following methods: masking, background separation, and occlusion area localization.

[0018] Furthermore, the micro-region division described in step S2 includes the following steps:

[0019] Based on the physical dimensions and characteristic distribution patterns of Qin bamboo slips, the physical space of Qin bamboo slips is divided into several local micro-regions using the grid division method, so as to achieve a refined regionalized representation of the morphological characteristics of Qin bamboo slips.

[0020] The degradation parameter construction involves constructing a set of degradation state parameters for each local micro-region, combining the physical laws of Qin bamboo slip damage and degradation with multimodal data characteristics. These parameters include one or more of the following indicators: handwriting fading, surface corrosion, crack propagation, and pollution occlusion. Each degradation state parameter is quantitatively calculated based on standardized multimodal data.

[0021] Furthermore, the inversion of the evolution path of the Qin bamboo slips described in step S2 includes the following steps:

[0022] Based on standardized multimodal data and degradation state parameter sets of various local micro-regions, combined with the historical evolution factors of Qin bamboo slips material aging and environmental erosion, an evolution function in the time dimension is constructed to inversely deduce the evolution path of Qin bamboo slips from their original writing state to their current damaged state, so as to realize the traceability and quantification of the damage process.

[0023] Furthermore, the construction of the current state twin layer and the original state inversion layer in step S2 includes the following steps:

[0024] Based on the evolution path of Qin bamboo slips and standardized multimodal data, and by integrating the physical characteristics of each local micro-region, a current twin layer is constructed through three-dimensional modeling and texture mapping to accurately restore the current preservation state of Qin bamboo slips.

[0025] By performing reverse inference based on the morphological evolution path, the influence of degradation factors on the original form of Qin bamboo slips is eliminated, the missing and damaged original features are estimated and reconstructed, and the original appearance inversion layer is constructed to realize the scientific estimation and visualization of the original form of Qin bamboo slips.

[0026] Furthermore, the three types of evidence chains mentioned in step S3 are physical codification, written evidence, and semantic evidence chains, including the following steps:

[0027] Feature sets for each evidence chain are constructed by extracting corresponding features from the bi-state digital twin basic model, and normalized matching values ​​of features within each evidence chain are obtained by quantitative calculation.

[0028] The feature weights within each evidence chain are determined by combining one or more of the following methods: analytic hierarchy process, coefficient of variation method, expert scoring method, and entropy weight method. The comprehensive matching degree of each evidence chain between any two Qin bamboo slips is calculated using the weighted summation method.

[0029] Furthermore, the establishment of the arrangement relationship between multiple Qin bamboo slips in step S3 includes the following steps:

[0030] The domain adaptation method is used to determine the weight coefficients of the three chains of evidence: physical binding, writing, and semantics. Different weight allocation ratios are adapted according to the type of Qin bamboo slips. The multi-source evidence fusion method is used to calculate the confidence of the three-chain coupling comprehensive splicing of any two Qin bamboo slips.

[0031] The percentile method was used to determine the confidence threshold for the assembly. Two Qin bamboo slips whose comprehensive assembly confidence reached the threshold were judged to have a valid assembly relationship. Based on all valid assembly relationships, the topology of the assembly of multiple Qin bamboo slips was established by the topology modeling method.

[0032] The information of at least one of the following—the source of evidence, the strength of evidence, the type of conflict, the candidate substitution relationship, and the historical revision trajectory—is written into the dual-state digital twin basic model to form a combined digital twin of Qin bamboo slips.

[0033] Furthermore, the identification of dispute type and dispute area in step S4 includes the following steps:

[0034] The system receives user interaction queries and dispute feedback commands through a human-computer interaction interface. It uses feature extraction and coordinate positioning technology to extract the spatial location information of the dispute focus and the dispute type from the commands. The dispute type includes one or more of the following: handwriting interpretation, broken edge matching, splicing relationship, and original appearance restoration result dispute.

[0035] The process of extracting core evidence sets involves extracting relevant evidence sets from the underlying standardized multimodal data based on the type of dispute, screening core evidence by calculating evidence discrimination, determining the evidence discrimination threshold using the K-means clustering method, and using evidence that reaches the threshold as the core evidence set.

[0036] Furthermore, the directional reconstruction and local re-rendering described in step S4 include the following steps:

[0037] Based on the core evidence set, corresponding algorithms are used to perform targeted reconstruction of the disputed area for different types of disputes. For handwriting interpretation disputes, image segmentation and stroke enhancement algorithms are used; for broken edge matching disputes, geometric fitting and contour comparison algorithms are used; for splicing relationship disputes, evidence weight recalculation and multi-source fusion algorithms are used; and for original appearance restoration result disputes, credibility quantification and hierarchical reconstruction algorithms are used.

[0038] By combining visualization rendering technology, the disputed area after targeted reconstruction is locally re-rendered, and a multi-dimensional discrimination view corresponding to the dispute type is output.

[0039] The dynamic updating of evidence weights and relationship confidence scores receives user confirmation, denial, and revision interaction results for the discriminative view, and formulates differentiated update strategies.

[0040] For negative operations, the evidence weights are updated according to the weight attenuation coefficient based on the level of dispute and the assembled confidence is recalculated. For revision operations, the core evidence set is supplemented and corrected and the calculation is updated in combination with the new weights specified by the user.

[0041] The interaction records and updated parameters are synchronized to the Qin bamboo slips digital twin to form a closed-loop operation mechanism.

[0042] The beneficial effects of this invention are as follows:

[0043] 1. Through standardized preprocessing and fusion of multimodal data, the current twin layer and original appearance inversion layer of Qin bamboo slips can be accurately reconstructed, enabling scientific deduction of the evolution path of Qin bamboo slips and ensuring the high precision and authenticity of the digital twin.

[0044] 2. Through a local evidence reconstruction interaction mechanism driven by the focus of the dispute, the system enables users to dynamically identify and respond to the dispute types and disputed areas of the Qin bamboo slips digital twin. Through targeted reconstruction and local re-rendering, it provides a multi-dimensional discriminative view and can dynamically update the evidence weight and relationship confidence based on user interaction, thus forming a closed-loop operation mechanism.

[0045] 3. A three-chain coupled self-evident splicing modeling method is adopted. Through the collaborative calculation of multiple evidence chains, physical splicing, writing and semantic evidence chains, the splicing relationship between multiple Qin bamboo slips is established, which effectively improves the accuracy and reliability of the splicing relationship between different Qin bamboo slips.

[0046] 4. By accurately identifying the types of disputes, this method can perform targeted reconstruction of disputed areas such as handwriting interpretation, broken edge matching, and splicing relationships. It also employs various repair algorithms for local re-rendering, effectively improving the interactive experience and user satisfaction of the Qin bamboo slips digital twin.

[0047] 5. By introducing digital twin technology and multimodal data processing, this invention provides a sustainable and accurate digital solution for the protection and restoration of cultural relics, which helps to improve the efficiency and accuracy of cultural relic protection work, especially for the protection and restoration of precious cultural relics such as Qin bamboo slips. Attached Figure Description

[0048] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0049] Figure 1 This is an overall flowchart of the method for constructing and interacting with a Qin Dynasty bamboo slip digital twin according to the present invention;

[0050] Figure 2 This is a flowchart illustrating the construction process of the dual-state digital twin basic model of the present invention.

[0051] Figure 3 This is a flowchart illustrating the construction process of the Qin Dynasty bamboo slips combined digital twin of the present invention. Detailed Implementation

[0052] The present application will now be described in further detail with reference to the accompanying drawings. It should be noted that the following specific embodiments are only used to further illustrate the present application and should not be construed as limiting the scope of protection of the present application. Those skilled in the art can make some non-essential improvements and adjustments to the present application based on the above application content.

[0053] Example 1:

[0054] Please see Figure 1-3 This invention provides a technical solution: a method for constructing and interacting with a Qin Dynasty bamboo slip digital twin based on multimodal data, comprising the following steps:

[0055] S1. Obtain multimodal data from Qin bamboo slips and perform standardized preprocessing;

[0056] Multimodal data refers to data containing various types of information. For Qin bamboo slips, this may include image data, such as photographs or scanned images of the slips' appearance; textual data, including the written content on the slips; and material analysis data, such as information on the material composition of the slips obtained through chemical analysis. Standardized preprocessing involves applying uniform formatting, scaling, noise removal, and data augmentation to the acquired multimodal data to ensure consistency and standardization, facilitating subsequent analysis and processing. For example, images with different resolutions are uniformly adjusted to the same resolution, and character recognition and formatting are performed on the textual data.

[0057] S2. Based on the degradation inversion-driven dual-state twin modeling method, the preprocessed multimodal data is divided into micro-regions and degradation parameters are constructed to invert the morphological evolution path of the Qin bamboo slips. The current twin layer representing the current true preservation state and the original appearance inversion layer representing the original morphological estimation result are constructed to form a dual-state digital twin basic model of a single Qin bamboo slip.

[0058] Among them, the degradation inversion-driven dual-state twin modeling method involves degradation inversion, which, based on the current state of the Qin bamboo slips, uses certain algorithms and models to reverse-engineer the morphological changes it has undergone from its original state to its current state, and analyzes the various factors and parameters that lead to these changes. Dual-state twin modeling constructs two different digital twin models: a current state twin layer and an original state inversion layer. Micro-region division and degradation parameter construction are also crucial. Micro-region division involves dividing the surface or the entire Qin bamboo slip into multiple small regions to analyze the characteristics and changes of each region in greater detail. Degradation parameter construction determines parameters that can describe the degree of degradation, such as wear, corrosion, and discoloration, occurring in each micro-region of the Qin bamboo slip. The system includes several parameters: parameters used to quantify the degradation process; the evolution path of the Qin bamboo slips' morphology, derived through degradation inversion to show the entire process and trajectory of the Qin bamboo slips' gradual change from their original form to their current preserved form; the current state twin layer, a digital model constructed using preprocessed multimodal data, accurately representing the current true preservation state of the Qin bamboo slips and reflecting their current appearance, structure, and other characteristics; the original form inversion layer, a digital model constructed based on the degradation inversion results, used to represent the inference results of the original form of the Qin bamboo slips, attempting to restore the appearance of the Qin bamboo slips when they were first made; and the dual-state digital twin basic model of a single Qin bamboo slip, composed of the current state twin layer and the original form inversion layer, comprehensively describing the morphological characteristics of a single Qin bamboo slip at different points in time.

[0059] S3. A self-evident splicing modeling method based on three-chain coupling is adopted. The collaborative calculation results of three types of evidence chains, namely physical splicing, writing, and semantics, are integrated to establish the splicing relationship between multiple Qin bamboo slips. The evidence source, evidence strength, conflict type, candidate substitution relationship, and historical revision trajectory corresponding to the splicing are written into the dual-state digital twin basic model to form a combined digital twin of Qin bamboo slips.

[0060] The three-chain coupling organically combines three types of evidence chains: physical linking, writing, and semantics. The physical linking evidence chain involves the actual physical connection methods between Qin bamboo slips, such as rope marks; the writing evidence chain focuses on the writing style, font, stroke order, and other characteristics of the characters on the Qin bamboo slips, as well as the logical relationships between the characters; the semantic evidence chain analyzes the semantic connections and logic based on the meaning expressed by the characters on the Qin bamboo slips. Self-proving splicing modeling automatically verifies and determines the splicing relationships between multiple Qin bamboo slips through the collaborative calculation of these three types of evidence chains, constructing a splicing model that can self-prove its rationality. The splicing relationship refers to the mutual relationships between multiple Qin bamboo slips in terms of physical connection and textual content coherence, such as which Qin bamboo slips were originally linked together, and whether they have a sequential order or mutual correspondence in content. The source of evidence refers to the origin of various types of evidence used to determine the splicing relationship of Qin bamboo slips, such as physical linking evidence obtained by observing rope marks on the Qin bamboo slips, or evidence obtained by analyzing textual style and... Semantic evidence includes written and semantic evidence; evidence strength, which measures the reliability and importance of various types of evidence in determining the composition relationship. The higher the evidence strength, the greater its role in determining the composition relationship; conflict type, which classifies contradictory situations that may arise when integrating different chains of evidence to determine the composition relationship, such as the conflict between physical linking evidence and written evidence; candidate alternatives, which propose alternatives to the current composition relationship when there are conflicts or uncertainties; historical revision trajectory, which records the revisions and changes in the composition relationship of Qin bamboo slips during the research process as new evidence emerges or analytical methods are improved; and a digital twin of Qin bamboo slips, which, based on the dual-state digital twin model of a single Qin bamboo slip, establishes the composition relationship between multiple Qin bamboo slips and writes relevant evidence information into the model, forming a digital model that can completely reflect the composition state and interrelationships of multiple Qin bamboo slips.

[0061] S4. Based on the dispute focus-driven local evidence reconstruction interaction mechanism, it automatically identifies the dispute type and dispute area of ​​the Qin bamboo slip digital twin, extracts the core evidence combination from the underlying multimodal data, performs targeted reconstruction and local re-rendering of the dispute area, outputs a multi-dimensional discrimination view, and dynamically updates the evidence weight and relationship confidence of the Qin bamboo slip digital twin according to the user's confirmation, denial or revision operation, forming a closed-loop operation mechanism.

[0062] The key points of contention are the areas of confusion or differing opinions users have regarding the Qin bamboo slips digital twin, such as disputes over the arrangement or interpretation of a particular slip. Other aspects include: partial evidence reconstruction (re-extracting and analyzing relevant evidence from underlying multimodal data to reconstruct and interpret the disputed areas); targeted reconstruction (reconstructing digital models specifically for the disputed areas based on core evidence combinations to more accurately reflect their potential state); partial re-rendering (re-rendering the reconstructed disputed areas within the digital twin for clearer visual presentation); and multi-dimensional discriminative views (viewing the disputed areas from different angles and dimensions, such as appearance, textual features, and semantics). The system includes logic, outputting a view to help users assess disputed areas and providing comprehensive information for decision-making; evidence weighting, an indicator that measures the importance of different pieces of evidence in determining relevant characteristics of the Qin bamboo slips, such as their arrangement and morphology, dynamically adjusted based on user actions; relationship confidence, representing the reliability of various relationships within the Qin bamboo slips digital twin, such as arrangement and morphological evolution relationships, also dynamically updated based on user feedback; and a closed-loop operation mechanism that automatically identifies user disputes, performs targeted reconstruction, receives user feedback, and dynamically updates evidence weighting and relationship confidence, forming a continuously cyclical, self-improving, and optimized system operation mode, enabling the Qin bamboo slips digital twin to continuously approximate reality and meet user needs.

[0063] It should be noted that, during use, S1 acquires and preprocesses multimodal data, providing a comprehensive and standardized data foundation for subsequent modeling, ensuring information integrity and accuracy. S2 constructs a dual-state model based on degradation inversion, which can accurately present the current state and original form of Qin bamboo slips, providing an intuitive basis for studying their evolution. S3 uses a three-chain coupling method to establish the splicing relationship and record detailed information, making the Qin bamboo slip combination relationship more scientific and reasonable, enhancing the model's credibility and traceability. S4's dispute focus-driven mechanism can automatically identify user disputes and output multi-dimensional views through targeted reconstruction, helping users to better understand and judge. It also dynamically updates according to user operations, forming a closed loop, allowing the model to be continuously optimized and improved with user feedback, improving interactivity and practicality. The overall design is comprehensive, scientific, and dynamically adaptable, effectively promoting the research and application of Qin bamboo slips.

[0064] In one embodiment, acquiring multimodal data from Qin bamboo slips and performing standardized preprocessing includes:

[0065] High-resolution scanning, 3D laser point cloud acquisition, microscopic imaging, and spectral analysis were used to collect multi-dimensional, multimodal data from the Qin bamboo slips. The acquired multimodal data covers the handwriting information, surface morphology information, material texture information, crack information, edge contour information, and contamination occlusion information of the Qin bamboo slips. The multimodal data set is denoted as:

[0066]

[0067] in, For handwriting information, For surface morphology information, For material texture information, For fracture information, For edge contour information, To obscure information due to pollution;

[0068] Targeted preprocessing operations are performed based on the characteristics and noise properties of different types of multimodal data:

[0069] The handwriting information is processed sequentially by image deblurring, grayscale enhancement, and stroke contour extraction to reduce the interference of background noise on handwriting features;

[0070] Point cloud denoising, coordinate registration, and texture mapping are performed on surface morphology and material texture information to improve the accuracy and completeness of point cloud data;

[0071] Feature point extraction, contour fitting, and fracture surface feature quantization are performed on crack and edge contour information to enhance the recognizability of geometric features;

[0072] The pollution occlusion information is processed by masking, background separation, and occlusion area localization to clarify the scope and type of pollution occlusion. After preprocessing, a standardized and fusionable multimodal dataset is obtained:

[0073]

[0074] This design utilizes multiple methods to collect multi-dimensional, multimodal data from Qin bamboo slips, and performs targeted preprocessing based on the characteristics and noise properties of different data types. The multi-method acquisition comprehensively gathers various types of information from the Qin bamboo slips, providing rich material for subsequent research. Targeted preprocessing effectively weakens background noise interference, improves data accuracy and completeness, enhances the identifiability of geometric features, and clarifies the scope and type of contamination. The preprocessed data results in a standardized and fusionable dataset, laying a solid foundation for subsequent modeling and analysis. This ensures that the model built based on this data can more accurately and realistically reflect the original state and characteristics of the Qin bamboo slips, improving the scientific rigor and reliability of the research.

[0075] In one embodiment, a degradation inversion-driven dual-state twin modeling method is used to perform micro-region division and degradation parameter construction on preprocessed multimodal data, invert the morphological evolution path of Qin bamboo slips, and construct a current twin layer representing the current true preservation state and an original appearance inversion layer representing the original morphological estimation result, including:

[0076] Based on the physical dimensions and characteristic distribution patterns of the Qin bamboo slips, a grid division method was used to uniformly divide the physical space of the Qin bamboo slips into [a specific grid]. Let the nth local micro-region be denoted as . Line number The local micro-region of the column is ,in , , A positive integer was set based on the actual size and analytical precision of the Qin bamboo slips to achieve a refined regionalized representation of the morphological features of the Qin bamboo slips.

[0077] For each local micro-region Combining the physical laws of damage and degradation of Qin bamboo slips with multimodal data characteristics, a degradation state parameter set was constructed, which includes four core indicators: handwriting fading, surface corrosion, crack propagation, and pollution occlusion.

[0078]

[0079] in, For the fading factor of the lettering, For the degree of surface corrosion, For crack propagation coefficient, To determine the percentage of pollution shielding, all degradation state parameters were quantitatively calculated based on standardized multimodal data. The calculation formula is as follows:

[0080]

[0081]

[0082]

[0083]

[0084] In the formula, For local micro-regions The average grayscale value of the handwriting information after internal preprocessing. The standard grayscale mean value represents the original writing state of Qin bamboo slips of the same type. For local micro-regions The pixel area or physical area of ​​the etched region on the inner surface. For local micro-regions The total pixel area or physical area, For local micro-regions The total pixel length or physical length of the internal crack. For local micro-regions The total pixel length or physical length of the edge. For local micro-regions The pixel area or physical area of ​​the area obscured by internal pollution;

[0085] Based on standardized multimodal data and the set of degradation state parameters for each local micro-region By combining historical evolution factors such as material aging and environmental erosion of Qin bamboo slips, an evolution function in the time dimension is constructed to reverse the stage-by-stage evolution path of Qin bamboo slips from their original writing state to their current damaged state:

[0086]

[0087] in, Using the completion time of the Qin bamboo slips as the starting point, a time dimension parameter is used to achieve traceability and quantification of the damage process;

[0088] Based on the morphological evolution path of the Qin bamboo slips and standardized multimodal data, and by integrating the actual damage state, geometric features, and texture information of each local micro-region, a current twin layer representing the true preservation state of the Qin bamboo slips is constructed through 3D modeling and texture mapping. ,

[0089]

[0090] in, The existing twin layer modeling function accurately restores all the existing physical features and damaged state of the Qin bamboo slips;

[0091] Based on the morphological evolution path, reverse inference calculations are performed to eliminate the influence of degradation factors such as fading of characters, surface corrosion, crack expansion, and contamination on the original form of Qin bamboo slips. The missing and damaged original features are reasonably inferred and reconstructed to construct an original appearance inversion layer that represents the inferred original form of characters, edges, and surface structure. ,

[0092]

[0093] in, Modeling functions for the original appearance inversion layer are used to realize the scientific estimation and visualization of the original form of Qin bamboo slips, and finally form a basic model of a single Qin bamboo slip with dual-state digital twin that has the ability to restore the current state and estimate the original appearance.

[0094] This design involves dividing the preprocessed data into micro-regions, constructing degradation parameters, reversing the morphological evolution path, and building a current state and original state inversion layer. The micro-region division enables refined regional characterization of morphological features, and the construction of a degradation parameter set allows for quantitative calculation of damage. Combining historical factors, an evolution function is constructed to achieve traceable and quantifiable damage processes. Based on this, a dual-state model is constructed. The current state twin layer accurately restores existing features, while the original state inversion layer scientifically infers the original form. It has both the ability to restore the current state and infer the original form, providing an intuitive and accurate digital model for the study of Qin bamboo slips and helping to explore the evolution and original state of Qin bamboo slips in depth.

[0095] In one embodiment, a self-evident splicing modeling method based on three-chain coupling is adopted, which integrates the collaborative calculation results of three types of evidence chains: physical linking, writing, and semantics, to establish the splicing relationship between multiple Qin bamboo slips, including:

[0096] Physical morphology-related features are extracted from the current twin layer of the Qin bamboo slips dual-state digital twin basic model to construct a physical coherent evidence chain feature set:

[0097]

[0098] in, For hole position coordinate matching degree, Hole spacing deviation rate, For edge fracture complementary coefficient, For fiber continuity similarity, For thickness wear synergy, each feature is quantitatively calculated based on the physical geometric parameters of the Qin bamboo slips;

[0099] Weights of each feature in a physically linked chain of evidence The Analytic Hierarchy Process (AHP) was used to determine the importance of physical binding characteristics based on research findings in bamboo and wooden slips. Normalized weights were obtained after a consistency test, satisfying the following conditions: The weighted summation method is used to calculate the weighted summation of any two Qin bamboo slips. Physically linked evidence chain comprehensive matching degree :

[0100]

[0101] in, Qin bamboo slips The The normalized matching value of each physical cohesion feature, with a value range of [value range missing].

[0102] Writing-related features are extracted from the original appearance inversion layer of the Qin bamboo slips dual-state digital twin basic model to construct a writing evidence chain feature set:

[0103]

[0104] in, For high character deviation rate, For consistent line spacing, For the similarity of the direction of the strokes, For ink color transition coefficient, To write the tilt angle matching degree, For character style similarity, each feature is quantitatively calculated based on the writing parameters of Qin bamboo slips.

[0105] Write the weights of each feature in the chain of evidence. The coefficient of variation method is used to determine the coefficient of variation of each writing feature in the sample set. The proportion of the coefficient of variation is used as the normalization weight, satisfying the following conditions: Then, the weighted summation method is used to calculate the weighted summation of any two Qin bamboo slips. Comprehensive matching degree of the written evidence chain :

[0106]

[0107] in, Qin bamboo slips The The normalized matching values ​​of each writing feature, with a value range of [value range missing]. ;

[0108] From the original inversion layer of the Qin bamboo slips dual-state digital twin basic model, textual content-related features are extracted, and semantic evidence chain feature sets are constructed by combining knowledge of paleography and bamboo slips:

[0109]

[0110] in, For syntactic coherence, For term co-occurrence frequency, For document format matching, For contextual logical consistency, To determine the continuity of the clause structure, each feature is quantitatively calculated based on the semantics of Qin bamboo slips and document rules;

[0111] Weights of each feature in the semantic evidence chain The method combines expert scoring with entropy weighting. First, experts in bamboo and wooden slips studies and paleography provide subjective weights. Then, the objective weights of each semantic feature are calculated using entropy weighting. These objective weights are then combined in a subjective:objective ratio of 7:3 to obtain the normalized weights, satisfying the following conditions: Then, the weighted summation method is used to calculate the weighted summation of any two Qin bamboo slips. semantic evidence chain comprehensive matching degree :

[0112]

[0113] in, Qin bamboo slips The The normalized matching values ​​of each semantic feature, with a value range of [value range missing]. ;

[0114] Weighting coefficients of three-chain evidence The domain adaptation method is used to determine the adaptation weights based on the Qin bamboo slips type:

[0115] For Qin bamboo slips containing official documents, a system was established. ;

[0116] For Qin bamboo slips containing private contracts, set up ;

[0117] For Qin bamboo slips containing classical texts, set up ,

[0118] All meet Then, the multi-source evidence fusion method is used to calculate the value of any two Qin bamboo slips. Three-chain coupled comprehensive splicing confidence :

[0119]

[0120] The range of the overall compilation confidence level is as follows: The higher the value, the higher the degree of matching between the two Qin bamboo slips;

[0121] Constructing confidence threshold The percentile method was used to determine the overall reconstruction confidence level based on the validated Qin bamboo slip reconstruction sample set. quantile value as threshold , ;

[0122] If any two Qin bamboo slips Comprehensive compilation confidence level If a valid splicing relationship exists between the two, a topological modeling method is used to establish the splicing topology of multiple Qin bamboo slips based on all valid splicing relationships, thereby realizing a visual representation of the Qin bamboo slips combination relationship.

[0123] The complete information, including the source of evidence, strength of evidence, type of conflict, candidate substitution relationships, and historical revision trajectory, corresponding to each compilation relationship is structured and stored, and written into the dual-state digital twin basic model to form a Qin bamboo slip combination digital twin with dual-state characteristics, compilation relationships, and evidence traceability capabilities.

[0124]

[0125] in, To assemble the topology, To compile a collection of relevant evidence information.

[0126] This design integrates three types of evidence chains—physical linking, writing, and semantics—to establish the composition relationship of the Qin bamboo slips. It extracts features from different levels to construct evidence chains, comprehensively considering various factors affecting the composition of the Qin bamboo slips. By determining the feature weights of each evidence chain and the weight coefficients of the three chains through different methods, the calculation is more scientific and reasonable. The multi-source evidence fusion method is used to calculate the composition confidence, which can accurately determine the composition relationship. The composition topology is established to achieve visual representation. All information is structured and stored and written into the model, forming a digital twin of the Qin bamboo slips combination with multiple capabilities. This provides a powerful tool for the study of Qin bamboo slip combinations and improves the systematicness and accuracy of the research.

[0127] In one embodiment, based on a dispute-driven local evidence reconstruction interaction mechanism, the system automatically identifies the dispute type and disputed area of ​​the Qin bamboo slips digital twin, extracts core evidence combinations from the underlying multimodal data, performs targeted reconstruction and local re-rendering of the disputed area, and outputs a multi-dimensional discriminative view, including:

[0128] The system receives user queries and feedback commands regarding the Qin Dynasty bamboo slips digital twin via a human-computer interface. Using feature extraction and coordinate positioning techniques, it extracts the two-dimensional / three-dimensional spatial location information of the points of contention from these commands. and types of disputes Dispute types It covers one or more of the following: handwriting interpretation disputes, broken edge matching disputes, splicing relationship disputes, and original appearance restoration result disputes, achieving accurate positioning of the focus of disputes and type identification;

[0129] Based on the identified types of disputes By combining evidence discrimination ability assessment models, multimodal data is standardized from the bottom layer. Extract evidence combinations that are highly relevant to the points of contention. Furthermore, the core evidence is screened by calculating the evidence discrimination index. The formula for calculating the evidence discrimination index is:

[0130]

[0131] in, , As evidence The eigenvalues ​​in the disputed region, This is the variance calculation function. The higher the evidence discrimination value, the more valuable the evidence is for determining disputes.

[0132] Evidence Discrimination Threshold K-means clustering was used to determine the extracted evidence combinations. Cluster analysis was performed on the discriminative power of the clusters, and the critical value between the two clusters was taken as the threshold. ,filter The evidence constitutes the core evidence set This provides precise evidentiary support for resolving disputes;

[0133] Based on the core evidence set To address the different types of disputes requiring different identification methods, targeted reconstruction operations are performed on the disputed areas:

[0134] To address the controversy surrounding handwriting interpretation, image segmentation and stroke enhancement algorithms were employed to separate handwriting from stains and reconstruct indented strokes, highlighting the original features of the handwriting.

[0135] To address the controversy surrounding the matching of fractured edges, a geometric fitting and contour comparison algorithm was used to perform edge interlocking fitting reconstruction and quantify the degree of edge matching.

[0136] For disputes over the splicing relationship, evidence weight recalculation and multi-source fusion algorithms are used to recalculate the evidence weight and reconstruct the candidate relationship to re-evaluate the splicing matching degree.

[0137] To address the controversy surrounding the original appearance restoration results, a credibility quantification and hierarchical reconstruction algorithm was employed to perform hierarchical reconstruction of the original appearance estimation credibility, thereby clarifying the reliability of the original morphology estimation.

[0138] For the disputed area after targeted reconstruction, local re-rendering is performed using visualization rendering technology. Based on the dispute type, a corresponding multi-dimensional discrimination view is output. This discrimination view includes at least one of the following: handwriting and stain separation view, indentation stroke enhancement view, edge stitching fitting view, candidate character shape comparison view, comparison view of supporting and opposing evidence, and layered view of the credibility of the original appearance presumption. The rendering feature values ​​of each discrimination view are calculated based on the core evidence set, using the following formula:

[0139]

[0140] in, The rendering function is set according to visualization requirements. It visualizes and intuitively presents the spatial scope of the disputed area and the evidence in dispute, providing clear visual support for users to determine the dispute.

[0141] This design automatically identifies the type and area of ​​dispute, extracts core evidence for targeted reconstruction, and re-renders the output discriminative view. It can accurately locate the focus of the dispute and identify the type, filter core evidence from massive amounts of data, provide precise support for dispute judgment, and reconstruct the data for different types of disputes to meet diverse judgment needs. Combined with visualization technology, it re-renders and outputs a multi-dimensional discriminative view, making the disputed evidence visual and intuitive, providing users with clear visual support, making it easier for users to understand the points of dispute, improving the efficiency and accuracy of dispute handling, and promoting the exchange and discussion of Qin bamboo slips research.

[0142] In one embodiment, the evidence weight and relationship confidence of the Qin bamboo slips digital twin are dynamically updated based on the user's confirmation, denial, or revision operations, forming a closed-loop operation mechanism, including:

[0143] The system receives the user's interactive operation results on the output discriminant view through a human-computer interaction interface. Operation results This includes one of the following: confirmation, denial, or revision of the current twin characteristics or splicing relationship, to achieve accurate capture of user interaction intent;

[0144] Develop differentiated twin update strategies based on different user operation results:

[0145] like To confirm this, it indicates that the user accepts the characteristic representation, arrangement relationship, or evidentiary presumption results of the current Qin bamboo slips digital twin, thus maintaining the evidentiary weight of the current Qin bamboo slips digital twin. And the confidence level of the composition The interaction remains unchanged, and the time, operation content, dispute results, and other information of the interaction are structured and written into the historical trajectory database of the twin, so as to realize the full recording of the interaction process;

[0146] like A negative result indicates that the user does not accept the current feature representation, arrangement relationship, or evidence presumption results of the Qin bamboo slips digital twin; the weight decay coefficient is applied. The dispute level matching method is used to determine:

[0147] For general disputes Set up important disputes Set up a core dispute , The feature weights of the evidence chain corresponding to the dispute type are updated by decay based on the dispute level. The update formula is as follows:

[0148]

[0149] in, The original weights, The updated weights are used to recalculate the concatenated confidence level after weight decay. If the updated composition confidence level If so, the splicing relationship between the two corresponding Qin bamboo slips is immediately terminated, realizing the dynamic adjustment of the splicing relationship;

[0150] like For revision, indicating that users have put forward new academic viewpoints or revision opinions on the feature representation, splicing relationship or evidence presumption results of the current Qin bamboo slips digital twin, the weight attenuation coefficient is used. Determined using the dispute level matching method, the core evidence set is analyzed based on the characteristics, weighting, or compilation relationships of the evidence revised by the user. After supplementation, correction, or reconstruction, the evidence weights are recalculated using the weighted fusion method, and the updated formula is:

[0151]

[0152] in, The user is assigned new weights based on academic research. These new weights must meet the normalization constraints of the corresponding chain of evidence.

[0153] At the same time, based on the revised core evidence set and the new weights, the overall matching degree of each chain of evidence was recalculated. And the confidence level of the composition This enables personalized updates of evidence weight and compilation confidence.

[0154] The updated evidence weights, compilation confidence, compilation relationships, and full records of this interaction are synchronized to the underlying data and model structure of the Qin bamboo slips digital twin, completing the dynamic update and iterative optimization of the twin. This ultimately forms a closed-loop operation mechanism of multimodal acquisition → degradation inversion → compilation modeling → dispute triggering → local evidence reconstruction → twin update, enabling the Qin bamboo slips digital twin to continuously evolve and improve as academic research deepens.

[0155] This design creates a closed loop by dynamically updating evidence weights and relationship confidence based on user actions. It accurately captures user interaction intentions, formulates differentiated update strategies for different actions, confirms the interaction process of confirmed actions, dynamically adjusts the compilation relationship for rejected actions, and revises actions to achieve personalized updates of evidence weights and compilation confidence. The updated information is synchronized to the underlying data and model structure, completing the dynamic update and iterative optimization of the digital twin, forming a complete closed-loop operation mechanism. This allows the Qin bamboo slip digital twin to continuously evolve and improve as academic research deepens, always keeping pace with the latest research results and improving the timeliness and accuracy of research.

[0156] Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0157] The above embodiments provide a detailed description of the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for constructing and interacting with a digital twin of Qin bamboo slips based on multimodal data, characterized in that, Includes the following steps: S1. Obtain multimodal data from Qin bamboo slips and perform standardized preprocessing to obtain a set of standardized multimodal data that can be fused. S2. Based on the degradation inversion-driven dual-state twin modeling method, the standardized multimodal data is divided into micro-regions and degradation parameters are constructed to invert the morphological evolution path of the Qin bamboo slips, construct the current twin layer and the original appearance inversion layer, and form a dual-state digital twin basic model of a single Qin bamboo slip. S3. Adopting a self-evident splicing modeling method based on three-chain coupling, integrating the collaborative calculation results of three types of evidence chains, establishing splicing relationships between multiple Qin bamboo slips, writing splicing-related evidence information into the dual-state digital twin basic model, and forming a combined digital twin of Qin bamboo slips; S4. Based on the dispute focus-driven local evidence reconstruction interaction mechanism, the system identifies the dispute type and dispute area of ​​the Qin bamboo slip digital twin, extracts the core evidence combination to perform targeted reconstruction and local re-rendering of the dispute area, outputs a multi-dimensional discrimination view, and dynamically updates the evidence weight and relationship confidence of the Qin bamboo slip digital twin according to the user's interaction operation, forming a closed-loop operation mechanism.

2. The method for constructing and interacting with a Qin Dynasty bamboo slip digital twin based on multimodal data according to claim 1, characterized in that, The acquisition of multimodal data from the Qin bamboo slips in step S1 includes the following steps: One or more of the following methods—high-resolution scanning, 3D laser point cloud acquisition, microscopic imaging, and spectral analysis—were used to collect Qin bamboo slips in all dimensions. The multimodal data covers at least one of the following: the handwriting, surface morphology, material texture, cracks, edge contours, and contamination / occlusion information of the Qin bamboo slips.

3. The method for constructing and interacting with a Qin Dynasty bamboo slip digital twin based on multimodal data according to claim 1, characterized in that, The standardization preprocessing described in step S1 includes the following steps: Targeted processing operations are performed on the features and noise characteristics of different types of multimodal data, including one or more of the following processing methods for handwriting information: deblurring, grayscale enhancement, and stroke outline extraction. Perform one or more of the following processing on the surface morphology and material texture information: point cloud denoising, coordinate registration, and texture mapping; One or more of the following processes are applied to the crack and edge contour information: feature point extraction, contour fitting, and fracture surface feature quantization. The pollution-occluded information is processed by one or more of the following methods: masking, background separation, and occlusion area localization.

4. The method for constructing and interacting with a Qin Dynasty bamboo slip digital twin based on multimodal data according to claim 1, characterized in that, The micro-region division described in step S2 includes the following steps: Based on the physical dimensions and characteristic distribution patterns of Qin bamboo slips, the physical space of Qin bamboo slips is divided into several local micro-regions using the grid division method, so as to realize the refined regionalized representation of the morphological characteristics of Qin bamboo slips. The degradation parameter construction involves constructing a set of degradation state parameters for each local micro-region, combining the physical laws of Qin bamboo slip damage and degradation with multimodal data characteristics. These parameters include one or more of the following indicators: handwriting fading, surface corrosion, crack propagation, and pollution occlusion. Each degradation state parameter is quantitatively calculated based on standardized multimodal data.

5. The method for constructing and interacting with a Qin Dynasty bamboo slip digital twin based on multimodal data according to claim 1, characterized in that, The inversion of the evolution path of the Qin bamboo slips described in step S2 includes the following steps: Based on standardized multimodal data and degradation state parameter sets of various local micro-regions, combined with the historical evolution factors of Qin bamboo slips material aging and environmental erosion, an evolution function in the time dimension is constructed to inversely deduce the evolution path of Qin bamboo slips from their original writing state to their current damaged state, so as to realize the traceability and quantification of the damage process.

6. The method for constructing and interacting with a Qin Dynasty bamboo slip digital twin based on multimodal data according to claim 1, characterized in that, The construction of the current state twin layer and the original state inversion layer in step S2 includes the following steps: Based on the evolution path of Qin bamboo slips and standardized multimodal data, and by integrating the physical characteristics of each local micro-region, a current twin layer is constructed through three-dimensional modeling and texture mapping to accurately restore the current preservation state of Qin bamboo slips. By performing reverse inference based on the morphological evolution path, the influence of degradation factors on the original form of Qin bamboo slips is eliminated, the missing and damaged original features are estimated and reconstructed, and the original appearance inversion layer is constructed to realize the scientific estimation and visualization of the original form of Qin bamboo slips.

7. The method for constructing and interacting with a Qin Dynasty bamboo slip digital twin based on multimodal data according to claim 1, characterized in that, The three types of evidence chains mentioned in step S3 are physical codification, written evidence, and semantic evidence chains, including the following steps: Feature sets for each evidence chain are constructed by extracting corresponding features from the bi-state digital twin basic model, and normalized matching values ​​of features within each evidence chain are obtained by quantitative calculation. The feature weights within each evidence chain are determined by combining one or more of the following methods: analytic hierarchy process, coefficient of variation method, expert scoring method, and entropy weight method. The comprehensive matching degree of each evidence chain between any two Qin bamboo slips is calculated using the weighted summation method.

8. The method for constructing and interacting with a Qin Dynasty bamboo slip digital twin based on multimodal data according to claim 7, characterized in that, Step S3, which establishes the arrangement relationship between multiple Qin bamboo slips, includes the following steps: The domain adaptation method is used to determine the weight coefficients of the three chains of evidence: physical binding, writing, and semantics. Different weight allocation ratios are adapted according to the type of Qin bamboo slips. The multi-source evidence fusion method is used to calculate the confidence of the three-chain coupling comprehensive splicing of any two Qin bamboo slips. The percentile method was used to determine the confidence threshold for the assembly. Two Qin bamboo slips whose comprehensive assembly confidence reached the threshold were judged to have a valid assembly relationship. Based on all valid assembly relationships, the topology of the assembly of multiple Qin bamboo slips was established by the topology modeling method. The information of at least one of the following—the source of evidence, the strength of evidence, the type of conflict, the candidate substitution relationship, and the historical revision trajectory—is written into the dual-state digital twin basic model to form a combined digital twin of Qin bamboo slips.

9. The method for constructing and interacting with a Qin Dynasty bamboo slip digital twin based on multimodal data according to claim 1, characterized in that, Step S4, which involves identifying the type of dispute and the area of ​​dispute, includes the following steps: The system receives user interaction queries and dispute feedback commands through a human-computer interaction interface. It uses feature extraction and coordinate positioning technology to extract the spatial location information of the dispute focus and the dispute type from the commands. The dispute type includes one or more of the following: handwriting interpretation, broken edge matching, splicing relationship, and original appearance restoration result dispute. The process of extracting core evidence sets involves extracting relevant evidence sets from the underlying standardized multimodal data based on the type of dispute, screening core evidence by calculating evidence discrimination, determining the evidence discrimination threshold using the K-means clustering method, and using evidence that reaches the threshold as the core evidence set.

10. The method for constructing and interacting with a Qin Dynasty bamboo slip digital twin based on multimodal data according to claim 1, characterized in that, The targeted reconstruction and local re-rendering described in step S4 include the following steps: Based on the core evidence set, corresponding algorithms are used to perform targeted reconstruction of the disputed area for different types of disputes. For handwriting interpretation disputes, image segmentation and stroke enhancement algorithms are used; for broken edge matching disputes, geometric fitting and contour comparison algorithms are used; for splicing relationship disputes, evidence weight recalculation and multi-source fusion algorithms are used; and for original appearance restoration result disputes, credibility quantification and hierarchical reconstruction algorithms are used. By combining visualization rendering technology, the disputed area after targeted reconstruction is locally re-rendered, and a multi-dimensional discrimination view corresponding to the dispute type is output. The dynamic updating of evidence weights and relationship confidence scores receives user confirmation, denial, and revision interaction results for the discriminative view, and formulates differentiated update strategies. For negative operations, the evidence weights are updated according to the weight attenuation coefficient based on the level of dispute and the assembled confidence is recalculated. For revision operations, the core evidence set is supplemented and corrected and the calculation is updated in combination with the new weights specified by the user. The interaction records and updated parameters are synchronized to the Qin bamboo slips digital twin to form a closed-loop operation mechanism.

Images