Method for constructing three-dimensional visual digital archives of cultural relic protection projects

Abstract

The application relates to the computer technical field and discloses a cultural relic protection engineering three-dimensional visualization digital archive construction method, which comprises archive initialization and bidirectional index construction, generation of a reference surface grid and a sparse voxel octree, construction of bidirectional indexes of the two, initialization of a time sequence attribute list, space-time calibration and data stream synchronization, establishment of a space conversion relationship between engineering equipment and a global coordinate system, space-time anchoring data collection, solving of an instantaneous action body by using the space conversion relationship, indexing of a target voxel, and encapsulation of a time sequence attribute data package, attribute writing and causal marking, writing of the time sequence attribute list in a data appending mode and generation of a causal relationship marking, and fusion visualization and query, wherein the bidirectional indexes and the causal marking are used for query execution. The grid and voxel bidirectional indexes are constructed, space-time anchoring and data appending are combined, and the analysis, visualization and query interactivity of the cultural relic protection digital archive are improved.

Description

Technical Field

[0001] This invention relates to the field of computer science and technology, specifically to a method for constructing three-dimensional visualization digital archives for cultural relic protection projects. Background Technology

[0002] Cultural relics are irreplaceable and precious cultural heritage, especially complex structures such as wooden buildings. The completeness, accuracy, and usability of their conservation project archives are crucial. Traditional cultural relic conservation project archives mainly rely on two-dimensional drawings, photographs, and written reports. These formats have significant limitations in terms of information capacity, retrieval efficiency, and intuitiveness. For example, traditional paper archives are voluminous and fragmented, making it difficult to efficiently access historical information during subsequent restoration work. Furthermore, two-dimensional drawings cannot comprehensively and accurately record and express the complex three-dimensional form and structural relationships of wooden buildings, failing to meet the needs of precise monitoring and quality traceability during construction.

[0003] To address these issues, 3D digitization technologies, such as 3D laser scanning, photogrammetry, and even Building Information Modeling (BIM), have begun to be introduced into the field of cultural relic preservation. These technologies can generate high-precision 3D surface models of cultural relics, achieving significant progress in visualizing and representing geometric forms, overcoming the dimensional limitations of traditional 2D drawings. However, when applying these 3D models to complex conservation projects, especially in recording dynamic data of the restoration process and reflecting the internal state of cultural relics, existing methods for constructing 3D digital archives generally have shortcomings.

[0004] Current mainstream 3D digital archiving technologies typically treat engineering data, such as restoration records and monitoring data, as independent files, managing them by adding hotspots or hyperlinks to the surface of the 3D model. In this model, the 3D geometric model and dynamic engineering processes, physical property sensing, and other multimodal data are statically separated at the data structure level. This structural deficiency firstly makes it impossible to accurately anchor and collect key operations, such as the instantaneous trajectory of restoration tools and process parameters, in the spatiotemporal context of the artifact, leading to significant difficulties in construction process supervision and quality traceability. Valuable construction techniques also face the risk of being lost due to the lack of standardized records. Simultaneously, this shallow linking mode based on surface meshes lacks an effective structure to carry internal physical property sensing data, such as ground-penetrating radar signals, causing the archives to fail to reflect the true state of the building's interior. This restricts risk warning and proactive protection, forcing conservation work into a passive mode of rescue restoration. Finally, because the operational data and the state data are stored separately in time and space, the system cannot automatically establish an inherent logical relationship between engineering intervention and changes in the artifact's state. This results in a lack of data support for the quantitative assessment of restoration effects, greatly weakening the application value of digital archives in decision support. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method for constructing three-dimensional visualized digital archives for cultural relic protection projects. This method aims to solve the problems of existing digital archives, which suffer from structural separation between geometric models and multimodal temporal data, resulting in difficulties in accurately anchoring key operational data in time and space, lack of effective carrying and penetrating query of internal state data, and inability to automatically establish causal relationships in engineering projects.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] The method for constructing three-dimensional visualization digital archives for cultural relic protection projects includes the following steps:

[0008] Step S1: Perform digital archive initialization and bidirectional index construction, generate a reference surface mesh and a sparse voxel octree, and construct a reference mesh and voxel bidirectional index between the reference surface mesh and the sparse voxel octree. The leaf node voxels of the sparse voxel octree are initialized with a temporal attribute list.

[0009] Step S2: Perform spatiotemporal calibration and multimodal data stream synchronization of engineering equipment and digital archives, establish real-time spatial transformation relationship between tools and non-destructive testing sensors and the global coordinate system of the sparse voxel octree, and assign a unified timestamp to the collected engineering process flow and physical property sensing flow.

[0010] Step S3: Perform spatiotemporally anchored digital archive data acquisition, calculate the instantaneous action volume using the real-time spatial transformation relationship, use the instantaneous action volume to query the sparse voxel octree to index the target voxel set, and encapsulate to generate a time-series attribute data package.

[0011] Step S4: Perform digital archive temporal attribute writing and causal relationship marking. Write the temporal attribute data packet into the temporal attribute list of the voxels in the target voxel set in the form of data appending. When the engineering process data packet and the physical property sensing data packet meet the spatiotemporal proximity determination condition, generate a causal relationship mark.

[0012] Step S5: Perform digital archive fusion visualization and multidimensional query, using the baseline grid and voxel bidirectional index and the causal relationship marker to perform visualization query, spatial penetration query and causal relationship query.

[0013] In one embodiment of the present invention, the reference grid and voxel bidirectional index include:

[0014] A positive mapping from the leaf node voxel to the surface texture coordinates of the reference surface mesh;

[0015] A reverse mapping from the surface triangular facets of the reference surface mesh to the voxel list of the leaf node voxels.

[0016] In one embodiment of the present invention, the file initialization and bidirectional index construction step S1 specifically includes:

[0017] The reference surface mesh is generated by acquiring raw data through 3D scanning;

[0018] The sparse voxel octree is constructed based on the bounding space of the reference surface mesh, and the temporal attribute list is initialized for each leaf node voxel of the sparse voxel octree.

[0019] In one embodiment of the present invention, the spatiotemporal calibration and multimodal data stream synchronization step S2 specifically includes:

[0020] Install a six-degree-of-freedom spatial tracker on the tool and the non-destructive testing sensor;

[0021] The real-time spatial transformation relationship between the tool, the non-destructive testing sensor, and the global coordinate system is established using a hand-eye calibration method.

[0022] In one embodiment of the present invention, the spatiotemporally anchored data acquisition step in step S3 specifically includes:

[0023] The instantaneous action volume is calculated using the real-time spatial transformation relationship and the predefined local action volume;

[0024] The transient actor is used to perform a spatial intersection query in the sparse voxel octree to index the target voxel set.

[0025] In one embodiment of the present invention, in the time-series attribute writing and causal relationship marking step S4, the data appending method is to add the time-series attribute data packet to the end of the time-series attribute list. This method is used to preserve the historical process and is different from data overwriting.

[0026] In one embodiment of the present invention, the spatiotemporal proximity determination condition includes:

[0027] Voxel clusters that are spatially determined to be composed of the same voxel or adjacent voxels;

[0028] The time-based determination is based on the difference between the timestamps of the engineering process data packet and the physical property sensing data packet being less than a preset time threshold.

[0029] In one embodiment of the present invention, in the S5 step of integrating visualization and multidimensional query, the visualization query includes:

[0030] The sparse voxel octree is searched to obtain voxels that meet the conditions;

[0031] Using the forward mapping in the reference mesh and the voxel bidirectional index, an attribute heatmap is dynamically rendered on the high-resolution texture of the reference surface mesh.

[0032] In one embodiment of the present invention, in the fusion visualization and multidimensional query step of step S5, the spatial penetration query includes:

[0033] In response to a user's query point on the reference surface grid, determine the triangular facet where the query point is located;

[0034] Using the inverse mapping in the reference grid and the voxel bidirectional index, an index is generated to the voxel list associated with the triangular facet;

[0035] Extract and display the time-series attribute list corresponding to the voxel list.

[0036] In one embodiment of the present invention, in the fusion visualization and multidimensional query step of step S5, the causal relationship query includes:

[0037] Retrieve the causal relationship markers generated in S4;

[0038] The system displays to the user the engineering process data package and the physical property sensing data package that are linked by the causal relationship.

[0039] This invention provides a method for constructing three-dimensional visualization digital archives for cultural relic protection projects. It has the following beneficial effects:

[0040] 1. This invention lays the structural foundation for digital archives by constructing a reference surface grid and a sparse voxel octree, and establishing a bidirectional index between the reference grid and voxels. Then, through spatiotemporal calibration, accompanying acquisition, and data appending, multimodal engineering process flow and physical property sensing flow data are accurately anchored to the voxel space in a way that preserves history, thus constructing a spatiotemporal archive of cultural relic protection projects that can be completely traced back.

[0041] 2. By setting spatiotemporal proximity judgment conditions and automatically generating causal relationship markers, this invention realizes the inherent logical connection between engineering operation data and cultural relic status data. This design enables the three-dimensional visualization digital archive of cultural relic protection projects to not only store data, but also provide direct data support for the effect evaluation and mechanism analysis of protection measures.

[0042] 3. This invention supports efficient multidimensional queries through a bidirectional index of a reference grid and voxels. The forward mapping of this index is used to realize the fusion visualization of attribute data on high-definition textures, and the reverse mapping of this index is used to realize spatial penetration queries from any surface point to its complete temporal attribute list, which helps to improve the interactivity, intuitiveness and ease of use of archival data. Attached Figure Description

[0043] Figure 1 This is a schematic diagram of the method flow of the present invention.

[0044] Figure 2 This is a schematic diagram of the digital archive initialization structure of the present invention;

[0045] Figure 3 This is a schematic diagram of the spatiotemporal calibration and data stream synchronization of the present invention;

[0046] Figure 4 This is a schematic diagram of data acquisition based on spatiotemporal anchoring according to the present invention;

[0047] Figure 5 A schematic diagram illustrating the addition of timing attributes and causal relationship markers to the present invention;

[0048] Figure 6 This is a schematic diagram illustrating the integration of visualization and multidimensional querying in this invention. Detailed Implementation

[0049] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0050] Please see the appendix Figure 1 This invention provides a method for constructing a three-dimensional visualization digital archive for cultural relic protection projects, including the following steps:

[0051] S1: Archive Initialization and Two-Way Index Construction. Executed by the Archive Initialization and Indexing module, this step first acquires the original data of the cultural relic protection project object through 3D scanning, generating a high-resolution reference surface mesh as a visualization base.

[0052] Subsequently, a sparse voxel octree is constructed based on the bounding space of the grid as the core data container, and an empty list of temporal attributes is initialized for each leaf node voxel of the octree.

[0053] Finally, this step constructs a crucial bidirectional index for the reference mesh and voxels: on the one hand, this index establishes a forward mapping from voxels to the texture coordinates of the reference mesh surface; on the other hand, it establishes a reverse mapping from the triangular facets of the reference mesh surface to the list of voxels they permeate.

[0054] S2: Spatiotemporal calibration and multimodal data stream synchronization. Performed by the spatiotemporal calibration and synchronization module. This step involves installing six-degree-of-freedom spatial trackers on all tools (such as laser cleaning heads) and non-destructive testing sensors (such as ground-penetrating radar) involved in the cultural relic conservation project.

[0055] Using a hand-eye calibration method, a real-time spatial transformation relationship is established and maintained between these devices and the voxel archive global coordinate system constructed in the previous step (S1). Simultaneously, this step defines the multimodal data streams to be acquired, mainly including: spatial attitude stream, engineering process stream (such as laser energy parameters), and physical property sensing stream (such as radar signals). The system employs a unified time synchronization protocol to assign high-precision and consistent timestamps to all data streams.

[0056] S3: Data Acquisition Based on Spatiotemporal Anchoring. Executed by the data anchoring and acquisition module. This step builds upon the spatiotemporal reference established in S2, performing dynamic data anchoring and acquisition. At any point during the project, the system utilizes the spatial transformation relationship established in S2 to calculate in real-time the position of the tool or sensor's instantaneous acting body (i.e., its influence range) in the global coordinate system.

[0057] Subsequently, the system uses this instantaneous agent to efficiently query the sparse voxel octree constructed in S1, quickly indexing all leaf node voxels covered or intersecting with this agent, forming a target voxel set. Finally, the system encapsulates the engineering process flow and physical property sensing flow data synchronously collected in S2, along with the timestamp at that moment, into a standardized time-series attribute data packet.

[0058] S4: Time-series attribute writing and causal relationship marking. Performed by the data writing and association module. This step processes the time-series attribute data packet generated in S3. The system appends this data packet to the end of the time-series attribute list of all voxels in the target voxel set determined in S3 (this list is initialized in S1). This appending mechanism ensures that the historical process of cultural relic protection operations is completely preserved, unlike data overwriting.

[0059] Simultaneously, a causal relationship marker unit is triggered: this unit monitors whether engineering process packets and physical property sensing packets are successively written within spatially adjacent voxel clusters and within temporally close thresholds. If this high spatiotemporal proximity condition is met, the system determines that there is a potential correlation between the two, automatically generates a causal relationship marker, which links the two data packets at the data structure level and stores them in the voxel metadata.

[0060] S5: Integration of Visualization and Multidimensional Query. Executed by the integration of query and visualization module. This step is the application phase of the archive, and its query capabilities heavily rely on the bidirectional index built in S1 and the causal relationship tags generated in S4. For visualization queries, the system retrieves the sparse voxel octree in S1 to obtain voxels that meet the criteria, and uses the forward mapping established in S1 to dynamically render an attribute heatmap on the high-resolution texture of the baseline surface mesh in S1.

[0061] For spatial penetration queries, when a user clicks on the reference surface mesh, the system immediately indexes the corresponding voxel list using the reverse mapping established in S1, and extracts and displays the complete temporal attribute list that was appended and written in S4.

[0062] For causal relationship queries, the system retrieves the causal relationship tags generated by S4 and displays the linked engineering process data packets and physical property sensor data packets to the user for effect analysis of cultural relic protection.

[0063] Please see the appendix Figure 2 Step S1 aims to construct an initial digital archive structure for the cultural relic conservation project, which includes a high-precision geometric base, a dynamic data container, and the mapping relationship between the two. The initial digital archive structure is the basic data structure for anchoring, storing, and querying (such as S5) all subsequent dynamic engineering data (such as S3 and S4).

[0064] In one embodiment, step S1 can be divided into three sub-steps: the generation step of the reference surface mesh (S110), the construction step of the sparse voxel octree (S120), and the construction step of the reference mesh and the bidirectional index of voxels (S130).

[0065] In step S110, which generates the reference surface mesh, the system first acquires the original point cloud data set of the cultural relic protection project object using a 3D scanning device (such as a laser scanner or structured light camera). The original point cloud data undergoes a series of preprocessing steps, including point cloud registration, denoising, and filtering.

[0066] Subsequently, a surface reconstruction algorithm (e.g., Poisson reconstruction or Delaunay triangulation) is used to convert the processed point cloud into a triangular mesh model.

[0067] Finally, using UV unwrapping and texture mapping techniques, high-resolution texture maps of the artifact surface are mapped onto the mesh, ultimately generating a baseline surface mesh. Reference surface mesh It is a structured data containing geometric, topological, and textural information, and its definition can be expressed as:

[0068] ;

[0069] In the formula, As a reference surface mesh; This provides the geometric and topological information of the mesh. High-resolution surface texture mapping; To be vertices mapped to UV coordinate mapping function in texture space.

[0070] In step S120, which involves constructing a sparse voxel octree, the system uses the reference surface mesh generated in step S110. To build the core data container.

[0071] First, calculate The three-dimensional bounding box.

[0072] Then, based on the preset file precision (i.e., the minimum size of the target voxel), the space represented by the bounding box is recursively partitioned into an octree. During the partitioning process, only [the following is considered]... Geometric entities intersect or allocate memory for nodes within them, thus forming a sparse voxel octree. .

[0073] Crucially, during the construction process, the system provides Each leaf node voxel Initialize an empty dynamic data structure called the timing attribute list. Timing attribute list It is a container used in the subsequent S4 step to append timing attribute data packets. Therefore, The structure can be defined as:

[0074] ;

[0075] In the formula, It is a sparse voxel octree; This is the root node of the octree; For the octree Each leaf node voxel; For leaf nodes A list of associated, empty time-series properties.

[0076] In step S130, which involves constructing the baseline mesh and the bidirectional voxel index, the system builds upon the data generated in S110. and generated by S120 Build a mapping file to connect the two. This index Includes forward mapping and reverse mapping .

[0077] Forward mapping The construction of this layer establishes a mapping from data containers (leaf node voxels) to the visualization substrate (texture), which is the foundation of blended rendering in S5. This process traverses... Each leaf node voxel in For each Calculate the coordinates of its center point. .

[0078] Subsequently, geometry Execute the nearest point query Find the distance nearest surface point Once this surface point is found The system utilizes the UV coordinate mapping function generated in S110. ,calculate In texture mapping UV coordinates on This mapping relationship is stored, and the process can be represented as follows:

[0079] ;

[0080] In the formula, leaf node voxels The corresponding positive mapping result is a UV coordinate; leaf node voxels The coordinates of the center point; The geometric and topological information of the reference surface mesh; This is a nearest-point lookup function, used to find the nearest point. Find the distance ; This refers to the UV coordinate mapping function defined in S110.

[0081] Reverse mapping The construction of this is used to establish a mapping from the visualization base (surface patches) to the data container (leaf node voxels), which is the basis for spatial penetration queries in S5.

[0082] This process iterates through Middle face set Each triangular facet For each The system is Execution space intersection test To find all related intersecting or adjacent leaf node voxels of geometry All indexed This forms a list of voxels. The mapping is stored, and the process can be represented as follows:

[0083] ;

[0084] In the formula, Triangular facet The corresponding reverse mapping result is a list of voxels; A sparse voxel octree constructed in S120; This is a function for testing spatial intersection.

[0085] After completing steps S110, S120, and S130 above, an initialized digital archive structure for cultural relic conservation projects is obtained. This structure includes a high-precision geometric base (…). ), ready-to-use dynamic data containers ( and ), and an efficient bidirectional index connecting the two ( This provides the necessary data foundation and structural support for achieving spatiotemporal anchoring, data appending, and fusion querying in subsequent steps S2 to S5.

[0086] Please see the appendix Figure 3 Step S2 aims to establish a unified spatiotemporal reference for all equipment involved in the cultural relic protection project, ensuring that all dynamically acquired data can be accurately aligned with the global coordinate system and unified timeline of the leaf node voxel archive constructed in S1. This is a necessary prerequisite for the subsequent step S3 to achieve spatiotemporal anchoring of data.

[0087] In one embodiment, step S2 can be divided into two sub-steps: device space calibration step (S210) and multimodal data stream synchronization step (S220).

[0088] In the equipment space calibration step S210, the system installs six-degree-of-freedom space trackers (e.g., optical markers or inertial measurement units) on all the tools involved in the project (e.g., laser cleaning heads, hardened syringes) and non-destructive testing sensors (e.g., ground-penetrating radar, hyperspectral cameras).

[0089] First, using the standard hand-eye calibration method, the static transformation relationship of the "point of action" or "body of action" coordinate system of each tool relative to its tracker coordinate system is pre-calculated and stored.

[0090] Subsequently, during the engineering implementation, the six-degree-of-freedom tracking system acquires in real time the dynamic transformation relationship between the tracker coordinate system and the global coordinate system of the leaf node voxel file defined in S1. The system performs matrix multiplication on these two transformation relationships to calculate the instantaneous pose of the tool's point of action or action body coordinate system in the S1 global coordinate system in real time. This process can be represented as:

[0091] ;

[0092] In the formula, for The transformation matrix from the time tool coordinate system to the global coordinate system; for The transformation matrix from the time tracker coordinate system to the global coordinate system (provided in real time by the tracking system); This is the static transformation matrix from the tool coordinate system to the tracker coordinate system (pre-calculated through hand-eye calibration).

[0093] In the multimodal data stream synchronization step S220, the system defines the various heterogeneous data streams that need to be collected, mainly divided into three categories: spatial attitude stream (i.e., the instantaneous pose calculated in S210). ), engineering process flow (such as tool operation parameters such as laser energy, frequency, and pulse width) and physical property sensing flow (such as the state of the measured object such as ground-penetrating radar echo signal array and spectral curve data).

[0094] To ensure time consistency of all data streams, the system uses a master clock source and synchronizes the time of all data acquisition modules (including the space tracking system, tool controller, and non-destructive testing sensors) via Network Time Protocol (NTP) or Precision Time Protocol (PTP).

[0095] Before entering the processing layer, all collected data is appended with a uniform, high-precision timestamp allocated by the master clock source, forming a synchronous data stream. The general structure of this synchronous data stream can be represented as follows:

[0096] ;

[0097] In the formula, for Any synchronous data stream at any given moment (such as engineering process stream or physical property sensing stream). for A high-precision timestamp assigned by a unified time protocol; A unique identifier for the device (tool or sensor) from which the data originates; for Real-time data (such as parameter values ​​or signal arrays) collected at all times.

[0098] After completing steps S210 and S220, the system obtains the real-time precise pose (spatial attitude stream) of all engineering equipment in the S1 archive space, as well as the engineering process and physical property sensing data stream with a unified high-precision timestamp. This provides all the necessary inputs for the more precise binding and accompanying acquisition of when, what kind of operation, and where it occurs in step S3.

[0099] Please see the appendix Figure 4 Step S3 is the core of the archive construction, which aims to use the spatiotemporal reference established in S2 to realize the real-time spatial positioning of engineering process and physical property status data, and encapsulate them into standardized time-series attribute data packages to prepare for data writing in S4.

[0100] In one embodiment, step S3 can be divided into three sub-steps: instantaneous action volume calculation step (S310), target voxel set efficient indexing step (S320), and time-series attribute data packet standardization encapsulation step (S330).

[0101] In the instantaneous action volume calculation step S310, the system uses the real-time spatial transformation relationship obtained in S2 to determine the precise range of action of the tool or sensor in the cultural relic archive space.

[0102] At any point during the project The system first obtains the instantaneous pose of the tool calculated in S2. At the same time, the tool predefines a local geometric range of action, i.e., the local action volume, within the system.

[0103] The system transforms the local action volume in real time to the global coordinate system of the voxel profile defined in S1 using an instantaneous pose transformation matrix, thus obtaining the instantaneous action volume. This instantaneous action volume is a well-defined geometric entity (e.g., a sphere, a cone, or a ray segment) representing the spatial extent of the operation or detection. This process can be represented as:

[0104] ;

[0105] In the formula, for The instantaneous action of a time-tracking tool or sensor (a geometric entity in the global coordinate system). The solution obtained in S2 The transformation matrix from the time tool coordinate system to the global coordinate system; A predefined local action body (a geometric entity in a local coordinate system) for a tool or sensor.

[0106] In the efficient indexing step S320 of the target voxel set, the system uses the transient action calculated in S310 as the spatial query condition and performs an efficient query in the sparse voxel octree constructed in S1. The query process utilizes the hierarchical structure of the octree, recursively descending from the root node and pruning nodes that are geometrically disjoint with the transient action and all their child nodes. This non-traversal pruning process greatly improves efficiency and quickly locates all leaf node voxels that intersect with or are contained within the transient action.

[0107] All locked leaf node voxels constitute a target voxel set, which represents the target voxel set. The precise spatial extent of the impact of a moment's operation or detection on the physical artifact. This process can be represented as:

[0108] ;

[0109] In the formula, for The target voxel set at time; The instantaneous action body calculated by S310; The sparse voxel octree constructed in S1; This is a spatial intersection query function performed on a sparse voxel octree.

[0110] In step S330, which involves standardizing and encapsulating time-series attribute data packets, the system integrates and standardizes the engineering data synchronously collected in S2. The system then obtains the corresponding data from S2. The process flow data and physical property sensing flow data at each moment are collected and encapsulated with a unified high-precision timestamp assigned by S2 and a unique identifier of the data source device. This results in a structured time-series attribute data packet. This data packet is the smallest file unit to be written into the voxel time-series attribute list in S1. The structure of this data packet can be represented as follows:

[0111] ;

[0112] In the formula, for Standardized time-series attribute data packets; for A high-precision timestamp of a moment; A unique identifier for the data source device; For data flow type (process or physical property); The actual data collected (parameter values ​​or signal arrays).

[0113] After completing steps S310, S320, and S330, the system achieves precise spatiotemporal anchoring of engineering operation and physical property testing data, and standardizes these anchored data units into time-series attribute data packets. This target voxel set... and data packets The data used in step S4 is written as the direct input to the causal relationship marker.

[0114] Please see the appendix Figure 5 Step S4 is the data storage and association stage of archive construction. Its core purpose is to store the time-series attribute data packets anchored and encapsulated in S3 into the voxel archive constructed in S1 in a way that preserves history, and automatically establish the inherent logical relationship between the conservation project process data and the cultural relic status data, so as to realize the coupling of process and status.

[0115] In one embodiment, step S4 can be divided into two sub-steps: the appending and writing step of the timing attribute list (S410) and the generation step of causal relationship tags (S420).

[0116] In step S410, which appends the timing attribute list, the system processes the timing attribute data packet and the target voxel set generated in step S3. The system iterates through each leaf node voxel in the target voxel set. For each voxel, the system accesses its timing attribute list initialized in step S1.

[0117] Crucially, the system performs a data append operation, which completely adds the time-series attribute data packet to the end of the list. This operation differs from data overwriting, meaning it does not replace existing data in the list. This append mechanism is the core of building a traceable history archive, ensuring that all engineering operations and state changes experienced by this spatial location (voxel) are permanently preserved. This append operation can be represented as:

[0118] ;

[0119] In the formula, leaf node voxels A list of timing attributes; Leaf node voxels; The data append function adds new data packets to the end of the list; for The timing attribute data packet at any given moment; for The target voxel set at time.

[0120] In step S420, which generates causal relationship tags, the system activates a causal relationship tagging unit simultaneously with or after the write operation in S410. This unit monitors the type of the data packet being written (from the data stream type field in the timing attribute data packet).

[0121] The causal relationship labeling unit defines a spatiotemporal proximity determination condition: spatially, they are voxel clusters consisting of the same voxel or its neighboring voxels; temporally, the difference in timestamps between the two data packets is less than a preset time threshold. When the system detects that an engineering process data packet (as the cause of the association) and a physical property sensing data packet (as the result of the association) are successively written within this spatiotemporal proximity window, the system determines that there is a potential causal relationship between the two.

[0122] At this point, the system automatically generates a causal relationship marker to link the two data packets at the data structure level. The structure of this marker can be defined as follows:

[0123] ;

[0124] In the formula, Label the generated causal relationships; This is the unique identifier for the tag; The identifier for the engineering process data packet that serves as the cause; This is the identifier for the resulting physical property sensing data packet.

[0125] The above-generated causal relationship tags It is stored in a separate causal index table, or stored as metadata in the relevant voxel, for the S5 step to perform causal query.

[0126] After completing steps S410 and S420, the sparse voxel octree initialized in S1 has been dynamically filled with data collected in S3. Its time-series attribute list stores the complete operation history. At the same time, logical coupling between data is established through causal relationship marking. This makes the archive not only a static data collection, but also a structured foundation for S5 to perform multidimensional queries and analysis.

[0127] Please see the appendix Figure 6 Step S5 is the application and interaction stage of the digital archive. Its core purpose is to provide users with three different dimensions of highly intuitive query and visualization methods by using the initial archive structure built in S1, the data written in S4, and the generated causal relationships, so as to achieve in-depth analysis and evaluation of the entire process of cultural relic protection projects.

[0128] In one embodiment, step S5 can be divided into three sub-steps: attribute texture fusion visualization step (S510), historical data backtracking step based on surface index (S520), and causal relationship correlation analysis step (S530).

[0129] In the attribute texture fusion visualization step (S510), the system responds to the user's attribute-based query request, such as querying the spatial distribution of a physical quantity at a specific point in time or within a time period.

[0130] First, the system parses the user query and generates a filter for time-series attribute data packets.

[0131] Subsequently, the system traverses the sparse voxel octree constructed in S1, checks the temporal attribute list of each leaf node voxel, and extracts the corresponding attribute values ​​of voxels that meet the filter conditions to form a result voxel set.

[0132] Next, for each voxel in the resulting voxel set, the system calls the forward mapping constructed in S1 to query its corresponding UV coordinates on the high-resolution texture of the reference surface mesh.

[0133] Finally, based on the voxel's attribute values, the system calculates a color value using a preset colormap function, and dynamically renders or writes this color value to the corresponding UV coordinate region on the texture map. This process avoids the blocky effect of directly rendering voxels, achieving data fusion and visualization on a smooth, high-precision geometric model. The generation process of this texture color value can be represented as:

[0134] ;

[0135] In the formula, UV coordinates on texture map The color value at that location; This is a color mapping function; Leaf node voxels that satisfy the query conditions The attribute value; voxels defined in S1 from leaf nodes The positive mapping to its corresponding UV coordinates.

[0136] In the historical data backtracking step based on surface index (S520), the system responds to the user's query for a specific location on the surface of the cultural relic.

[0137] When a user specifies a query point on the reference surface mesh of S1 using an interactive device (such as a mouse), the system first determines the triangular facet containing that point. Subsequently, the system uses the reverse mapping constructed in S1, with that triangular facet as an index, to directly query and retrieve a list of all leaf node voxels associated with that facet, without traversing the entire octree.

[0138] Finally, the system extracts a complete temporal attribute list of all leaf node voxels in the list, sorts and integrates the data packets in these lists by timestamp, and presents them to the user in the form of a timeline. This method enables instantaneous access from any surface point to its entire historical record (including all engineering operations and physical property sensing data). The process of obtaining this historical record set can be represented as follows:

[0139] ;

[0140] In the formula, Selected triangular facets for the user The corresponding complete historical record set; The triangular facet defined in S1 The reverse mapping to its associated voxel list; leaf node voxels A list of timing attributes; This indicates that the data packets in all associated leaf node voxel lists are merged.

[0141] In the causal relationship analysis step (S530), the system responds to the user's exploratory query regarding the relationship between operation results. When the user wishes to analyze the impact of a certain type of engineering operation (e.g., laser cleaning with specific parameters) on the condition of a cultural relic (e.g., surface cleanliness or internal moisture content), the system first parses the query to form filtering conditions for the engineering process data package.

[0142] Subsequently, the system queries the causal relationship index table generated in S4 to find all causal relationship tags in the cause part (i.e., the engineering process data package) that meet the user's filtering conditions. For each matching tag, the system extracts the identifier of its linked result part (i.e., the physical property sensing data package).

[0143] Finally, the system extracts and displays these paired engineering process data packets and physical property sensing data packets side-by-side, providing users with direct, data-driven quantitative analysis of the effectiveness of cultural relic protection. The query result of this causal data pair can be represented as:

[0144] ;

[0145] In the formula, A set of causal data pairs that satisfy the conditions; For all users who meet the filtering criteria Engineering process data package A set; For all engineering-property data packages linked by causality tags A set of.

Claims

1. A method for constructing a three-dimensional visual digital archive of a cultural heritage protection project, characterized in that, Includes the following steps: S1: Digital archive initialization and bidirectional index construction, generating a reference surface mesh and a sparse voxel octree, and constructing a reference mesh and voxel bidirectional index between the reference surface mesh and the sparse voxel octree. The leaf node voxels of the sparse voxel octree are initialized with a temporal attribute list. S2: The spatiotemporal calibration of engineering equipment and digital archives is synchronized with the multimodal data stream to establish a real-time spatial transformation relationship between the tools and non-destructive testing sensors and the global coordinate system of the sparse voxel octree, and to assign a unified timestamp to the collected engineering process flow and physical property sensing flow. S3: Accompanying acquisition of digital archive data based on spatiotemporal anchoring, using the real-time spatial transformation relationship to calculate the instantaneous action volume, using the instantaneous action volume to query the sparse voxel octree to index the target voxel set, and encapsulating it to generate a time-series attribute data package; S4: Writing the temporal attributes of digital archives and marking causal relationships. The temporal attribute data packet is written into the temporal attribute list of the voxels in the target voxel set by appending data. When the engineering process data packet and the physical property sensing data packet meet the spatiotemporal proximity determination condition, a causal relationship mark is generated. S5: Digital archives integrate visualization and multidimensional querying, utilizing the baseline grid and voxel bidirectional index and the causal relationship markers to perform visualization queries, spatial penetration queries and causal relationship queries.

2. The method according to claim 1, wherein, The baseline grid and voxel bidirectional index include: A positive mapping from the leaf node voxel to the surface texture coordinates of the reference surface mesh; A reverse mapping from the surface triangular facets of the reference surface mesh to the voxel list of the leaf node voxels.

3. The method according to claim 1, wherein, The file initialization and bidirectional index construction steps in step S1 specifically include: The reference surface mesh is generated by acquiring raw data through 3D scanning; The sparse voxel octree is constructed based on the bounding space of the reference surface mesh, and the temporal attribute list is initialized for each leaf node voxel of the sparse voxel octree.

4. The method according to claim 1, wherein, The spatiotemporal calibration and multimodal data stream synchronization steps in step S2 specifically include: Install a six-degree-of-freedom spatial tracker on the tool and the non-destructive testing sensor; The real-time spatial transformation relationship between the tool, the non-destructive testing sensor, and the global coordinate system is established using a hand-eye calibration method.

5. The method of claim 1, wherein, The data acquisition step based on spatiotemporal anchoring in step S3 specifically includes: The instantaneous action volume is calculated using the real-time spatial transformation relationship and the predefined local action volume; The transient actor is used to perform a spatial intersection query in the sparse voxel octree to index the target voxel set.

6. The method of claim 1, wherein, In step S4, the timing attributes are written into the causal relationship marker. The data appending method is to add the timing attribute data packet to the end of the timing attribute list to preserve the historical process, which is different from data overwriting.

7. The method according to claim 1, wherein, The spatiotemporal proximity determination criteria include: A voxel cluster consisting of the same voxel or adjacent voxels in space; In terms of time, the difference between the timestamps of the engineering process data packet and the physical property sensing data packet is less than a preset time threshold.

8. The method of claim 1, wherein, In the S5 step of fusion visualization and multidimensional query, the visualization query includes: The sparse voxel octree is searched to obtain voxels that meet the conditions; Using the forward mapping in the reference mesh and the voxel bidirectional index, an attribute heatmap is dynamically rendered on the high-resolution texture of the reference surface mesh.

9. The method of claim 1, wherein, In the S5 step of fusion visualization and multidimensional query, the spatial penetration query includes: In response to a user's query point on the reference surface grid, determine the triangular facet where the query point is located; Using the inverse mapping in the reference mesh and the voxel bidirectional index, the voxel list associated with the triangular facet is indexed; the temporal attribute list corresponding to the voxel list is extracted and displayed.

10. The method of claim 1, wherein, In the S5 step of fusion visualization and multidimensional query, the causal relationship query includes: Retrieve the causal relationship markers generated in S4; The system displays to the user the engineering process data package and the physical property sensing data package that are linked by the causal relationship.

Images