A real-scene 3D data management system based on cloud computing

Abstract

The present invention relates to the field of data management technology, specifically a real-life three-dimensional data management system based on cloud computing, the system including a model verification module, a hierarchy generation module, a label collection module, a permission configuration module and a request sorting module. The present invention, through the calculation of connectivity based on boundary segments and the analysis of directional cosine similarity, can accurately identify problem points with closure defects or directional mutations in component boundaries, and build a consistency index set based on this, thereby improving the accuracy of model integrity judgment. In combination with facet structure parameters and weighted hierarchical evaluation methods, it can accurately delineate the level of regional detail levels, realize spatially targeted differential expression of detail level configuration, perform priority calculation based on the number of access requests and functional task matching results, improve component call efficiency and rendering response performance, and ensure service load optimization under multi-terminal access conditions and the rationality of cloud resource scheduling.

Description

Technical Field

[0001] The present invention relates to the technical field of data management, and in particular to a real-scene three-dimensional data management system based on cloud computing. Background Art

[0002] The field of data management technology involves systematic solutions for the collection, organization, storage, retrieval, sharing, and maintenance of data, aiming to improve data availability, consistency, security, and lifecycle control capabilities. This field covers data modeling, database management systems, data warehouses, metadata management, data quality control, data access control, and the formulation and implementation of data governance strategies. Especially in the context of the coexistence of complex heterogeneous data sources and the growing demand for multi-source data fusion, data management systems need to have high concurrent processing capabilities, data consistency assurance mechanisms, distributed storage architecture support capabilities, and scalable data structure organization forms. In addition, for data types with distinct structural characteristics such as spatial data and graph data, targeted indexing mechanisms and compression encoding methods are also required to support efficient spatial relationship queries and large-scale data visualization needs.

[0003] The real-world 3D data management system is used to process geospatial 3D model data generated through methods such as oblique photogrammetry, LiDAR scanning, and multi-source image modeling. Its purpose is to efficiently store, quickly load, meticulously organize, index, and manage real-world 3D models. It supports functional requirements such as model retrieval, spatial analysis, real-time rendering, and business overlays. The system typically includes a model data partitioning mechanism, a Level of Dimension (LOD) hierarchy, spatial indexing, a metadata system, and data permission control, serving a variety of practical scenarios, including digital twin cities, smart planning, emergency response, and engineering construction.

[0004] When traditional management systems are faced with real-life 3D models generated by oblique photogrammetry or laser scanning, boundary closure defects or structural anomalies are often not effectively identified. This is especially true in areas with high component complexity or accumulated modeling errors. Existing indexing mechanisms lack the ability to analyze boundary connectivity and directional consistency, resulting in models being prone to surface breakage and misalignment during loading or rendering. Furthermore, because access permission configuration fails to integrate component functions and spatial hierarchies, the granularity of user requests cannot be effectively restricted, resulting in redundant loaded data and increasing pressure on bandwidth and computing resources. This problem is particularly pronounced in scenarios with concurrent access by multiple users, impacting overall system response efficiency and resource utilization. Summary of the Invention

[0005] The purpose of the present invention is to solve the shortcomings of the existing technology and propose a real-scene three-dimensional data management system based on cloud computing.

[0006] In order to achieve the above-mentioned object, the present invention adopts the following technical solution: a real-scene three-dimensional data management system based on cloud computing, the system comprising:

[0007] The model verification module obtains the boundary line segment set of the real-life 3D component submitted to the cloud service platform by the access terminal, filters the endpoints of the line segments with sudden changes in boundary direction, integrates the set of unclosed points and the information of nodes with sudden changes in direction, and generates a set of 3D boundary consistency indicators on the cloud.

[0008] The hierarchical generation module reads the number of face edges, the normal vector angle, and the adjacent face angle from the cloud parsing task based on the cloud 3D boundary consistency index set, calculates the structural difference score, assigns the corresponding LOD detail level to each area, and generates a LOD structure configuration list for the real scene area;

[0009] The label collection module calls the LOD structure configuration list of the real scene area, establishes a label dictionary mapping relationship table according to the area number, and encodes and binds the label value with the component path to generate a cloud component label index table;

[0010] The permission configuration module identifies the path ID range of components accessible to each type of access end user based on the cloud component tag index table and the access end access request and access permission of the current cloud component, and establishes a user access component path list.

[0011] The present invention has the following improvements: the cloud-based three-dimensional boundary consistency index set includes a boundary closure label, a direction consistency identifier, an abnormal boundary record, and a component space index code; the real-scene area LOD structure configuration list includes an LOD classification identification code, a reconstruction area code, a detail density parameter, and an area grid group ID; the cloud-based component label index table specifically includes a function class label set, a location class label set, a layer type code, and a component index key-value pair; and the user access component path list includes a component access path ID set, an authority level label, a role category identifier, and a path authorization mapping table.

[0012] The present invention is improved in that the model verification module includes:

[0013] The boundary extraction submodule obtains the boundary line segment set of the real-life 3D component submitted by the access end to the cloud service platform, extracts the coordinates of the two ends of each line segment to construct a vertex list, counts the line segment connection values ​​corresponding to the vertices, calls the boundary connectivity threshold and compares it with the vertex connection value to determine whether there are boundary anomalies with low connection counts, records the coordinates of vertices with connection counts less than the threshold and establishes a vertex number index to generate a boundary closure anomaly point set;

[0014] The structure identification submodule obtains the direction vector value of the associated boundary segment based on the boundary closure anomaly point set and the spatial position coordinates of the corresponding vertex. Adjacent vectors are extracted in pairs according to the segment connection order, and the cosine values ​​of the angles between adjacent vectors are calculated to obtain the direction cosine values ​​at the connection points. The direction cosine values ​​are compared with the direction cosine similarity threshold, and the node groups with prominent direction offsets are screened. The direction cosine deviation degree value is obtained by calculation to determine whether it exceeds the cosine offset set threshold. If the conditions are met, it is marked as a structural direction anomaly point and a direction anomaly identification list is established.

[0015] The consistency integration submodule calls the boundary closure anomaly point set and the direction anomaly identification list, merges the corresponding points according to the component number, constructs an anomaly coordinate set based on the component structural unit, integrates the index value and anomaly type label of the point in the set, and establishes a cloud-based three-dimensional boundary consistency indicator set.

[0016] The present invention is improved in that the hierarchy generation module includes:

[0017] The structure acquisition submodule obtains the 3D mesh structure data of the model facets in the area based on the three-dimensional boundary consistency index set in the cloud and the spatial positioning coordinates, extracts the number of edges, facet normal vector angle, and adjacent facet angle values ​​of each facet from the cloud parsing task, and establishes a facet geometric feature parameter table;

[0018] The difference calculation submodule is based on the facet geometric feature parameter table, and according to the number of edges of each facet, the angle between the facet normal vectors, the angle between adjacent faces, and the corresponding position index number, calls the structural layering control parameter to perform a scoring calculation, calculates the structural difference score value, determines the relative variation range of the facet geometric characteristics within the same area, and generates a structural difference score list;

[0019] The level allocation submodule calls the structural difference score list, performs segmented judgment on the score value and the set LOD detail level threshold, assigns an LOD detail level identifier to each spatial area face according to the score range, and establishes a real scene area LOD structure configuration list.

[0020] The present invention is improved in that the label collection module includes:

[0021] The regional parsing submodule calls the LOD structure configuration list of the real scene area, obtains the cloud component storage path information according to the area number and LOD level identification item, extracts the three fields of the corresponding component: building function label, geospatial location code and layer attribute, merges them according to the area number, and establishes a component label data set;

[0022] The label construction submodule extracts the function type name, spatial coordinate code and layer flag value based on the component label dataset and the area number index field and the three label field contents, combines the three data contents to construct a multi-field index label, and generates a unique value code for each combination to generate a component label code value list;

[0023] The path binding submodule calls the component label code value list, establishes a key-value binding table of component path and label code according to the path index identifier corresponding to the code item and the component storage path, constructs a mapping structure from label code to component path, and establishes a cloud component label index table.

[0024] The present invention is improved in that the permission configuration module includes:

[0025] The permission reading submodule is based on the cloud component tag index table, according to the building function tag items, according to the correspondence between the function tag items and the component type numbers, calls the access control list in the cloud service platform, filters the access permission level values ​​corresponding to the function tag matching items, and binds the results to the component type index structure to generate a component permission level list;

[0026] The access identification submodule calls the component permission level list, extracts the user role identifier based on the access request information of the current cloud component, performs a corresponding judgment process on the role identifier and the permission level, outputs the access component number interval that each type of access terminal user has within the current permission range, and generates role access number interval information;

[0027] The path screening submodule traces back the corresponding path ID value of the index component in the cloud path table according to the role access number interval information, screens the path ID and number matching results, collects the accessible path index data according to the user role, and establishes a user access component path list.

[0028] The present invention is improved in that the system further comprises:

[0029] The request sorting module calls the user access component path list, extracts the business function parameters in the access request, matches them with the task priority configuration table built into the cloud platform, performs a weighted evaluation based on the access frequency and task priority value, sets the loading order based on the score, and generates a three-dimensional loading priority path set on the cloud;

[0030] The cloud-based three-dimensional loading priority path set includes a loading priority number, a regional access hotspot index, a task urgency level code, and a path scheduling index set.

[0031] The present invention is improved in that the request sorting module includes:

[0032] The path extraction submodule calls the user access component path list, obtains the geospatial location code associated with each path, extracts the cumulative request count value of the component path ID in the access log, merges and records the request path and access frequency after merging by path ID, and generates a path request frequency distribution value;

[0033] The task evaluation submodule extracts the business function parameters submitted by the user based on the path request frequency distribution value, compares them with the task level code in the cloud service task priority configuration table, obtains the task function level value and role identity information value of each path, calls the regional heat level item corresponding to the path spatial location, calculates and obtains the path comprehensive priority score value, determines the ranking level of the component path in the multi-path scheduling system, and generates a path priority score sequence;

[0034] The priority allocation submodule calls the path priority score sequence, sorts the paths in descending order according to the priority score value corresponding to each path ID, writes the sorting results into the cache loading task queue according to the path index table, and establishes a cloud-based 3D loading priority path set.

[0035] Compared with the prior art, the advantages and positive effects of the present invention are:

[0036] In the present invention, through the connection calculation based on boundary segments and the direction cosine similarity analysis, it is possible to accurately identify problem points with closure defects or directional mutations in the component boundaries, and based on this, a consistency index set is constructed to improve the accuracy of model integrity judgment. Combined with the patch structure parameters and the weighted hierarchical evaluation method, it is possible to accurately delineate the regional detail level and achieve spatial targeted differential expression of detail level configuration. The label index is established by combining the building function labels, location codes and layer attributes extracted from the component path, and the accurate execution of subsequent access control logic is optimized. Priority calculation is performed based on the number of access requests and the matching results of functional tasks, supporting dynamic sorting of component access requests and loading priority strategy arrangement in high-frequency areas, improving component call efficiency and rendering response performance, and ensuring service load optimization under multi-terminal access conditions and the rationality of cloud resource scheduling. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 is a system flow chart of the present invention;

[0038] Figure 2 This is a flow chart of the model verification module of the present invention;

[0039] Figure 3 A flowchart of the hierarchical generation module of the present invention;

[0040] Figure 4 This is a flowchart of the label collection module of the present invention;

[0041] Figure 5 This is a flowchart of the permission configuration module of the present invention;

[0042] Figure 6 Flowchart of the request sorting module of the present invention. DETAILED DESCRIPTION

[0043] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0044] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings and are only for the convenience of describing the present invention and simplifying the description. They do not indicate or imply that the devices or elements referred to must have a specific direction, be constructed and operate in a specific direction, and therefore should not be understood as limiting the present invention. In addition, in the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0045] See also Figure 1 The present invention provides a technical solution: a real-scene three-dimensional data management system based on cloud computing, the system comprising:

[0046] The model verification module obtains the boundary line segment set of the real-life 3D component submitted by the access end to the cloud service platform, extracts the coordinates of the two ends of the line segment to form a vertex group, counts the number of line segments connected to the vertex, calls the boundary connectivity threshold set in the stitching requirement, compares the difference between the boundary connectivity threshold and the vertex connection value, records the vertex coordinates below the connection threshold, constructs an unclosed point set, obtains the direction vector value of each boundary line segment, calculates the angle difference of the direction vector according to the segment connection order, and judges the angle difference with the direction cosine similarity threshold. It screens the endpoints of the line segments with sudden changes in boundary direction, integrates the unclosed point set and the direction mutation node information, and generates a cloud-based 3D boundary consistency indicator set.

[0047] The boundary connectivity threshold is used to determine whether a vertex is a closed structure point, which comes from the definition of degree in graph theory; the direction cosine similarity threshold is used to determine whether adjacent edges are in the same direction;

[0048] The hierarchical generation module, based on the cloud-based 3D boundary consistency indicator set, obtains the model face mesh structure data of the corresponding area according to the spatial positioning coordinates. It reads the three parameters of the face edge number, normal vector angle, and adjacent face angle from the cloud-based parsing task, calls the regional structure hierarchical control parameters for weighted evaluation, calculates the structural difference score, compares the score value with the detail level setting threshold, assigns the corresponding LOD detail level to each area, and generates the real-life area LOD structure configuration list.

[0049] The label collection module calls the LOD structure configuration list of the real scene area. Based on each area number and LOD level identification item, combined with the three parameters of building function label, geospatial location code and layer attribute read from the component cloud storage path, it establishes a label dictionary mapping relationship table by area number, and encodes and binds the label value with the component path to generate a cloud component label index table.

[0050] Building function labels are component classifications defined in international building information modeling standards, such as IFCWall, IFCSlab, and IFCDoor. Geospatial location coding is a geographic coding method that converts longitude and latitude into string codes for easy spatial positioning and indexing. Layer attributes in GIS and BIM systems refer to the semantic level of a component, such as the structural layer, water and electricity layer, etc.

[0051] The permission configuration module is based on the cloud component tag index table and the building function tag item, and obtains the access permission level of the corresponding component type in the access control list set in the cloud service platform. It then performs parallel screening based on the permission level and the LOD level value in the tag. Based on the access request and access permission of the current cloud component access end, it identifies the path ID range of the component that can be accessed by each type of access end user and establishes a user access component path list.

[0052] The request sorting module calls the user's access component path list. Based on the user's target component path ID, it obtains the geospatial location code of the path corresponding to the area and the number of access requests in the log. It extracts the business function parameters in the access request and matches them with the task priority configuration table built into the cloud platform. It performs a weighted evaluation of the access frequency and task priority value, sets the loading order by score, and generates a set of cloud-based 3D loading priority paths.

[0053] The cloud-based 3D boundary consistency indicator set includes boundary closure labels, direction consistency identifiers, abnormal boundary records and component space index codes. The real-scene area LOD structure configuration list includes LOD classification identification codes, reconstruction area codes, detail density parameters and regional grid group IDs. The cloud-based component label index table specifically includes functional label sets, location label sets, layer type codes and component index key-value pairs. The user access component path list includes component access path ID sets, permission level labels, role category identifiers and path authorization mapping tables. The cloud-based 3D loading priority path set includes loading priority numbers, regional access hotspot indicators, task urgency levels and path scheduling index sets.

[0054] See also Figure 2 , the model verification module includes:

[0055] The boundary extraction submodule obtains the boundary line segment set of the real-life 3D component submitted by the access end to the cloud service platform, extracts the coordinates of the two ends of each line segment to construct a vertex list, counts the line segment connection values ​​corresponding to the vertices, calls the boundary connectivity threshold and compares it with the vertex connection value to determine whether there are boundary anomalies with low connection counts, records the coordinates of vertices with connection counts less than the threshold and establishes a vertex number index to generate a boundary closure anomaly point set;

[0056] Obtain the boundary line segment set of the real-scene 3D component submitted to the cloud service platform by the access end, collect the 3D coordinate values ​​of the two ends of each boundary line segment in the component set, map each set of start and end coordinate data into a bidirectional edge vector, and add them to the line segment table in order. For example, a component contains 12 facets, and after boundary extraction, 36 boundary line segments are obtained. Each line segment has a starting point. ,end Build a line segment set, further build an index list for each vertex, scan all the endpoints of the line segments, count the number of times each vertex is referenced in the entire set of line segments, obtain the connectivity, and set the boundary connectivity threshold This value is based on the boundary closure requirement of the spatial geometric surface. If the number of connections of a vertex is less than 2, that is, the boundary of the component is not closed at this point, it is determined to be a boundary abnormal point. For example, it only appears in one line segment, and the connectivity is , should be considered as an abnormal point, its coordinates are , the component number is F007. According to the component structure layer classification, this point is marked as a closed abnormal point, added to the number index table, and classified as an abnormal point set.

[0057] Embed the table text as follows:

[0058] Table 1 Direction deviation judgment data table

[0059]

[0060] As shown in Table 1, the number of edges connected to node N2 is 1, which is lower than the connectivity threshold. , and thus classified into the boundary closure anomaly point set. Threshold The setting logic is based on the minimum edge connectivity requirement for three-dimensional space surfaces to form a closed area in the Euler topology rule. If the value is lower than this, it constitutes a geometric disconnected point, and the corresponding geometric body cannot be completely closed.

[0061] The structure recognition submodule is based on the boundary closure anomaly point set. According to the spatial position coordinates of the corresponding vertices, the direction vector value of the associated boundary segment is obtained. Adjacent vectors are extracted in pairs according to the segment connection order. The cosine value of the angle between adjacent vectors is calculated to obtain the direction cosine value at the connection point. The direction cosine value is compared with the direction cosine similarity threshold to screen the node group with prominent direction deviation. The formula is used:

[0062] ;

[0063] The direction cosine deviation value is obtained by calculation to determine whether it exceeds the cosine deviation threshold. If the condition is met, it is marked as a structural direction abnormal point and a direction abnormality identification list is established;

[0064] in, Indicates the degree of deviation of the direction cosine, reflecting the degree of imbalance of the directional continuity at the component node. Represents the cosine value of the theoretical direction vector between the starting edge and the adjacent edge, represents the theoretical angle between the starting edge and the adjacent edge, Indicates the normalized value of the modulus length of the starting edge direction vector, Represents the normalized value of the modulus length of the adjacent edge direction vector, Indicates the normalized value of the mesh density of the component where the current node is located, which is used to identify whether there is a sudden change in direction;

[0065] Based on the node numbering in the boundary closure outlier set, e.g. 、 Etc., through the component index mapping, query the associated line segment set, extract the direction vector coordinate difference of each line segment, calculate the unit vector and normalize the modulus value, and record it as the starting side modulus With adjacent side mold , such as node The direction vectors of the two adjacent sides are 、 , then:

[0066] ;

[0067] ;

[0068] ;

[0069] Node mesh density , set this parameter to the normalized statistical value of the average number of edges per node in the unit body component. Substitute the above parameters into the formula:

[0070] ;

[0071] Substitute the data:

[0072] ;

[0073] ;

[0074] ;

[0075] Denominator: ;

[0076] Overall expression: ;

[0077] Set the direction deviation threshold This value is set according to the national standard GB / T24356 3D assembly error limit design guidance document, which is the boundary direction continuity error limit, approximately radians The cosine value difference of . ,node Nodes that are judged to have abnormal directions are included in the abnormal direction identification list.

[0078] Formula structure analysis:

[0079] This formula uses the direction vector modular product The denominator of the square root of the sum of squares and logarithmic terms constitutes the directional coupling consistency evaluation standard. The nonlinear adjustment factor introduced is used to enhance the sensitivity of high-density grid areas to deviation perception, and the overall direction offset is processed by absolute value to finally form a unified dimension of abnormality index. The benefit of the formula is that by combining the directional product with the density adjustment term, the ability to identify boundary disturbances in local high-density areas is improved. This result shows that Nodes with significant directional deviations should be identified as structural directional anomalies, which are directly related to the integrity of the subsequent consistency indicator set.

[0080] The consistency integration submodule calls the boundary closure anomaly point set and the direction anomaly identification list, merges the corresponding points according to the component number, constructs an anomaly coordinate set based on the component structural unit, integrates the index value and anomaly type label of the point in the set, and establishes a cloud-based 3D boundary consistency indicator set;

[0081] The spatial coordinates of each node in the boundary closure anomaly point set and the direction anomaly identification list are called, and the numbers are mapped through the component structure list. Each anomaly point is assigned to the component index and merged. A subset of anomaly points is established under each component, and its anomaly type is marked as closure anomaly or direction anomaly. The type field is constructed and the corresponding point index is marked. After merging, it is packaged according to the component number to form an anomaly coordinate set structure. The number of indexes and anomaly classifications in the set are retained in the database to support subsequent annotation display and loading filtering logic. The overall structure composed of the mapping relationship between all component numbers and their corresponding anomaly points is the cloud-based three-dimensional boundary consistency indicator set. This structure records the information of all nodes in the space where geometric boundaries are disconnected and direction angles suddenly change, providing a panoramic index benchmark for subsequent LOD reconstruction and component availability verification.

[0082] See also Figure 3 , the hierarchy generation module includes:

[0083] The structure acquisition submodule obtains the 3D mesh structure data of the model facets in the area based on the spatial positioning coordinates based on the 3D boundary consistency index set in the cloud. It extracts the number of edges, the angle between the facet normal vectors, and the angle between adjacent faces of each facet from the cloud parsing task, and establishes a facet geometric feature parameter table.

[0084] Based on the spatial positioning coordinates in the cloud-based 3D boundary consistency index set, the 3D model facet data within the coordinate coverage range is extracted, the edge connection records of each facet are obtained from the structural topology relationship in the model file, and the number of edges of each facet is determined by counting the number of edge connection nodes. , and call the surface normal vector sampling module to obtain the surface normal vector information, select the unit normal vector direction of each face, perform angle calculation operation with its adjacent face normal vector, obtain the angle angle and then perform normalization to obtain , such as the angle between a face and its adjacent face is , normalized to the interval The result is In addition, the absolute value of the angle difference between each face and the adjacent face is extracted as , if the angles between the normal vectors of two adjacent faces are and ,but On this basis, according to the total number of facets in the area and the range of the system preset standard facet number, the number of facets in the area is standardized and defined as the normalized density index , if the system base is 100 facets, and if the current area contains 64 facets, then , summarize the above parameters into a structure array to construct a patch geometric feature parameter table;

[0085] The difference calculation submodule is based on the facet geometric feature parameter table. According to the number of edges of each facet, the angle between the face normal vectors, the angle between adjacent faces and the corresponding position index number, it calls the structural hierarchical control parameters to perform the scoring calculation. The formula is:

[0086] ;

[0087] Calculate and obtain the structural difference score value, determine the relative change range of the patch geometric characteristics in the same area, and generate a structural difference score list;

[0088] in, represents the structural difference score, Indicates the number of edges of the patch, represents the normalized value of the angle between adjacent faces, Represents the absolute value of the angle difference between the normal vectors of adjacent faces, A normalized value representing the number of patches in the region;

[0089] According to the number of edges of each patch extracted from the patch geometric feature parameter table , angle normalized value , normal vector difference And the normalized value of regional patch density , call the system structure hierarchical control logic and calculate the structural difference score using the following formula:

[0090] ;

[0091] Taking a certain patch as an example, the parameter settings are: , , , , the calculation process is as follows:

[0092] Numerator calculation: , , then the product term is , the total numerator is ;

[0093] Denominator calculation: , add 1 to the denominator and we get ;

[0094] The structural difference scores are: .

[0095] The formula structure is described as follows:

[0096] The numerator uses the square of the number of face edges coupled with the angle information to highlight the relationship between edge connection complexity and angle perturbation. The square root term amplifies the response degree of the angle difference. The denominator introduces the regional density normalization value. The logarithm is taken to buffer the molecular expansion caused by high-density areas and maintain the stability of the scoring standard. The formula is beneficial in that it improves the recognition weight of structural abnormal areas by combining square terms and square roots, and avoids misjudgment caused by disturbance amplification in dense areas by using logarithmic adjustment terms. This result shows that the structural score value The image is already above the first-level detail retention range defined by the system. You need to configure higher-level LOD detail parameters and record them in the structural difference score list.

[0097] The level allocation submodule calls the structural difference score list, performs segmented judgment on the score value and the set LOD detail level threshold, assigns LOD detail level identifiers to each spatial area face according to the score range, and establishes the LOD structure configuration list of the real scene area;

[0098] Call the score value of each patch recorded in the structural difference score list , according to the LOD grading judgment criteria set by the system, perform the scoring interval segmentation operation, and classify each face into the level of detail within the matching level range. The grading intervals are set as follows: LOD-3, LOD-2, For LOD-1, high scores correspond to high precision requirements, such as the aforementioned patch score value , classified into the LOD-1 interval, logically aggregate all face level identifiers in the same spatial range according to regional division, establish a mapping structure between the face number and the corresponding LOD level in the component area, and generate a real-scene area LOD structure configuration list to drive subsequent data loading scheduling and component layered rendering operations.

[0099] See also Figure 4 , the label collection module includes:

[0100] The regional parsing submodule calls the LOD structure configuration list of the real-life region, obtains the cloud component storage path information based on the region number and LOD level identification item, extracts the corresponding component's building function label, geospatial location code, and layer attribute fields, merges them by region number, and establishes a component label dataset;

[0101] The region ID and LOD level identifier items in the real-world region LOD structure configuration list are retrieved. The region zone ID field stored in the list is parsed, and the corresponding component's cloud path ID value is retrieved using the ID index. The three component attribute fields contained in the path data file are extracted: the building function tag field, the geospatial location code field, and the layer attribute field. The function tag field typically represents the function type as a string, such as "residential," "commercial," or "underground parking." The geospatial location code field represents the region's center point as a standard latitude and longitude number or a six-digit grid code. The layer attribute field represents the layer depth or layer type as an integer, for example, ground level is set to 0, underground is -1, and aerial components are +1. For example, from the path " / zone32 / lod1 / component_158.fbx," the building function is extracted as "commercial," the geospatial location code is "JX3276E," and the layer attribute value is +1. After field extraction, components with the same ID are merged into a single data record, creating a component label dataset based on the region dimension.

[0102] The label construction submodule is based on the component label dataset. According to the area number index field and the three label field contents, it extracts the function type name, spatial coordinate code and layer flag value respectively. It combines the three data contents to construct a multi-field index label, generates a unique value code for each combination, and generates a component label code value list.

[0103] Based on the area number field and three attribute tag fields of each component in the component tag dataset, the area number, function type name, spatial coordinate code, and layer identifier values ​​are sequentially read as the function attribute, spatial location information, and structural layer identifier fields, respectively. Multi-field index tags are constructed through combination. This combination uses a sequential concatenation method, with the order "function type_coordinate code_layer value." For example, if a component's function is "residential," its location code is "HZ4162N," and its layer identifier is "0," the multi-field combination result is "residential_HZ4162N_0." A standard hash mapping method is then used to generate unique values ​​for each combination of fields. The encoding process uses MD5 or SHA256 as the standard hashing algorithm. Each tag combination generates a unique hash code to ensure that tag codes do not conflict within the same area. For example, the SHA256 encrypted unique value of the above example tag combination is "94D95B3A73B..." (the first 14 digits are truncated for display). Each tag combination and its encoded value are written into an index record to generate a list of component tag encoding values.

[0104] The path binding submodule calls the component label code value list, establishes a key-value binding table of component path and label code according to the path index identifier corresponding to the code item and the component storage path, constructs a mapping structure from label code to component path, and establishes a cloud component label index table;

[0105] Call each record in the component tag code value list, read the tag unique code field, and match the path index identifier in the corresponding component path in sequence. The path index identifier is the file number or standard path ID extracted from the component path. For example, the path index identifier of the path " / zone28 / lod2 / component_347.obj" is "component_347". This identifier is mapped to the tag code field in a key-value pair to form a "tag code value → component path ID" structure. For example, the code "3EC1F4BC987..." is mapped to "component_347". This binding record is used to reversely resolve the correspondence between the tag and the storage path in subsequent queries. Finally, all binding records are written to the key-value mapping structure to establish a cloud-based component tag index table for cloud platform tag-driven component loading management, permission control, or layer retrieval tasks.

[0106] See also Figure 5 , the permission configuration module includes:

[0107] The permission reading submodule is based on the cloud component tag index table. According to the building function tag items and the correspondence between the function tag items and the component type numbers, it calls the access control list in the cloud service platform, filters the access permission level values ​​corresponding to the function tag matching items, and binds the results to the component type index structure to generate a component permission level list;

[0108] To retrieve building function tags from the cloud-based component tag index table, the component path is first used as the query key to extract the corresponding function tag value for each component. For example, the component with the path " / zone24 / lod1 / C456.fbx" has the tag "Industrial Plant." The tag "Industrial Plant" is then bound to the component type number "CT24" based on the component type number mapping table in the cloud platform's resource library. The system then invokes an access control list (ACL), which contains preset access permission levels for each function type. These levels are typically represented numerically, ranging from 1 to 5, with 1 representing the lowest level and 5 the highest. Based on the component function tag matching results, the permission level value set in the ACL is extracted. For example, "Industrial Plant" corresponds to level 3. This permission level value is used as a component access control constraint parameter and is bound to the component type number and function tag key-value structure to generate a unified component permission level list. Each record includes the function tag, component number, and permission level value. A typical record is as follows: "Industrial Plant, CT24, 3." If multiple components share the same tag, the shared level value is written into the permission mapping structure.

[0109] The access identification submodule calls the component permission level list, extracts the user role identifier based on the access request information of the current cloud component, matches the role identifier with the permission level, outputs the access component number range that each type of access user has within the current permission range, and generates role access number range information;

[0110] The system uses the component type index and permission level values ​​in the component permission level list. The system receives access request packets from the user access point, parses the user identity field in the packet, and extracts the user role identification field. The role field is typically a standard character label, such as "guest," "engineer," or "admin." The system searches the user permission configuration table to find the corresponding upper limit for each user role, such as 2 for "guest," 3 for "engineer," and 5 for "admin." The role permission level is then compared with the component permission level, and a conditional judgment is used to determine whether the role has permission to access the component. If the user permission level is greater than or equal to the component permission level, the user is allowed to access the corresponding component. The system performs this judgment on a component-by-component basis and outputs component number ranges that meet the criteria, such as "C450-C470" or "C100-C109." Each range indicates the corresponding user role. Ultimately, a role access number range is generated, which is used in the subsequent component access path screening stage.

[0111] The path screening submodule traces back the corresponding path ID value of the index component in the cloud path table based on the role access number interval information, filters the path ID and number matching results, collects the accessible path index data according to the user role, and establishes a user access component path list;

[0112] Based on the component number range specified in the role access number interval, the system performs a reverse component path index operation, tracing back to the cloud-based path index table in numerical order. The path number field matches the component number and extracts the path ID. If the number range is "C450-C470," the system searches the path table for all path ID entries containing the specified number. For example, the path " / zone24 / lod1 / C456.fbx" corresponds to the ID "P456" and selects it into the collection. To differentiate user access rights, the system archives the path index sets owned by different roles and constructs a role-based access path list structure. Examples of entries include "guest: [P101, P102]" and "engineer: [P150, P151, P456]." This list serves as the user's component access path list, guiding path determination and data loading during actual access control. It provides path ID-level precision control. Subsequent modules that require the path list for tag extraction or component loading sorting can directly access this list structure.

[0113] See also Figure 6 , the request sorting module includes:

[0114] The path extraction submodule calls the user's access component path list, obtains the geospatial location code associated with each path, extracts the cumulative request count value of the component path ID in the access log, merges and records the request path and access frequency after merging them by path ID, and generates the path request frequency distribution value;

[0115] After accessing the component path IDs from the user's access component path list, the system first reads the path index table and extracts the geospatial location code corresponding to each path ID. For example, the component with path ID "P321" corresponds to the spatial code "Z17-X4-Y2." The system then accesses the access log table to retrieve the cumulative number of accesses for each path ID within a specified time window. The system merges logs using a fixed sampling period (e.g., 24 hours) and combines repeated access records by path ID. Based on the number of records in the user access log, the system aggregates the path IDs. For example, path "P321" is recorded as accessed 32 times, path "P326" is recorded as accessed 8 times, and so on. A mapping between path ID and access frequency is constructed and numbered. All path frequencies are aggregated to form a complete data distribution. For example, the frequency distribution data might be: "P321: 32 times," "P326: 8 times," and "P330: 15 times." Finally, the path ID is associated with the access frequency to form a path frequency distribution structure, and the path request frequency distribution value is generated. This distribution is used for subsequent task level fusion judgment and priority evaluation.

[0116] The task evaluation submodule extracts the business function parameters submitted by the user based on the path request frequency distribution value, compares them with the task level code in the cloud service task priority configuration table, obtains the task function level value and role identity information value of each path, and calls the regional heat level item corresponding to the path spatial location using the formula:

[0117] ;

[0118] Calculate and obtain the comprehensive priority score of the path, determine the ranking level of the component path in the multi-path scheduling system, and generate a path priority score sequence;

[0119] in, Indicates the comprehensive priority score of the path, Indicates the task level value, which comes from the task priority configuration table. Indicates the path access frequency value, which comes from the path access log record. Indicates the geographical heat level value of the area where the path is located, which is derived from the cumulative access index of the access space area. Indicates the role permission level value of the user requesting access, which comes from the permission configuration module record. Indicates the normalized preloading priority value of the path in the global cache list, which is used to measure the storage priority of the path in the system cache system;

[0120] According to the path request frequency distribution value, the system reads the access frequency data of each path. For example, the access frequency of "P321" is 32 times. At the same time, the business request parameter field of the user to which the path ID belongs is retrieved, and the parameter value is "modeling scheduling". In the task priority configuration table, find the task level value corresponding to the task type "modeling scheduling" and find it is 4. The system extracts the user role level value "engineer", and the corresponding permission level value is 3. At the same time, the regional heat level value of the path spatial location "Z17-X4-Y2" is obtained from the regional heat table, which is 5. The preload level corresponding to this path in the system cache priority list is 0.65. Call all the above values ​​and substitute them into the following formula:

[0121] ;

[0122] The path "P321" has a comprehensive priority score of 6.759. The system scores all paths one by one in this way. This score represents the priority index of the path in scheduling and loading after the fusion of multiple factors, reflecting the fusion effect of access frequency, task importance, regional popularity and cache scheduling weight. The higher the score, the more priority it needs to be cached. The formula is beneficial in that by introducing Enhanced sensitivity to frequency changes, Construct access role and region heat difference items to penalize heat mismatch scenarios. The impact of cache structure on path scheduling efficiency is incorporated to achieve multi-factor balanced fusion of path priority.

[0123] in, Indicates the comprehensive priority score of the path, Indicates the task level value, Indicates the path access frequency value, Indicates the geographic heat level value of the area where the path is located. Indicates the role permission level value of the user requesting access. Represents the normalized preload priority of a path in the global cache list. The multiplication function in the formula structure enhances the linkage between task level and frequency. The denominator deviation term acts as a negative factor to penalize inconsistencies between roles and popularity. The square and root signs are used to integrate differences in multiple dimensions into a unified indicator. The following lists the parameter values ​​for different paths in the sample test:

[0124] Table 2 Example of path comprehensive priority scoring parameters

[0125]

[0126] As shown in Table 2, path "P321" scored 6.759, path "P326" scored approximately 3.728, and path "P330" scored approximately 4.948. This result indicates that P321 has a significantly higher priority among similar tasks, making it suitable for scheduling and preloading into the cache, providing a ranking basis for subsequent construction of the scheduling sequence.

[0127] The priority allocation submodule calls the path priority score sequence, sorts the paths in descending order according to the priority score value corresponding to each path ID, writes the sorting results into the cache loading task queue according to the path index table, and establishes a cloud-based 3D loading priority path set;

[0128] After invoking the path priority scoring sequence, the system sorts each path ID based on its score, ranking them from highest to lowest, such as "P321: 6.759," "P330: 4.948," and "P326: 3.728." This sorting structure serves as the task enqueue index for cache scheduling management. Based on the sorting results, the system generates cache load queue task instructions, synchronously writing each path into the cache load table based on its score and index number. These instructions are assigned a queue index and marked with the loading priority for the current scheduling cycle. The queue structure takes the form of a mapping table between path ID and score, with each record structured as: [path ID, priority score, queue index], for example, "P321, 6.759, Q1." Finally, the system outputs a scheduling ranking structure for all paths, constructing a cloud-based 3D loading priority path set, which provides a scheduling basis for subsequent scene rendering and loading processes. During dynamic access, the system can call this path set structure in real time to rapidly complete cache replacement and prioritized loading operations.

[0129] The above are merely preferred embodiments of the present invention and do not limit the present invention in any other form. Any technician familiar with the profession may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes and apply it to other fields. However, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall still fall within the scope of protection of the technical solution of the present invention.

Claims

1. A real-scene 3D data management system based on cloud computing, characterized in that: The system comprises: The model verification module obtains the boundary line segment set of the real-life 3D component submitted to the cloud service platform by the access terminal, filters the endpoints of the line segments with sudden changes in boundary direction, integrates the set of unclosed points and the information of nodes with sudden changes in direction, and generates a set of 3D boundary consistency indicators on the cloud. The hierarchical generation module reads the number of face edges, the normal vector angle, and the adjacent face angle from the cloud parsing task based on the cloud 3D boundary consistency index set, calculates the structural difference score, assigns the corresponding LOD detail level to each area, and generates a LOD structure configuration list for the real scene area; The label collection module calls the LOD structure configuration list of the real scene area, establishes a label dictionary mapping relationship table according to the area number, and encodes and binds the label value with the component path to generate a cloud component label index table; The permission configuration module identifies the path ID range of the component that each type of access end user can access based on the cloud component tag index table and the access end access request and access permission of the current cloud component, and establishes a user access component path list; The request sorting module calls the user access component path list, extracts the business function parameters in the access request, matches them with the task priority configuration table built into the cloud platform, performs a weighted evaluation based on the access frequency and task priority value, sets the loading order based on the score, and generates a three-dimensional loading priority path set on the cloud; The cloud-based three-dimensional boundary consistency index set includes boundary closure labels, direction consistency identifiers, abnormal boundary records and component space index codes. The real-scene area LOD structure configuration list includes LOD classification identification codes, reconstruction area codes, detail density parameters and regional grid group IDs. The cloud-based component label index table specifically includes a function class label set, a location class label set, a layer type code and a component index key-value pair. The user access component path list includes a component access path ID set, a permission level label, a role category identifier and a path authorization mapping table. The cloud-based three-dimensional loading priority path set includes a loading priority number, an area access hotspot indicator, a task urgency level code and a path scheduling index set.

2. The cloud computing-based real-scene 3D data management system according to claim 1, characterized in that: The model verification module includes: The boundary extraction submodule obtains the boundary line segment set of the real-life 3D component submitted by the access end to the cloud service platform, extracts the coordinates of the two ends of each line segment to construct a vertex list, counts the line segment connection values ​​corresponding to the vertices, calls the boundary connectivity threshold and compares it with the vertex connection value to determine whether there are boundary anomalies with low connection counts, records the coordinates of vertices with connection counts less than the threshold and establishes a vertex number index to generate a boundary closure anomaly point set; The structure identification submodule obtains the direction vector value of the associated boundary segment based on the boundary closure anomaly point set and the spatial position coordinates of the corresponding vertex. Adjacent vectors are extracted in pairs according to the segment connection order, and the cosine values ​​of the angles between adjacent vectors are calculated to obtain the direction cosine values ​​at the connection points. The direction cosine values ​​are compared with the direction cosine similarity threshold, and the node groups with prominent direction offsets are screened. The direction cosine deviation degree value is obtained by calculation to determine whether it exceeds the cosine offset set threshold. If the conditions are met, it is marked as a structural direction anomaly point and a direction anomaly identification list is established. The consistency integration submodule calls the boundary closure anomaly point set and the direction anomaly identification list, merges the corresponding points according to the component number, constructs an anomaly coordinate set based on the component structural unit, integrates the index value and anomaly type label of the point in the set, and establishes a cloud-based three-dimensional boundary consistency indicator set.

3. The cloud computing-based real-scene 3D data management system according to claim 2, characterized in that: The hierarchy generation module includes: The structure acquisition submodule obtains the 3D mesh structure data of the model facets in the area based on the cloud-based 3D boundary consistency index set and spatial positioning coordinates, extracts the number of edges, normal vector angles, and adjacent face angles of each facet from the cloud-based parsing task, and establishes a facet geometric feature parameter table; The difference calculation submodule is based on the facet geometric feature parameter table, and according to the number of edges, normal vector angle, adjacent face angle and corresponding position index number of each facet, calls the structural layer control parameter to perform score calculation, calculates and obtains the structural difference score value, determines the relative variation range of the facet geometric characteristics in the same area, and generates a structural difference score list; The level allocation submodule calls the structural difference score list, performs segmented judgment on the score value and the set LOD detail level threshold, assigns an LOD detail level identifier to each spatial area face according to the score range, and establishes a real scene area LOD structure configuration list.

4. The cloud computing-based real-scene 3D data management system according to claim 3, characterized in that: The label collection module includes: The regional parsing submodule calls the LOD structure configuration list of the real scene area, obtains the cloud component storage path information based on the area number and LOD level of detail identification item, extracts the three fields of the corresponding component: building function label, geospatial location code and layer attribute, merges them according to the area number, and establishes a component label data set; The label construction submodule extracts the function type name, spatial coordinate code and layer flag value based on the component label dataset and the area number index field and the three label field contents, combines the three data contents to construct a multi-field index label, and generates a unique value code for each combination to generate a component label code value list; The path binding submodule calls the component label code value list, establishes a key-value binding table of component path and label code according to the path index identifier corresponding to the code item and the component storage path, constructs a mapping structure from label code to component path, and establishes a cloud component label index table.

5. The cloud computing-based real-scene 3D data management system according to claim 4, characterized in that: The authority configuration module includes: The permission reading submodule is based on the cloud component tag index table, according to the building function tag items, according to the correspondence between the function tag items and the component type numbers, calls the access control list in the cloud service platform, filters the access permission level values ​​corresponding to the function tag matching items, and binds the results to the component type index structure to generate a component permission level list; The access identification submodule calls the component permission level list, extracts the user role identifier based on the access request information of the current cloud component, performs a corresponding judgment process on the role identifier and the permission level, outputs the access component number interval that each type of access terminal user has within the current permission range, and generates role access number interval information; The path screening submodule traces back the corresponding path ID value of the index component in the cloud path table according to the role access number interval information, screens the path ID and number matching results, collects the accessible path index data according to the user role, and establishes a user access component path list.

6. The cloud computing-based real-scene 3D data management system according to claim 5, characterized in that: The request sorting module includes: The path extraction submodule calls the user access component path list, obtains the geospatial location code associated with each path, extracts the cumulative request count value of the component path ID in the access log, merges and records the request path and access frequency after merging by path ID, and generates a path request frequency distribution value; The task evaluation submodule extracts the business function parameters submitted by the user based on the path request frequency distribution value, compares them with the task level code in the cloud service task priority configuration table, obtains the task function level value and role identity information value of each path, calls the regional heat level item corresponding to the path spatial location, calculates and obtains the path comprehensive priority score value, determines the ranking level of the component path in the multi-path scheduling system, and generates a path priority score sequence; The priority allocation submodule calls the path priority score sequence, sorts the paths in descending order according to the priority score value corresponding to each path ID, writes the sorting results into the cache loading task queue according to the path index table, and establishes a cloud-based 3D loading priority path set.

Images