Large-scale three-dimensional model data rendering method based on multi-level LOD organization dynamic scheduling

Abstract

The application provides a large-scale three-dimensional model data rendering method based on multi-level LOD organization dynamic scheduling, and relates to the field of data processing. The original three-dimensional model data is acquired and the observation area is divided, the contour basis, the fold basis, the occlusion basis and the neighborhood basis are extracted to construct a geometric record table; multi-level LOD candidate clusters are generated based on the observation area and attribute mapping relationship is established; the observation maintenance interval is generated combined with the viewpoint position, the viewpoint movement trend and the occlusion change trend and written into the geometric record table; when the target LOD enters the observation area, temporary takeover is performed and item-by-item comparison and historical consistency verification are performed; after the verification passes, function step-by-step takeover is performed, the geometric record table is updated and the original LOD is reserved as a backup layer; after the takeover is completed, the level switching is completed and the current frame rendering result is output. Therefore, it is convenient to establish a stable cross-scale geometric consistency constraint mechanism in the multi-level LOD dynamic scheduling process, and the problems of normal flip, curvature fracture and topological mismatch in the level switching process are avoided.

Description

Technical Field

[0001] This application relates to the technical field of data processing, specifically to a method for rendering large-scale 3D model data based on dynamic scheduling of multi-level LOD organization. Background Technology

[0002] In the rendering of large-scale 3D model data, multi-level Levels of Depth (LOD) are used to dynamically adjust model accuracy under different view distances and resource conditions. However, since each LOD level is usually generated independently and lacks unified cross-scale geometric constraints, inconsistencies in geometric representation can easily occur at the boundary regions of different levels. This problem is more prominent in scenarios with model block organization and independent region scheduling, easily leading to phenomena such as inconsistencies in normals, curvature discontinuities, and topological adjacency offsets at the stitching boundaries.

[0003] In dynamic rendering processes involving rapid viewpoint movement or frequent changes in occlusion relationships, the geometrical differences between multi-level Levels of Detail (LOD) are amplified. When the system performs LOD switching on the same spatial region, issues such as abrupt normal changes, surface flipping, or boundary tearing can easily occur at the boundaries. This is especially true in self-similar structures or high-curvature folded surface regions, where these problems are more likely to evolve into obvious geometric inversions or failed gap reconstructions, thus affecting rendering stability. While existing technologies alleviate these problems by adding transition levels, progressive blending, or boundary smoothing, they largely rely on local patching or post-processing methods, making it difficult to fundamentally guarantee cross-level geometric consistency. Under conditions of multi-region parallel scheduling and complex dynamic scene changes, it remains difficult to balance rendering efficiency and geometric stability.

[0004] Therefore, how to establish a stable cross-scale geometric consistency constraint mechanism in the dynamic scheduling of multi-level LODs, and avoid problems such as normal flipping, curvature breakage and topology mismatch during level switching, has become an urgent technical problem to be solved. Summary of the Invention

[0005] This application provides a method for rendering large-scale 3D model data based on dynamic scheduling of multi-level LOD organization, which facilitates the establishment of a stable cross-scale geometric consistency constraint mechanism during multi-level LOD dynamic scheduling, and avoids problems such as normal flipping, curvature breakage and topology mismatch during level switching.

[0006] The first aspect of this application provides a method for rendering large-scale 3D model data based on dynamic scheduling of multi-level LOD organization. The method includes: acquiring original 3D model data and performing spatial continuity and visible region segmentation processing to form an observation region; simultaneously extracting contour criteria, facet criteria, occlusion criteria, and neighborhood criteria to construct a corresponding geometric record table; constructing a multi-level LOD candidate cluster based on the observation region, and functionally decomposing each level of LOD into contour processing attributes, facet processing attributes, occlusion processing attributes, and neighborhood processing attributes to establish a mapping relationship with the geometric record table; generating an observation holding interval for the observation region based on the viewpoint position, viewpoint movement trend, and occlusion change trend, and then... The observation holding interval is written into the geometry record table as a functional takeover constraint. When the target LOD enters the corresponding observation area, a temporary takeover process is performed based on the geometry record table to generate a comparison result. The comparison result is then compared item by item with the current judgment result in the geometry record table, and historical consistency is verified. After the comparison result passes the item-by-item comparison and historical consistency verification, a step-by-step functional takeover is performed on the target LOD, and the geometry record table is updated item by item, while the original LOD is retained as a backup layer. After the target LOD completes the takeover, the target LOD is converted to the current layer, and the original LOD is downgraded to a backup layer. The current frame rendering result is then output based on the updated geometry record table.

[0007] A second aspect of this application provides a large-scale 3D model data rendering apparatus based on multi-level LOD organization dynamic scheduling. The apparatus includes an acquisition module and a processing module. The acquisition module is used to acquire raw 3D model data and perform spatial continuity and visible area segmentation processing to form an observation region. Simultaneously, it extracts contour criteria, facet criteria, occlusion criteria, and neighborhood criteria to construct a corresponding geometric record table. The processing module is used to construct multi-level LOD candidate clusters based on the observation region and perform functional decomposition of contour processing attributes, facet processing attributes, occlusion processing attributes, and neighborhood processing attributes on each level of LOD to establish a mapping relationship with the geometric record table. The processing module is also used to generate an observation hold for the observation region based on the viewpoint position, viewpoint movement trend, and occlusion change trend. The processing module is further configured to: ...

[0008] A third aspect of this application provides an electronic device including a processor, a memory, a user interface, and a network interface. The memory is used to store instructions, and both the user interface and the network interface are used to communicate with other devices. The processor is used to execute the instructions stored in the memory to cause the electronic device to perform the method described above.

[0009] A fourth aspect of this application provides a non-transitory computer-readable storage medium storing instructions that, when executed, perform the method described above.

[0010] In summary, one or more technical solutions provided in this application have at least the following technical effects or advantages: By introducing a geometric record table as a unified carrier in the multi-level LOD dynamic scheduling process, the observation region possesses complete judgment criteria from the initial construction stage, including contour, facet, occlusion, and neighborhood criteria. This avoids the boundary inconsistency problem caused by the independent expression of LODs at each level in traditional schemes, establishing a unified constraint foundation for cross-level geometric consistency from the source. During the multi-level LOD candidate cluster construction stage, by functionally decomposing contour processing attributes, facet processing attributes, occlusion processing attributes, and neighborhood processing attributes for each LOD level and establishing a mapping relationship with the geometric record table, level switching no longer relies on overall replacement but is transformed into a fine-grained takeover process around different processing functions. This reduces the risk of geometric abrupt changes caused by one-time switching and improves the controllability and stability of level switching.

[0011] By generating observation hold intervals and writing them into the geometric record table as functional takeover constraints, the system possesses temporally stable window constraints during continuous frame scheduling. It can dynamically limit the timing of level switching based on viewpoint position, viewpoint movement trends, and occlusion change trends, avoiding level replacement during highly sensitive phases and effectively suppressing geometric consistency reversal issues caused by sudden changes in viewpoint or occlusion. A temporary takeover mechanism is introduced when a target LOD enters. By generating comparison results and performing item-by-item comparisons and historical consistency checks, candidate levels undergo multi-dimensional verification before formal takeover. The impact on the current observation area is assessed from multiple levels, including contour, facets, occlusion, and neighborhood, and continuity is checked in conjunction with historical judgment trajectories. This proactively intercepts levels that may cause normal flipping, curvature breakage, or topology mismatch, significantly improving the reliability of scheduling decisions.

[0012] By employing a phased takeover mechanism, the original overall hierarchical replacement process is broken down into a progressive takeover process that handles each function individually. This ensures that the target LOD remains under control throughout the takeover process, while allowing the original LOD to be retained as a backup layer. This enables rapid rollback in case of anomalies at any stage, preventing abrupt changes in the overall rendering result and improving the system's robustness and fault tolerance. After the takeover is complete, a clear distinction between the current and backup layers is established, and the current frame's rendering result is output based on an updated geometry record table. This ensures that the final output result is based on a verified and constrained stable judgment, while simultaneously writing the current frame's result into the geometry record table to form a historical benchmark for subsequent frames, maintaining continuity across frames. Therefore, it facilitates the establishment of a stable cross-scale geometric consistency constraint mechanism during multi-level LOD dynamic scheduling, avoiding issues such as normal flipping, curvature breakage, and topology mismatch during hierarchical switching. Attached Figure Description

[0013] Figure 1 A schematic diagram illustrating the process of rendering large-scale 3D model data based on dynamic scheduling of multi-level LOD organization, as provided in an embodiment of this application; Figure 2 A schematic diagram of a large-scale 3D model data device based on dynamic scheduling of multi-level LOD organization provided in an embodiment of this application; Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0014] Explanation of reference numerals in the attached figures: 21. Acquisition module; 22. Processing module; 31. Processor; 32. Communication bus; 33. User interface; 34. Network interface; 35. Memory. Detailed Implementation

[0015] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0016] In the description of the embodiments of this application, the words "for example" or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design that is described as "for example" or "for instance" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design options. Rather, the use of the words "for example" or "for instance" is intended to present the relevant concepts in a specific manner.

[0017] In the description of the embodiments of this application, the term "multiple" means two or more. For example, multiple systems means two or more systems, and multiple screen terminals means two or more screen terminals. In addition, the terms "comprising," "including," "having," and variations thereof all mean "including but not limited to," unless otherwise specifically emphasized.

[0018] To address the aforementioned technical problems, this application provides a method for rendering large-scale 3D model data based on dynamic scheduling of multi-level LOD organizations, referring to... Figure 1 , Figure 1 This is a flowchart illustrating a large-scale 3D model data rendering method based on multi-level LOD organization dynamic scheduling provided in an embodiment of this application. The method is applied to a server and includes steps S110 to S160, as follows:

[0019] S110. Obtain the original 3D model data and perform spatial continuity and visible area division processing to form the observation area. At the same time, extract the contour basis, fold basis, occlusion basis and neighborhood basis to construct the corresponding geometric record table.

[0020] Specifically, after the server acquires the raw 3D model data, it first performs a unified scene processing procedure to form a scene representation set. The raw 3D model data refers to a collection of 3D object data generated from different modeling sources, sampling accuracies, and mesh organization methods. It typically includes vertex coordinates, edge connectivity, facet topology, normal information, material attachment information, and object hierarchy. The purpose of the unified scene processing procedure is to transform the raw 3D model data, which originally had inconsistent coordinate systems, scale references, orientation definitions, and object organization methods, into a standardized scene representation that can be continuously analyzed within the same processing flow. During processing, first, coordinate references are unified for all 3D objects, ensuring that all objects fall into the same global scene coordinate system; then, model orientation is standardized to ensure that normal orientation and surface outward rules are consistent; subsequently, scale is normalized to ensure that objects from different sources have the same quantitative standard for spatial distance judgment and region division; finally, vertex sets, edge sets, face sets, and object reference relationships are written into a unified data organization structure to form the scene representation set. The scene representation set can be understood as a standardized scene container for subsequent analysis, where each 3D object not only retains its own geometric information but also establishes traceable relationships with neighboring objects. To improve the computability of the processed results, a standardized state vector can be defined for each object:

[0021] in, Indicates the first A standardized state vector of an object; Indicates the first The set of spatial positions of an object after a unified coordinate transformation, the value of which is the coordinate result of all vertices of the object in the global scene coordinate system; Indicates the first The set of normal distributions of an object after orientation unification is taken as the unit normal vector corresponding to each facet or vertex of the object. Indicates the first The topological connection set of an object, which takes the values ​​of the connection relationships between vertices, edges, and faces, as well as the internal adjacency relationships of the objects; Indicates the first An object's additional description set typically includes material type, texture attachment relationships, object category markers, and hierarchy identifiers. The purpose of this expression is to unify geometric location, surface orientation, topological connectivity, and additional attributes into a single structure, enabling subsequent spatial continuity identification to focus on the structured state of the object rather than fragmented data.

[0022] After forming a scene representation set, spatial continuity recognition processing is performed based on the scene representation set to form spatial continuity candidate regions. Spatial continuity does not simply refer to the proximity between two objects, but rather to the continuous connection between objects in terms of spatial location, boundary contact, surface extension, normal transition, and visual association. A spatial continuity candidate region is a candidate region composed of multiple spatially continuous, visually related, and topologically merging objects. Its role is to provide a more stable intermediate organizational unit for subsequent observation region division than traditional grid partitioning. In practice, the server calculates the continuity strength between objects around each object in the scene representation set. The higher the continuity strength, the more suitable the two objects are to be classified into the same spatial continuity candidate region. The continuity strength can be determined by comprehensively considering the boundary proximity, normal similarity, surface extension, and occlusion propagation correlation between objects, and can be expressed by the following continuity evaluation expression:

[0023] in, Representation Object With object The strength of continuity between them; the larger the value, the more suitable they are to be merged into the same spatial continuous candidate region. Indicates boundary proximity, used to reflect the proximity of objects With object The spatial proximity of the edge regions is usually calculated by inversely based on the distance between the nearest boundary points of two objects; the closer the distance, the larger the value. Normal similarity is used to reflect the similarity of objects. With object The degree to which surfaces in contact or adjacent areas are aligned is usually determined by how close the angles between their corresponding normal vectors are; the smaller the angle, the larger the value. Indicates surface extensibility, used to describe an object With object Whether a continuous transition is formed on the local surface trend is usually calculated based on the trend of surface change on both sides of the boundary or the distribution of connecting edges; Indicates occlusion correlation, used to describe objects With object Whether they frequently participate in the front-to-back display relationship in the main line of sight direction; if there is stable occlusion transmission between the two in multiple viewpoints, then this value is higher. , , , These represent the weight coefficients corresponding to boundary proximity, normal similarity, surface extensibility, and occlusion correlation, respectively. Their values ​​are typically non-negative and are set according to the specific scenario's emphasis on spatial continuity. The principle behind this expression is to uniformly convert four factors—spatial proximity, surface consistency, geometric continuity, and visual correlation—into a single continuity intensity value. Objects with high continuity relationships are then merged based on this value to form spatial continuity candidate regions. After merging, the objects within each spatial continuity candidate region are more spatially stable, which is beneficial for subsequently forming the observation area.

[0024] After the formation of spatially contiguous candidate regions, the server performs visual region segmentation processing on these regions to form observation regions. Then, it performs region registration processing on these observation regions to generate region identifiers, spatial boundary descriptions, neighborhood connection descriptions, and object member indexes. Visual region segmentation processing means not simply dividing candidate regions according to geometric distance, but further combining the consistency of appearance under viewpoint changes, sensitivity to occlusion changes, and the degree of common response of objects within the region to dynamic scheduling to further subdivide the spatially contiguous candidate regions. The observation region is the basic unit for subsequent multi-level LOD dynamic scheduling. It requires that objects within the same region have high consistency in outline appearance, facet exposure, occlusion changes, and neighborhood linkage, so that the region as a whole can remain relatively stable during subsequent level switching. In implementation, the server first performs multi-view projection analysis on the spatially contiguous candidate regions under several preset viewing directions and preset viewing distances to extract the stability of the outer contour, the frequency of main facet appearance, and the frequency of occlusion switching under different viewpoints. Then, sub-regions with similar responses are assigned to the same observation region. To quantify whether a sub-region is suitable as an independent observation region, a visual consistency index can be defined:

[0025] in, Indicates the first The visual consistency index of candidate sub-regions indicates that the larger the value, the more suitable the sub-region is as a unified observation area under the change of viewpoint. This indicates the profile stability, which reflects the degree to which the candidate sub-region maintains the position and closure tendency of the outer profile under multi-view conditions. It is usually determined based on the degree of overlap between multiple projected profiles. This indicates the consistency of the fold lines, which reflects the stability of the order and transition relationship of the main fold lines in the candidate sub-region under different viewpoints. The more consistent the exposure pattern of the main fold lines, the higher this value. This indicates the occlusion stability, reflecting whether the appearance relationship of the candidate sub-region changes smoothly during multi-viewpoint switching. The more concentrated and predictable the occlusion switching, the higher this value. This indicates the degree of regional linkage, which reflects whether objects within the candidate sub-region tend to appear or disappear together when the viewpoint changes. The stronger the degree of joint response, the higher the value. , , , These represent the weighting coefficients for contour stability, facet consistency, occlusion stability, and region linkage, respectively. Their values ​​are determined based on the current scene's emphasis on boundary stability, facet stability, occlusion stability, and region linkage. The principle behind this expression is to use a unified metric to measure the display consistency of candidate sub-regions under multi-viewpoint conditions, thus providing a quantitative basis for visible region segmentation. After completing the observation region segmentation, a region registration process is further performed. The region identifier uniquely identifies the current observation region; the spatial boundary description records the bounding range and boundary position of the observation region in the 3D scene; the neighborhood connectivity description records the contact position, continuation direction, and connection strength between the observation region and adjacent observation regions; and the object member index records all 3D objects, local regions, and boundary fragments contained within the current observation region. Through the region registration process, the observation region is transformed from a temporary segmentation result into a formal region entity that can be continuously referenced and updated in subsequent steps.

[0026] After completing the observation area division and registration, the server performs contour basis extraction, facet basis extraction, occlusion basis extraction, and neighborhood basis extraction processes around the observation area to form contour basis, facet basis, occlusion basis, and neighborhood basis. These, along with the area identifier, spatial boundary description, neighborhood connection description, and object member index, are written into the geometric record table. Contour basis refers to the set of boundary information that stably represents the appearance pattern of the outer boundary of the observation area, including main boundary segments, boundary extension directions, and boundary closure relationships. Facet basis refers to the set of structural information that stably represents the surface transitions, concavity / convexity relationships, and facet continuation sequence within the observation area. Occlusion basis refers to the set of visual relationships that represent the front-to-back appearance relationship, occlusion entry path, and occlusion exit path of the observation area under different viewing directions. Neighborhood basis refers to the set of relationships that represent the boundary continuation, surface transition, and linkage change patterns between the current observation area and adjacent observation areas. In practice, the server first performs outer boundary scanning and multi-view projection comparison on the observation area, filtering out outer contour edges and surfaces that remain stably visible under multiple views to form contour basis. Then, it performs normal turning detection and main fold plane clustering within the area to identify fold plane bands and turning sequences that determine the main shape relationships of the area, forming fold plane basis. Subsequently, it statistically analyzes the before-and-after visibility relationship and occlusion switching positions between the observation area and surrounding areas under multiple preset viewpoints and dynamic viewpoint paths, forming occlusion basis. Finally, it combines the neighborhood connection description obtained during area registration to analyze the contact boundary position, transition direction, and linkage change trend, forming neighborhood basis. To improve the expressiveness of basis extraction, the geometric record table can be organized into the following structure:

[0027] in, Indicates the first Geometric record table corresponding to each observation area; This represents a region identifier, used to uniquely identify the observation area; This indicates a spatial boundary description, used to describe the boundary extent, enclosing shape, and distribution of boundary segments of the observed area in the scene; This indicates a neighborhood connectivity description, used to describe the contact relationships and connection directions between the observed region and its neighboring observed regions; Represents an object member index, used to indicate the set of objects and local geometric units contained in the observation region; The outline is used to represent the stable boundary information of the outer contour of the observation area; The basis for indicating the folds is used to characterize the stable transition information of the fold structure within the observation area; This indicates the basis for occlusion and is used to characterize the information on the front and back visibility and occlusion switching of the observation area under different observation directions; The neighborhood criteria are used to characterize the continuity and linkage information between the observed region and its neighboring observed regions. The principle behind this expression is to uniformly write region identity information, region boundary information, region member information, and four types of stability determination criteria into the same record structure. This allows subsequent multi-level LOD candidate cluster construction, mapping relationship establishment, observation-holding interval generation, temporary takeover processing, and functional step-by-step takeover processing to all revolve around the same geometric record table. Thus, the aforementioned three steps form a clear sequential relationship in the technology chain: first, a standardized scene expression is formed through unified scene processing; then, an observed region is formed through spatial continuity recognition and visible area division; and finally, a geometric record table is formed through the extraction of the four types of criteria and unified table writing. This provides a complete prerequisite foundation for subsequent multi-level LOD dynamic scheduling based on observed regions.

[0028] S120. Construct multi-level LOD candidate clusters based on the observation region, and perform functional decomposition of contour processing attributes, facet processing attributes, occlusion processing attributes, and neighborhood processing attributes for each level of LOD in order to establish a mapping relationship with the geometric record table.

[0029] Specifically, after the server has completed the division of the observation region and constructed the geometric record table, it first performs intra-regional hierarchical identification processing on the object member indexes within the observation region to divide them into core-preserving segments and adjustable segments. The object member index refers to the set of indexes registered in the geometric record table for all 3D objects, local regions, boundary segments, and facet connection units belonging to the current observation region. Its function is to provide a unified retrieval entry point for subsequent intra-regional analysis. The purpose of intra-regional hierarchical identification processing is not to simply re-divide the observation region, but to identify which geometric parts in the observation region directly determine the region's contour closure, main facet stability, occlusion order, and neighborhood continuity, and which geometric parts mainly undertake detail supplementation and local modification. The former is defined as core-preserving segments, and the latter as adjustable segments. Core-preserving segments refer to geometric segments that directly support contour, facet, occlusion, and neighborhood criteria in the current observation region and should not be significantly rewritten during subsequent layer reduction. Adjustable segments refer to geometric segments that, without compromising the overall geometric judgment criteria of the observation region, are allowed to be simplified, merged, or weakened with changes in layer level. In practice, the server reads the corresponding boundary positions, normal changes, curvature transitions, occlusion participation frequencies, and neighborhood contact strengths one by one around the object member index, and calculates the support strength of each geometric segment for the four types of criteria. To uniformly measure the importance of segments, a segment support strength expression can be defined:

[0030] in, Indicates the first The support strength of a geometric segment; the larger the value, the more likely the segment should be classified as a core retaining segment. Indicates the first The support of a geometric segment to the contour is used to reflect whether the segment is located in the main segment of the outer contour, the boundary closure link, or the main extension direction of the contour. If the segment plays a decisive role in the stability of the contour of the observation area, the value is higher. Indicates the first The support of a geometric segment to the fold surface is used to reflect whether the segment is located at the main fold surface position, the fold surface turning zone, or the fold surface connection node. If it has obvious support for the regional shape transition, the value is higher. Indicates the first The support of a geometric segment to the occlusion basis is used to reflect whether the segment frequently participates in the formation of the front-to-back display relationship, the occlusion entry boundary, or the occlusion exit boundary. If it has a significant impact on the stability of the occlusion order, the value is higher. Indicates the first The support of a geometric segment to the neighborhood is used to reflect whether the segment is located at the regional contact boundary, surface transition zone, or linkage position of adjacent observation areas. If it plays a decisive role in the stability of the neighborhood, the value is higher. , , , These represent the weight coefficients corresponding to contour support, facet support, occlusion support, and neighborhood support, respectively. Their values ​​are set based on the current scene's emphasis on boundary stability, shape stability, occlusion stability, and continuity stability. The principle behind this expression is to uniformly convert the contribution of geometric segments to the four types of support into support strength. Then, based on the comparison between the support strength and a preset threshold, high-support segments are classified as core-preserving segments, and the remaining segments are classified as adjustable segments. Through this processing, the geometric organization within the observation area is transformed from the original set of objects into a segment structure with a constraint hierarchy, providing a clear structural foundation for subsequent multi-level progressive construction.

[0031] After dividing the core-preserving fragments and adjustable fragments, the server constructs contour-preserving constraints, facet-preserving constraints, occlusion-preserving constraints, and neighborhood-preserving constraints based on contour, facet, occlusion, and neighborhood criteria. Then, based on the core-preserving fragments and adjustable fragments, it performs multi-level progressive construction processing under these constraints to generate multi-level LOD candidate clusters. Contour-preserving constraints require candidate levels to maintain the position of the outer contour boundary of the observation region, the contour closure trend, and the main extension direction of the contour as the level decreases. Facet-preserving constraints require candidate levels to maintain the position of the main facet, the facet turning relationship, and the facet succession order. Occlusion-preserving constraints require candidate levels to maintain the front-to-back display order of the observation region, the occlusion entry boundary, and the occlusion exit boundary. Neighborhood-preserving constraints require candidate levels to maintain the position of the contact boundary with adjacent observation regions, the surface transition direction, and the range of linkage influence. The goal of multi-level progressive construction processing is to generate several candidate levels with different levels of representational completeness but consistent region determination criteria, rather than generating several independent simplified models. In practice, the server first preserves the geometric position and topological relationships of the core preservation fragments without disrupting them. Then, it progressively performs boundary merging, region dilution, local detail reduction, and secondary structure weakening around the adjustable fragments, gradually generating lower integrity levels from high integrity levels. To determine whether a candidate level still satisfies the four types of preservation constraints, a hierarchical constraint satisfaction expression can be defined:

[0032] in, Indicates the first The constraint satisfaction of the candidate level is such that the higher the value, the better the level meets the existing geometric judgment criteria of the observation area. Indicates the first The degree to which the candidate level satisfies the contour preservation constraints is used to reflect the degree to which the position of the outer contour boundary, the contour closure tendency, and the main extension direction are preserved relative to the contour basis at that level. Indicates the first The degree to which the candidate level satisfies the constraint of the folding surface is used to reflect the degree to which the position of the main folding surface, the folding surface turning relationship and the succession sequence are maintained relative to the folding surface basis at that level; Indicates the first The degree to which the candidate level satisfies the occlusion preservation constraint is used to reflect the degree to which the order of appearance before and after, the occlusion boundary, and the occlusion propagation direction are preserved relative to the occlusion criteria at that level. Indicates the first The degree to which the candidate level satisfies the neighborhood preservation constraints is used to reflect the degree to which the region contact boundary, surface transition direction and linkage range are preserved relative to the neighborhood criteria at that level. , , , These represent the weight coefficients corresponding to the contour preservation constraint, facet preservation constraint, occlusion preservation constraint, and neighborhood preservation constraint, respectively. Their values ​​are determined based on the dependence of the observation region on the stability of the outer contour, the stability of the principal facet, the stability of occlusion, and the stability of neighborhood continuity. The principle of this expression is to uniformly organize the satisfaction status of a certain level of candidate hierarchy on the four types of preservation constraints into a single constraint satisfaction value, and use this value to filter out effective levels that can be included in the multi-level LOD candidate cluster. The final generated multi-level LOD candidate cluster refers to a set of candidate levels that belong to the same observation region, share the same geometric record table constraints, and are arranged from high to low in terms of representational completeness.

[0033] After forming a multi-level LOD candidate cluster, the server extracts contour processing attributes, facet processing attributes, occlusion processing attributes, and neighborhood processing attributes for each LOD level within the cluster and performs attribute classification processing. Contour processing attributes describe a LOD level's ability to maintain the outer contour boundary, contour closure relationship, and main extension direction of the observation area. Facet processing attributes describe a LOD level's ability to maintain the position, turning relationship, and sequence of main faces within the observation area. Occlusion processing attributes describe a LOD level's ability to maintain the order of appearance of the observation area, the position of occlusion boundaries, and the stability of occlusion switching. Neighborhood processing attributes describe a LOD level's ability to maintain the contact boundary, surface transition relationship, and linkage stability between the observation area and adjacent observation areas. The purpose of attribute classification processing is not only to determine whether a LOD level belongs to a high-level or low-level category, but also to clarify the support level of that level in the four types of region determination tasks. This allows subsequent takeovers to avoid a complete replacement and instead be called step-by-step based on different processing attributes. During implementation, the server first calculates the preservation results of each level of LOD in terms of contour preservation, facet preservation, occlusion preservation, and neighborhood preservation, and then organizes these results into corresponding processing attribute values. Subsequently, based on the relative strength of the processing attribute values ​​within the candidate cluster, each level of LOD is assigned a high support level, medium support level, or low support level in the four processing attributes. To uniformly express the attribute ranking results, a processing attribute strength expression can be defined:

[0034] in, Indicates the first Level LOD in The class handles the attribute strength of the attribute; where, the first The class processing attributes can correspond to contour processing attributes, facet processing attributes, occlusion processing attributes, or neighborhood processing attributes in sequence. Indicates the first Level LOD in The degree of retention of the benchmark in the class processing attributes is used to reflect the degree of fit of this level to the geometric record table benchmark in the corresponding judgment task; Indicates the first Level LOD in The stability of the class processing attribute is used to reflect the degree of fluctuation of the corresponding judgment result under continuous viewpoint changes and level switching conditions. The smaller the fluctuation, the higher the value. Indicates the first Level LOD in The continuity of the class processing attribute is used to reflect the smoothness of the connection between this level and the adjacent decision parts in the corresponding decision task. The more continuous the connection, the higher the value. , , These represent retention, stability, and continuity, respectively, in the [number]th [year]. The weight coefficient in the class-based processing attribute is set according to whether the processing attribute prioritizes baseline preservation, process stability, or boundary alignment, and ranges from 0 to 1. The principle behind this expression is to uniformly convert the baseline fit, process stability, and boundary alignment of a certain level of LOD in a certain type of judgment task into attribute strength, and then classify the attribute levels based on the attribute strength. Through this process, each level of LOD is no longer just a simple hierarchical number, but rather has a set of four quantifiable, comparable, and callable processing attributes.

[0035] After extracting and classifying the processing attributes of each level of LOD, the server performs a class-by-class matching process based on the contour, facet, occlusion, and neighborhood criteria in the geometric record table, and the contour, facet, occlusion, and neighborhood processing attributes in each level of LOD to establish a mapping relationship. Class-by-class matching means that instead of mapping a single LOD level to the entire observation area, it determines whether the level is suitable as a candidate support level for the corresponding decision task based on each of the four dimensions: contour, facet, occlusion, and neighborhood. The mapping relationship refers to the traceable correspondence between a certain type of criterion in the geometric record table and the corresponding processing attribute in a certain level of LOD. This is used to quickly determine which level is suitable for which decision task in subsequent observation-holding interval generation, temporary takeover processing, and functional step-by-step takeover processing. In practice, the server first compares the contour criteria in the geometry record table with the contour processing attributes of each level of LOD, selecting levels with high contour preservation and stable contour continuity to form a contour mapping relationship. Next, it compares the facet criteria with the facet processing attributes of each level of LOD, selecting levels with high main facet preservation and stable facet turning to form a facet mapping relationship. Then, it compares the occlusion criteria with the occlusion processing attributes of each level of LOD, selecting levels with stable before-and-after display order and smooth occlusion transitions to form an occlusion mapping relationship. Finally, it compares the neighborhood criteria with the neighborhood processing attributes of each level of LOD, selecting levels with accurate contact boundaries and stable linkage relationships to form a neighborhood mapping relationship. To uniformly measure the matching degree between a certain level of LOD and a certain type of criteria, a mapping matching degree expression can be defined:

[0036] in, Indicates the first Level LOD in The mapping matching degree based on the class is used; the larger the value, the more suitable the level is as a candidate support level for this class determination task; where, the first Class criteria can correspond to contour criteria, facet criteria, occlusion criteria, or neighborhood criteria in sequence; Indicates the first Level LOD in The baseline fit of the class is used to reflect the degree of consistency between its processing attributes and the corresponding basis of the geometric record table; Indicates the first Level LOD in The strength of the attribute on the class processing attribute, that is, the result of the strength of the processing attribute after being extracted and classified in the previous stage; Indicates the first Level LOD in Historical reliability in class determination tasks reflects the stable performance of this level when undertaking class determination tasks in similar observation areas or preceding consecutive frames. The more stable the performance, the higher the value. , , These represent the weighting coefficients corresponding to baseline fit, attribute strength, and historical reliability, respectively. Their values ​​are set according to whether the current mapping prioritizes immediate matching, hierarchical capability, or historical stability, ranging from 0 to 1. The principle behind this expression is to uniformly convert the geometric record table's baseline requirements, the hierarchical's own processing attributes, and historical stability performance into a mapping matching degree. Then, based on the mapping matching degree results, appropriate candidate support levels are selected according to contour basis, facet basis, occlusion basis, and neighborhood basis, respectively. After completing the class-by-class matching, the server writes the results into the geometric record table, forming contour mapping items, facet mapping items, occlusion mapping items, and neighborhood mapping items.

[0037] S130. Based on the viewpoint position, viewpoint movement trend, and occlusion change trend, generate the observation holding interval of the observation area and write the observation holding interval into the geometric record table as a function takeover constraint.

[0038] Specifically, after the server has completed the division of the observation region, the construction of the geometric record table, and the establishment of the mapping relationship between multi-level LOD candidate clusters, it generates the observation holding interval of the observation region based on the viewpoint position, viewpoint movement trend, and occlusion change trend, and writes the observation holding interval into the geometric record table as a functional takeover constraint. The core function of this processing step is to prevent the target LOD from directly requesting takeover at any time. Instead, it first establishes a stable holding window around the current observation region, which is constrained by both viewpoint changes and occlusion evolution. This ensures that subsequent temporary takeover processing and functional step-by-step takeover processing occur within this window according to the defined boundaries. The observation holding interval refers to the effective interval within which the observation region needs to maintain the stability of the current judgment structure and is not suitable for sudden hierarchical rewriting within a continuous time range. The functional takeover constraint refers to the restrictions imposed on the takeover timing, takeover order, and takeover range of contour processing attributes, facet processing attributes, occlusion processing attributes, and neighborhood processing attributes within the observation holding interval. Since the previous step has established the mapping relationship between the geometry record table and each level of LOD, this step no longer re-determines the level switch from the original geometry level. Instead, it directly maintains control at the continuous temporal level around the contour basis, facet basis, occlusion basis, neighborhood basis, as well as the contour mapping item, facet mapping item, occlusion mapping item, and neighborhood mapping item in the geometry record table.

[0039] When performing recognition processing based on viewpoint position, viewpoint movement trend, and occlusion change trend, the following results are obtained first: region visibility status localization processing result, continuous observation trend recognition processing result, and occlusion evolution recognition processing result. Viewpoint position refers to the observer's spatial position and gaze direction in the scene coordinate system under the current rendering frame; viewpoint movement trend refers to the changes in the viewpoint's speed, direction, acceleration, and proximity to or distance from the observation area in consecutive frames; occlusion change trend refers to the evolution direction of the visibility relationship between the current observation area and adjacent observation areas or nearby objects in consecutive frames. Region visibility status localization processing result refers to the comprehensive localization result of the current observation area's visibility level, projection coverage, main outline exposure, and main facet exposure under the current viewpoint conditions; continuous observation trend recognition processing result refers to the recognition result of whether the viewpoint will continue to focus on the current observation area in subsequent consecutive frames, whether it will circle the observation area, or whether it will quickly pass over the observation area; occlusion evolution recognition processing result refers to the recognition result of whether the current observation area maintains stable occlusion with adjacent areas, gradually removes occlusion, or enters a frequent occlusion switching state in subsequent consecutive frames. To achieve a unified recognition and quantification result, an expression for the intensity of the region manifestation state can be defined:

[0040] in, Indicates the first The intensity of the visibility of an observation area at the current viewpoint position; the larger the value, the more prominent the observation area is at present. This indicates the projection coverage, which describes the proportion of the observed area in the current screen space or the range of projection influence. The larger the coverage, the higher the value. This value represents the outline exposure, describing the completeness and clarity of the outer outline of the observation area at the current viewpoint. The more complete the outer outline, the higher the value. This indicates the exposure of the folded surface, which describes the degree of exposure and the discernibility of the transition of the main folded surface within the observation area at the current viewpoint. The more obvious the main folded surface, the higher the value. This indicates visibility dominance and describes whether the observed area is in the dominant position in the current line of sight. If it is located in the core of the line of sight and is not severely obstructed, this value is higher. , , , These represent the weighting coefficients for projection coverage, contour exposure, planar exposure, and visibility dominance, respectively. Their values ​​are determined based on whether the current system prioritizes the area influence, contour influence, planar influence, or dominant visibility position, and are set between 0 and 1. The principle behind this expression is to uniformly convert the visual influence of the observed area at the current viewpoint position into a visibility state intensity, thereby determining whether the observed area is in a high visibility, transitional visibility, or weak visibility state.

[0041] While obtaining the location processing result of the region display state, it is also necessary to form a continuous observation trend recognition processing result based on the viewpoint movement trend. Continuous observation trend recognition processing does not simply determine whether the viewpoint is currently facing the observation area, but rather whether the viewpoint will maintain continuous attention to the current observation area in several subsequent consecutive frames. If the viewpoint is steadily approaching the observation area, it indicates that the observation area will gradually enlarge and its detail sensitivity will increase in subsequent frames; if the viewpoint is laterally rotating around the observation area, it indicates that the contours and surfaces will undergo continuous exposure changes; if the viewpoint is rapidly passing over the observation area, it indicates that the current display state may only exist for a short time and should not trigger large-scale hierarchical takeover. To quantify the continuous observation trend of the viewpoint, a continuous observation trend strength expression can be defined:

[0042] in, Indicates the first The intensity of the continuous observation trend corresponding to each observation area, the larger the value, the more likely the viewpoint is to continue to pay attention to the observation area in subsequent consecutive frames; This indicates the alignment of the line of sight, which describes the degree of consistency between the forward direction of the viewpoint and the center direction of the observation area. The higher the alignment, the higher this value. This indicates the proximity trend, describing whether the distance between the viewpoint and the observation area changes in a continuous manner. If the distance continues to decrease, this value is higher. This indicates dwell stability, which describes whether the viewpoint orientation and velocity remain stable over the current observation area in consecutive frames. A higher value indicates smaller fluctuations. This indicates the degree of attention around the viewpoint, which describes whether the viewpoint maintains significant attention despite continuous angular changes around the observation area. A higher value indicates that the viewpoint continues to observe the area during the rotation. , , , These represent the weighting coefficients for view alignment, proximity trend, dwell stability, and rotational attention, respectively. Their values ​​are set based on whether the system prioritizes forward observation, continuous approach, local dwell, or rotational observation. These values ​​can be determined through least squares fitting or manual experience, and range from 0 to 1. The principle behind this expression is to transform the spatial motion of the viewpoint in consecutive frames into the degree of sustained attention to the observed area, thereby determining whether the observed area warrants a strong hold constraint.

[0043] In parallel with the two identification results mentioned above, it is also necessary to form an occlusion evolution identification processing result based on the occlusion change trend. Occlusion evolution identification processing is used to determine whether the currently observed region will experience stable occlusion maintenance, occlusion release, or frequent occlusion switching in subsequent consecutive frames, because changes in occlusion relationships directly affect whether the target LOD is suitable for takeover at the current moment. If a high-frequency switching of occlusion relationships is imminent, it is not advisable to formally take over the occlusion processing attributes and neighborhood processing attributes at this stage. To quantify the occlusion change trend, an occlusion evolution intensity expression can be defined:

[0044] in, Indicates the first The occlusion evolution intensity of an observation region is such that the larger the value, the more likely the observation region is to undergo a significant occlusion change process in subsequent consecutive frames. This indicates the occlusion boundary proximity, used to describe whether adjacent occlusion sources are approaching the key display boundary of the current observation area; the higher the proximity, the higher the value. This indicates the frequency of occlusion switching, used to describe the activity level of changes in the display relationship between preceding and following frames. The more frequent the switching, the higher this value. This indicates the trend of occlusion removal, describing whether the original occlusion is gradually being removed. If the observation area is moving from behind the occlusion to the front of the display, the value is higher. This indicates the neighboring occlusion linkage degree, which describes whether changes in the visibility of adjacent observation areas will affect the occlusion order of the current observation area. The more significant the linkage, the higher the value. , , , These represent the weighting coefficients corresponding to the occlusion boundary proximity, occlusion switching frequency, occlusion removal trend, and neighborhood occlusion linkage, respectively. Their values ​​are set based on the system's sensitivity to occlusion boundaries, frequent occlusion switching, occlusion removal, or neighborhood linkage. The principle behind this expression is to uniformly convert the antecedent factors related to occlusion changes into occlusion evolution intensity, thereby determining whether the currently observed area is in an occlusion-sensitive stage in time.

[0045] After obtaining the results of the region display status localization, continuous observation trend recognition, and occlusion evolution recognition, a retention requirement merging process is performed on these results to form the primary retention requirement for the observation region. The retention requirement merging process combines the three recognition results from different dimensions to determine whether the current observation region most needs to maintain stability in the following continuous period, specifically contour determination, planar determination, occlusion determination, or neighborhood determination. The primary retention requirement refers to the core determination requirement that is least likely to be disrupted by sudden layer switching under the combined influence of the current observation region's display status, continuous observation trend, and occlusion evolution status. In specific implementation, if the region display status intensity and continuous observation trend intensity are both high, it indicates that the observation region will continue to be exposed to the observation center in subsequent consecutive frames. In this case, contour determination and planar determination should often be identified as the primary retention requirement. If the occlusion evolution intensity is high, it indicates that the preceding and following display order and the occlusion boundaries between regions are in an active change phase. In this case, occlusion determination and neighborhood determination should often be identified as the primary retention requirement. To unify and merge multiple recognition results, a master retention requirement strength expression can be defined:

[0046] in, Indicates the first The observation area in the first The class maintains the intensity of demand; among them, the first Class preservation requirements can be corresponding to outline preservation requirements, facet preservation requirements, occlusion preservation requirements, or neighborhood preservation requirements, in turn. This indicates the intensity of the visibility status of the observed area, reflecting its current visual importance; This indicates the strength of a continuously observed trend, reflecting the level of continued attention required. It indicates the intensity of occlusion evolution and is used to reflect the sensitivity to changes in occlusion; Indicates the geometric record table and the first The strength of the mapping support corresponding to the class requirement is used to reflect the sufficiency of support between the current layer and the candidate layer in the class's judgment task. , , , These represent the intensity of the regional manifestation state, the intensity of the continuous observation trend, the intensity of the occlusion evolution, and the intensity of the mapping support, respectively, in the first... The weighting coefficients in the retention requirements are set based on the dependence of different retention requirements on visibility importance, continuous observation, occlusion evolution, and existing hierarchical mapping. The principle behind this expression is to combine the visibility importance, temporal continuity, occlusion activity, and existing mapping support of the current observation region with the four types of retention requirements, thereby identifying the primary retention requirement that should be prioritized. Subsequently, the server can determine the combination of primary retention requirements based on the strength of the four types of requirements; for example, it can determine only one primary retention requirement, or it can determine two parallel primary retention requirements when their strengths are similar.

[0047] After establishing the master hold requirement for the observation region, an observation hold interval generation process is performed based on the master hold requirement to determine the starting trigger point and duration of the observation hold interval. Functional takeover restrictions within the observation hold interval are determined based on the mapping relationship in the geometric record table. The starting trigger point refers to the moment when the current observation region begins to enter a state where the master hold requirement needs to be prioritized for stability. The duration refers to the continuous time range or continuous frame range within which the stability requirement needs to remain effective. Functional takeover restrictions refer to the limitations on which processing attributes the target LOD can request to take over, which processing attributes must be delayed for takeover, and which processing attributes are prohibited from direct takeover within the observation hold interval. The generation of the observation hold interval is not a fixed length but is determined by the strength of the master hold requirement, the continuous observation trend, and the occlusion evolution state. If an observation region is about to continuously enter a high visibility and high attention state, the observation hold interval should trigger earlier and cover a longer continuous frame range; if the observation region is only in a short-term grazing state, the observation hold interval can be shortened. To quantify the strength of the observation hold interval, an interval hold strength expression can be defined:

[0048] in, Indicates the first The interval preservation strength of an observation region is defined by the value of the interval preservation strength. A larger value indicates that a stronger observation preservation interval needs to be generated for that observation region. This represents the final determined intensity of the primary retention demand in the observed region, which is usually the maximum value among the four types of retention demand intensities or the composite value after combining the primary retention demands. This indicates the strength of a continuously observed trend, used to reflect the sustainability of subsequent demand. This indicates the intensity of occlusion evolution, used to reflect whether occlusion sensitivity requires an extended hold window; This indicates the stability support level of the current layer, reflecting whether the current layer has sufficient stability support capability. If the current layer has high stability, it is more suitable to maintain the existing state and delay takeover. , , , These represent the weighting coefficients corresponding to the strength of the primary maintenance demand, the strength of the continuous observation trend, the strength of occlusion evolution, and the stability support of the current layer, respectively. Their values ​​are set according to the system's emphasis on demand strength, persistence, occlusion changes, or the stability of the current layer. The principle behind this expression is to uniformly express the necessity of maintaining the stability of the current observation area over a future period as the interval maintenance strength. The server can determine the starting trigger point and the duration of the observation maintenance interval based on the interval maintenance strength; for example, the higher the interval maintenance strength, the earlier the starting trigger point and the longer the duration. Furthermore, it is necessary to combine the mapping relationships in the geometric record table to determine functional takeover restrictions, i.e., whether the candidate layer corresponding to the contour mapping item is allowed to apply for contour processing attribute takeover within the current interval, whether the candidate layer corresponding to the facet mapping item is allowed to apply for facet processing attribute takeover, and whether the candidate layers corresponding to the occlusion mapping item and the neighborhood mapping item must postpone takeover or be prohibited from takeover. Through this processing, the observation maintenance interval is not only a maintenance window in a time sense but also a takeover restriction window directly bound to the mapping relationship.

[0049] After the observation and hold interval is generated, a takeover constraint decomposition process is performed on the observation and hold interval to form contour takeover constraints, facet takeover constraints, occlusion takeover constraints, and neighborhood takeover constraints. The observation and hold interval and the corresponding takeover constraints are then written into the geometric record table. Takeover constraint decomposition refers to breaking down the overall hold requirements in the observation and hold interval into specific control rules that can be directly applied to the four types of processing attributes. Contour takeover constraints are rules that restrict whether contour processing attributes are allowed to apply for takeover, which candidate levels are allowed to takeover, and what verification conditions must be met before takeover. Facet takeover constraints are rules that restrict the takeover conditions, takeover order, and takeover boundaries of facet processing attributes within the interval. Occlusion takeover constraints are rules that restrict whether occlusion processing attributes can enter formal takeover during the occlusion-sensitive phase. Neighborhood takeover constraints are rules that restrict whether unilateral takeover is allowed for neighborhood processing attributes during the active change phase of the region contact boundary. During implementation, the server refines the overall constraints in the observation and hold interval to the four types of processing attributes based on the primary hold requirements and mapping relationships. For example, if the master hold requirement points to contour and facet determination, then contour takeover constraints and facet takeover constraints will be more inclined to delay sudden switching, requiring the target LOD to only enter temporarily; if the master hold requirement points to occlusion and neighborhood determination, then occlusion takeover constraints and neighborhood takeover constraints will be more inclined to freeze formal takeover, only allowing it to open after the occlusion evolution has stabilized. To uniformly express the strength of various takeover constraints, a takeover constraint strength expression can be defined:

[0050] in, Indicates the first The observation area in the first The strength of the takeover constraint on the class handling attribute; where, the first The class processing attributes can correspond to contour processing attributes, facet processing attributes, occlusion processing attributes, or neighborhood processing attributes in sequence. This indicates that the observation area is in the [number]th [year]. The strength of the retention requirement is used to reflect the retention priority of the task to be determined by that class; This indicates the strength of the interval holding, used to reflect the overall stability constraint level of the observed holding interval; Indicates the current layer is at the 1st level. The stability of a class's processing attributes reflects whether the current layer should continue to prioritize handling the functions of that class. If the current layer has a high stability, this value will be higher. Indicates the candidate level at the th The takeover maturity level on the class processing attribute is used to reflect whether the candidate level has been supported by sufficient mapping and verified by historical reliability. The lower the maturity level, the stronger the takeover constraint needs to be. , , , These represent the weighting coefficients corresponding to the strength of the holding requirement, the holding strength of the interval, the stability of the current layer, and the maturity of the candidate layer takeover, respectively. Their values ​​are determined based on the system's focus on holding decisions, overall window control, current layer stability, and candidate layer maturity. The principle behind this expression is to uniformly convert the holding necessity, overall window constraints, current layer reliability, and candidate layer maturity of a certain type of processing attribute within the current observation holding interval into takeover constraint strength. Then, based on the takeover constraint strength, this type of processing attribute is divided into different states such as allowable takeover, delayed takeover, or frozen takeover. After decomposition, the server writes the starting trigger point, duration, main holding requirement, and the four types of takeover constraints of the observation holding interval into the holding interval field, holding requirement field, and takeover constraint field of the geometry record table.

[0051] S140. When the target LOD enters the corresponding observation area, a temporary takeover process is performed based on the geometric record table to generate a comparison result. The comparison result is then compared item by item with the current judgment result in the geometric record table, and historical consistency is verified.

[0052] Specifically, when a target LOD enters the corresponding observation region and passes the verification preprocessing, the server first performs shadow loading processing on the target LOD to form a shadow participation state, and generates a shadow region representation based on the shadow participation state. The target LOD refers to a candidate layer that has passed the region ownership consistency check, mapping relationship existence check, and observation-holding interval entry condition check; verification preprocessing refers to the preliminary checks performed before formally entering the temporary takeover process to determine whether the target LOD belongs to the current observation region, whether it has a corresponding mapping entry in the geometry record table, and whether it meets the entry timing allowed by the current takeover constraints; the shadow participation state refers to an intermediate state in which the target LOD does not directly rewrite the formal output result of the current layer, but only participates in the region representation reconstruction in an isolated verification path; the shadow region representation refers to the contour representation, facet representation, occlusion representation, and neighborhood representation independently formed by the target LOD around the current observation region in the shadow participation state. In practice, the server first reads the region identifier, object member index, contour mapping item, facet mapping item, occlusion mapping item, neighborhood mapping item, and the takeover constraint field corresponding to the observation-holding interval from the geometry record table. Then, it loads the target LOD into the shadow access bit of the current observation region, allowing it to participate in reconstruction using only the registered object member index and spatial boundary description within the current observation region, without accessing the formal rendering output path. Subsequently, based on the target LOD's contour processing attributes, facet processing attributes, occlusion processing attributes, and neighborhood processing attributes, shadow contour representation, shadow facet representation, shadow occlusion representation, and shadow neighborhood representation are generated within the current observation region, and these four are unified into a shadow region representation. To quantify the overall reconstruction completeness of the current observation region by the target LOD in shadow participation mode, a shadow representation completeness expression can be defined:

[0053] in, Indicates the first The first observation area The shadow representation completeness corresponding to the LOD of the target is as follows: the larger the value, the more completely the region representation of the observed area can be reconstructed when the target LOD is in shadow participation state; It indicates the completeness of the shadow contour representation of the target LOD in the current observation area, and is used to describe the degree of completeness of its reconstruction of the outer boundary, contour closure relationship and the main extension direction of the contour; It indicates the completeness of the shadow facet representation of the target LOD in the current observation area, and is used to describe the degree of completeness of its reconstruction of the main facet position, facet turning relationship and facet succession sequence; It represents the completeness of the shadow occlusion representation of the target LOD in the current observation area, and is used to describe the degree of completeness of its reconstruction of the order of appearance before and after, the position of the occlusion boundary, and the occlusion switching path; It represents the completeness of the shadow neighborhood representation of the target LOD in the current observation area, and is used to describe the degree of completeness of its reconstruction of the contact boundary, surface transition direction and the linkage relationship between adjacent observation areas; , , , These represent the weight coefficients corresponding to the completeness of shadow contour representation, shadow facet representation, shadow occlusion representation, and shadow neighborhood representation, respectively. Their values ​​are set according to the current system's focus on contour reconstruction, facet reconstruction, occlusion reconstruction, and neighborhood reconstruction. The principle behind this expression is to uniformly convert the reconstruction status of the target LOD in the four types of shadow representation into shadow representation completeness, which is used to confirm that the target LOD has the basic representational foundation to enter the subsequent differential extraction.

[0054] After the shadow region representation is formed, the server performs a comparison and result extraction process on the shadow region representation and the current judgment result to form contour difference items, facet difference items, occlusion difference items, and neighborhood difference items. The current determination result refers to the formal region determination result supported by the updated geometric record table of the current layer, which usually includes the current contour determination result, the current facet determination result, the current occlusion determination result, and the current neighborhood determination result. The comparison result extraction process refers to comparing the shadow region expression with the current determination result one by one in the four determination dimensions, and extracting the consistent and different parts between the two. The contour difference item refers to the difference between the shadow contour expression and the current contour determination result in terms of boundary position, boundary closure relationship, and main extension direction of the contour. The facet difference item refers to the difference between the shadow facet expression and the current facet determination result in terms of main facet position, facet turning relationship, and facet succession order. The occlusion difference item refers to the difference between the shadow occlusion expression and the current occlusion determination result in terms of front and back display order, occlusion boundary, and occlusion switching path. The neighborhood difference item refers to the difference between the shadow neighborhood expression and the current neighborhood determination result in terms of contact boundary position, surface transition direction, and linkage relationship. In practice, the server first performs boundary registration and main segment comparison between the shadow contour representation and the current contour determination result, extracting contour boundary offset, contour closure link change, and contour main extension direction deflection to form contour difference items; then, it performs facet positioning and turning relationship comparison between the shadow facet representation and the current facet determination result, extracting main facet offset, facet front-to-back relationship reversal, and continuation sequence perturbation to form facet difference items; subsequently, it performs front-to-back display order and occlusion boundary comparison between the shadow occlusion representation and the current occlusion determination result, extracting occlusion boundary forward movement, backward movement, and switching path change to form occlusion difference items; finally, it performs contact boundary and surface transition comparison between the shadow neighborhood representation and the current neighborhood determination result, extracting boundary misalignment, transition direction change, and linkage breakage to form neighborhood difference items. To quantify the overall deviation of the target LOD from the current determination result, a difference intensity expression can be defined:

[0055] in, Indicates the first The first observation area The overall difference strength between the target LOD and the current judgment result; the larger the value, the more significant the regional expression changes that may be caused after the target LOD enters. Indicates the intensity of contour difference, used to reflect the degree of deviation between the shadow contour representation and the current contour determination result in terms of boundary position, closure relationship, and main extension direction; Indicates the intensity of the facet difference, used to reflect the degree of deviation between the shadow facet representation and the current facet determination result in terms of facet position, turning relationship, and succession order; Indicates the intensity of occlusion difference, which reflects the degree of deviation between the shadow occlusion representation and the current occlusion determination result in terms of the order of appearance before and after, occlusion boundary and switching path; Indicates the intensity of neighborhood difference, used to reflect the degree of deviation between the shadow neighborhood representation and the current neighborhood determination result in terms of contact boundary, surface transition direction and linkage relationship; , , , These represent the weighting coefficients corresponding to the intensity of contour difference, the intensity of planar difference, the intensity of occlusion difference, and the intensity of neighborhood difference, respectively. Their values ​​are set according to the system's sensitivity to different difference types. The principle of this expression is to uniformly convert the four types of difference items into a comprehensive difference intensity, providing a quantitative basis for subsequent item-by-item comparison. However, subsequent judgments do not replace the four types of difference items themselves with this comprehensive difference intensity, but still revolve around the four types of difference items separately.

[0056] After extracting contour differences, facet differences, occlusion differences, and neighborhood differences, the server performs a step-by-step comparison process on these differences based on the contour, facet, occlusion, and neighborhood criteria in the geometric record table, combined with the takeover constraints in the observation hold interval. This process yields step-by-step comparison results. The step-by-step comparison process means that it doesn't just determine whether the target LOD as a whole can replace the current layer, but rather separately determines whether each of the four types of differences violates the existing regional determination criteria for the current observation area and whether it exceeds the allowed takeover boundaries in the observation hold interval. The step-by-step comparison results are the comparison judgment results given for each of the four types of differences, used to characterize whether each type of difference belongs to the allowed difference, warning difference, or out-of-bounds difference category. In practice, the server first checks whether contour differences have altered the position of the main segment of the outer boundary, the contour closure link, or the main extension direction of the contour, based on contour criteria, and then determines whether such changes are allowed within the current observation holding interval, in conjunction with contour takeover constraints. Next, it checks whether facet differences have altered the position of the main facet, the facet turning relationship, and the facet succession order, and determines whether such changes are allowed at the current moment, in conjunction with facet takeover constraints. Then, it checks whether occlusion differences have altered the before-and-after display order, the occlusion boundary position, and the occlusion switching path, and determines whether such changes are allowed under the current occlusion evolution state, in conjunction with occlusion takeover constraints. Finally, it checks whether neighborhood differences have altered the contact boundary position, the surface transition direction, and the linkage relationship, and determines whether unilateral changes are allowed when adjacent observation areas have not yet changed collaboratively, in conjunction with neighborhood takeover constraints. To uniformly measure the acceptability of a certain type of difference relative to the baseline and constraints, a step-by-step comparison fit expression can be defined:

[0057] in, Indicates the first The first observation area Level 1 target LOD in the 1st item-by-item match fit on class differences; where, the first The type of difference item can correspond to the contour difference item, the facet difference item, the occlusion difference item, or the neighborhood difference item in turn; Indicates the target LOD at the 1st position. The degree of fit between the differential items and the geometric record table reference; the higher the degree of fit, the higher the value. Indicates the current observation holding interval. The takeover constraint adaptation value corresponding to the class processing attribute is used to reflect the tolerance level of the class for changes at the current moment. The higher the tolerance level, the higher the value. Indicates the target LOD at the 1st digit. The controllability of the boundary on the class difference item is used to reflect whether the class difference is limited to the local area and whether it can be smoothly absorbed by subsequent step-by-step takeover. The higher the controllability, the higher the value. , , These represent the weighting coefficients corresponding to the baseline fit, the adaptation value of the takeover constraint, and the boundary controllability, respectively. Their values ​​are set based on whether the system prioritizes baseline matching, current temporal tolerance, or local controllability. The principle behind this expression is to uniformly convert the overall adaptation of a certain type of difference relative to existing baselines, current interval constraints, and local controllability into an item-by-item comparison fit, and then categorize this type of difference into different acceptable levels based on the item-by-item comparison fit. The final item-by-item comparison result is not a single conclusion, but rather provides judgments based on four dimensions: contour, facet, occlusion, and neighborhood, providing fractal input for subsequent historical consistency verification.

[0058] After obtaining the item-by-item comparison results, the server performs historical consistency verification processing on the item-by-item comparison results based on the historical decision trajectory in the geometric record table, and generates verification and classification results for passed items, restricted items, and rejected items. The historical decision trajectory refers to the evolution sequence of contour determination, facet determination, occlusion determination, and neighborhood determination of the current observation area that has been sealed in the geometric record table in previous consecutive frames, used to characterize the stable change trend of the observation area in the continuous time dimension; the historical consistency verification processing refers to projecting the current item-by-item comparison results onto the historical decision trajectory to determine whether these changes continue along the existing stable evolution direction or constitute an abrupt break of the existing evolution chain; passed items refer to difference items that do not violate the current geometric record table benchmark and takeover constraints, and are continuous and consistent with the historical decision trajectory; restricted items refer to difference items that, although they do not directly violate the benchmark and takeover constraints, have a potential risk of discontinuity with the historical decision trajectory, and need to be delayed or partially restricted in subsequent functional step-by-step takeover processing; rejected items refer to difference items that have violated the current benchmark and takeover constraints or obviously disrupted the continuity of the historical decision trajectory. In practice, the server first compares the contour comparison results with the historical contour determination trajectory to check whether the current contour change still follows the historical direction of outer contour appearance and closure change. Then, it compares the facet comparison results with the historical facet determination trajectory to check whether the current facet change still follows the main facet exposure sequence and turning evolution direction. Next, it compares the occlusion comparison results with the historical occlusion determination trajectory to check whether the current change in the appearance order remains consistent with the known occlusion entry and exit paths. Finally, it compares the neighborhood comparison results with the neighborhood historical determination trajectory to check whether the current changes in contact boundary and surface transition remain coordinated with the historical linkage of adjacent observation areas. To quantify the consistency between a certain type of comparison results and the historical determination trajectory, a historical consistency expression can be defined:

[0059] in, Indicates the first The first observation area Level 1 target LOD in the 1st The degree of historical consistency on class differences; where, the first The type of difference item can correspond to the contour difference item, the facet difference item, the occlusion difference item, or the neighborhood difference item in turn; This indicates the item-by-item compatibility of the difference item, reflecting its degree of conformity to the benchmark and constraints at the current moment; This indicates the degree to which this type of difference item fits the historical evolution trend, and is used to reflect whether the current change is progressing along the existing stable evolution direction. If the degree of fit is high, the value is higher. This indicates the degree of continuity between this type of difference item and historical stable segments, reflecting whether the current change will interrupt the established stable judgment segment. If the continuity is continuous, this value is higher. , , These represent the weighting coefficients corresponding to item-by-item compatibility, alignment with historical evolution trends, and continuity of historical stable segments. Their values ​​are determined based on the system's emphasis on current rationality, continuity of historical direction, or continuity of historical stability. The principle behind this expression is to uniformly convert the rationality of a certain type of difference under the current comparison with its continuity in the historical trajectory into a degree of historical consistency, and then classify it into approved, restricted, or rejected items based on this degree of historical consistency. The final verification and classification results are essentially a takeover risk distribution formed around the four judgment dimensions, used to directly guide subsequent step-by-step functional takeover processing.

[0060] After generating the verification and classification results for approved, restricted, and rejected items, the server writes the comparison results, item-by-item comparison results, and historical consistency verification results into the geometry record table. Based on these verification and classification results, the server provides a takeover basis for subsequent functional phased takeover processing. The takeover basis refers to the direct judgment foundation provided for subsequent functional phased takeover processing regarding whether the target LOD can enter formal takeover, whether takeover needs to be postponed, and whether takeover should be frozen, based on its contour processing attributes, facet processing attributes, occlusion processing attributes, and neighborhood processing attributes, respectively. In specific implementation, the server establishes a temporary takeover record field in the geometry record table, writes the summary information of the current shadow region expression into the shadow expression subfield, writes the contour difference items, facet difference items, occlusion difference items, and neighborhood difference items into the comparison result subfield, writes the item-by-item comparison results formed around the four types of difference items into the item-by-item comparison subfield, and writes the historical consistency verification results formed around the four types of difference items, along with approved, restricted, and rejected items, into the verification and classification subfield. Simultaneously, the server further marks the corresponding processing functions for each item as takeoverable in the takeover criteria field, marks the processing functions for each restriction item as delayed takeover, and marks the processing functions for each rejection item as frozen takeover. To quantify the overall takeoverability level of the target LOD as it enters the subsequent phased takeover stage, a takeover criteria strength expression can be defined:

[0061] in, Indicates the first The first observation area The takeover of a target LOD is based on its strength; the higher the value, the more suitable the target LOD is for subsequent functions to take over and process in stages. This indicates the combined strength of the pass items for the target LOD across the four categories of differences, reflecting the proportion and stability of its clearly manageable processing capabilities; This indicates the overall strength of the restrictions on the target LOD across the four categories of differences. It reflects the proportion of its processing functions that can be accessed but require controlled takeover and its potential value. The higher the value, the more likely it can be further absorbed through phased takeover. This indicates the overall strength of the rejection items for the target LOD across the four categories of differences. It reflects the proportion and risk level of the processing functions that should not be formally taken over. The higher the value, the greater the overall risk of entry. , , These represent the weighting coefficients corresponding to the comprehensive strength of the approved items, the comprehensive strength of the restricted items, and the comprehensive strength of the rejected items, respectively. Their values ​​are set according to the system's emphasis on the directly manageable portion, the delayed manageable portion, or the degree of risk mitigation in rejection. The principle behind this expression is to uniformly convert the positive and negative impacts of the approved, restricted, and rejected items on the subsequent takeover value into a takeover basis strength, thereby providing a quantitative foundation for the development of takeover qualifications, the arrangement of takeover order, and the freezing of takeover functions in the subsequent phased takeover process.

[0062] S150. After the comparison results are verified to be consistent with the historical data item by item, the target LOD is taken over step by step and the geometric record table is updated item by item, while the original LOD is retained as a backup layer.

[0063] Specifically, after the comparison results are verified for consistency with historical data item by item, the server first performs takeover qualification processing based on the verification and classification results in the geometric record table. This maps the passed, restricted, and rejected items to takeoverable functions, delayed takeover functions, and prohibited takeover functions, respectively. The verification and classification results refer to the set of classification conclusions formed around contour processing attributes, facet processing attributes, occlusion processing attributes, and neighborhood processing attributes after the temporary takeover processing is completed. Passed items indicate that the processing attribute has simultaneously met the current geometric record table benchmark, the observation-maintaining interval takeover constraint, and the historical judgment trajectory continuity requirement, and is qualified to enter the formal takeover process. Restricted items indicate that although the processing attribute does not directly violate the current judgment benchmark, there is still a need for observation in terms of temporal continuity, neighborhood linkage, or local fluctuation amplitude, so it is only suitable for controlled takeover in subsequent stages. Rejected items indicate that if the processing attribute is formally taken over, it will disrupt contour stability, facet stability, occlusion order, or neighborhood continuity, so it cannot enter the formal participation state in this round of takeover cycle. The essence of takeover qualification processing is to transform the verification conclusions formed in the previous verification stage into an executable functional permission structure for the formal takeover stage. This ensures that the target LOD no longer participates in the takeover as a whole, but rather with hierarchical permissions for different processing functions. To quantify the distribution of functional takeover qualifications for the target LOD within the current observation area, a functional takeover qualification strength expression can be defined:

[0064] in, Indicates the first The first observation area Level 1 target LOD in the 1st The strength of takeover eligibility in class processing functions; among which, the first The class processing functions can be sequentially corresponding to the contour processing function, the facet processing function, the occlusion processing function, or the neighborhood processing function; Indicates the first The strength of the pass items corresponding to a class of processing functions reflects the degree to which the class of processing functions has been confirmed as suitable for formal takeover; the more pass items there are, the higher the value. Indicates the first The strength of the restrictions corresponding to a class of processing functions reflects the degree to which such processing functions, although available, still need to be controlled in their implementation. If the restrictions are light and there is room for subsequent absorption, the value is higher. Indicates the first The strength of the rejection item corresponding to the type of processing function is used to reflect the degree of risk that the type of processing function should not be formally taken over in the current cycle. The higher the risk, the higher the value. , , These represent the weighting coefficients corresponding to the strength of the approved item, the strength of the restricted item, and the strength of the rejected item, respectively. Their values ​​are set according to the system's emphasis on confirmed takeoverable portions, deferred absorption portions, or prohibited risk suppression portions. The principle behind this expression is to uniformly convert the positive support and negative risks for a certain type of processing function in the verification and classification results into takeover eligibility strength, and then mark the corresponding processing function as a takeoverable function, a deferred takeover function, or a prohibited takeover function based on the takeover eligibility strength. After this processing is completed, the function eligibility field will be added or updated in the geometry record table, clearly defining the target LOD's formal authority scope for the four types of processing functions.

[0065] After the takeover qualification process is completed, the server performs takeover sequence orchestration based on the observation hold interval, hold requirement field, takeover constraint field, current layer state, and standby layer state in the geometry record table. This determines the takeover order of contour processing attributes, facet processing attributes, occlusion processing attributes, and neighborhood processing attributes. The observation hold interval refers to the time window during which the current observation area needs to maintain a stable decision structure. The hold requirement field refers to the set of decision requirements that should be prioritized for stability within the observation hold interval, typically manifested as the primary hold requirement or parallel primary hold requirement among contour hold requirement, facet hold requirement, occlusion hold requirement, and neighborhood hold requirement. The takeover constraint field specifies the restrictions on whether, when, and how the four types of processing functions are allowed to be taken over in the current interval. The current layer state refers to the LOD level state currently undertaking the main decision-making tasks for the observation area. The standby layer state refers to the support state that can be restored and activated when the original LOD or other bottom-level layers experience an anomaly. The purpose of the takeover sequence arrangement is not to arrange the four types of processing functions according to a fixed template, but to determine which processing functions are suitable for takeover first, which should be takenover later, and which must remain frozen until adjacent areas are completed collaboratively, based on the sensitivity and priority of the current observation area in this round of timing. Generally, processing functions that are more likely to cause overall contour jumps, facet flips, or neighborhood tearing should not be taken over first, while processing functions with relatively localized impacts on overall stability and already possessing high takeover qualifications can be taken over first. To establish a unified evaluation criterion for the takeover sequence, a takeover priority expression can be defined:

[0066] in, Indicates the first The first observation area Level 1 target LOD in the 1st The takeover priority of a class of processing functions; the larger the value, the more suitable the class of processing functions is to be formally taken over first. This indicates the strength of the takeover qualification for this type of processing function, reflecting the entry conditions that this type of function has already met; This indicates the release degree of the primary hold requirement corresponding to this type of processing function. It is used to reflect whether this type of function is no longer the most sensitive priority hold object within the current observation hold interval. The higher the release degree, the higher this value. This indicates the strength of the takeover constraint for this type of processing function, reflecting the degree of restriction imposed on this type of function by the current interval. The stronger the restriction, the higher the value. This indicates the degree of collaborative risk of this type of processing function, reflecting whether premature takeover of this type of function is likely to cause conflicts with other processing functions or adjacent observation areas. The higher the risk, the higher the value. , , , These represent the weighting coefficients corresponding to the strength of takeover eligibility, the degree of release of primary retention requirements, the strength of takeover constraints, and the degree of collaborative risk. Their values ​​are set according to the system's emphasis on takeover feasibility, sensitivity release, timing constraints, or collaborative security. The principle behind this expression is to uniformly convert the takeover maturity, retention release degree, constraint strength, and collaborative risk of a certain type of processing function within the current interval into a takeover priority, and then arrange the takeover order of the four types of processing functions according to the order of their takeover priorities. After processing, a takeover order field will be formed in the geometric record table, providing a direct basis for subsequent initial function imports and progressive takeovers.

[0067] According to the takeover order, the server performs initial function import processing on the target LOD to put the corresponding processing function into a formal participation state, and performs single-function stability verification processing on the observation area. Initial function import processing refers to selecting the first type of processing function with the highest priority and marked as takeoverable from the takeover order field, and promoting the target LOD's supporting status on this type of processing function from shadow participation state to formal participation state, rather than promoting it as a whole to the current layer; formal participation state means that the decision responsibility of this type of processing function in the current observation area has been formally assumed by the target LOD and has begun to affect the current decision results in the geometry record table; single-function stability verification processing refers to immediately verifying whether the formal takeover of a certain type of processing function will cause new impacts on other processing functions that have not yet been taken over, on the current contour or faceted structure, or on occlusion order or neighborhood continuity after the formal import of a certain type of processing function is completed. In practice, the server first selects the first processing function based on the takeover order field. For example, in some scenarios, the facet processing function might be prioritized, or the contour processing function might be prioritized. Then, only the target LOD is granted formal write permissions for that type of processing function, transforming the corresponding shadow region representation sub-item into a formal region representation sub-item. Simultaneously, other processing functions remain under the formal responsibility of the current layer, while the original LOD maintains full support and remains in a standby state. After the import is complete, the server regenerates the corresponding decision items around the single processing function that has been taken over and checks whether these decision items are consistent with the currently untaken items. To quantify the local stability after the formal import of a single function, a single-function stability expression can be defined:

[0068] in, Indicates the first The first observation area Level 1 target LOD in the 1st The stability of a single function after the class processing function is first imported; the larger the value, the more stable the class processing function is after it is officially taken over. This indicates the degree of fit between the imported processing function and the geometric record table reference; the higher the degree of fit, the higher the value. This indicates the degree of coordination between this type of processing function and other unmanaged processing functions after the function is imported; the better the coordination, the higher the value. This indicates the degree of continuity of the historical judgment trajectory after the import of this type of processing function; the more natural the continuity, the higher the value. This indicates the degree of risk of additional disturbances caused by the import of this type of processing function, including the risk of local contour tearing, the risk of misalignment of facets, the risk of occlusion jump, or the risk of neighborhood misalignment. The higher the risk, the higher the value. , , , These represent the weighting coefficients corresponding to the degree of baseline fit, functional coordination, historical continuity, and the degree of additional disturbance risk, respectively. Their values ​​are set according to the system's focus on current accuracy, local coordination, historical continuity, and suppression of additional risks. The principle behind this expression is that when a single-type processing function has just completed its formal takeover, it not only examines whether the function itself is correct, but also whether it remains compatible with the current overall judgment structure, thereby determining whether to allow it to proceed to the next step of takeover and update processing.

[0069] After the single-function stability review is passed, the server performs the first takeover update process on the geometry record table to update the current judgment results and hierarchical responsibility distribution status of the corresponding processing function. The first takeover update process refers to writing the officially approved takeover results from the temporary state to the official record after the initial takeover function is imported, so that the current judgment result field, hierarchical responsibility distribution field, and takeover progress field in the geometry record table are simultaneously updated for the first time. The current judgment result refers to the set of contour judgments, facet judgments, occlusion judgments, and neighborhood judgments that have taken effect in the current observation area at the time of this round of takeover advancement. The hierarchical responsibility distribution status refers to the official responsibility allocation status of the current layer, target LOD, and original LOD on the four types of processing functions. In specific implementation, the server rewrites the judgment items corresponding to the initial takeover processing function to be supported by the target LOD, while adjusting the responsibility originally undertaken by the current layer to a released state, and retaining the rollback mapping of this function for the original LOD in the spare layer support field. Simultaneously, the takeover progress field registers the number of processing functions that have been taken over, the type of processing function, and the corresponding time segment. The advantage of this approach is that, from this moment on, the geometric record table no longer only contains temporary takeover records, but truly reflects the formal progress of the step-by-step takeover, allowing subsequent progressive takeover and phase summary verification to be executed directly based on the updated records.

[0070] After completing the initial takeover update, the server performs a progressive takeover process for the target LOD, repeatedly performing single-function stability verification and geometric record table update after each function is imported. Simultaneously, the original LOD is retained as a backup layer to provide fallback support. The progressive takeover process means that it doesn't just formally import the initial function, but rather proceeds with the takeover of remaining manageable functions and delayed takeover functions in the order given by the takeover sequence field. The progressive aspect means that each step requires three consecutive stages: formal import, stability verification, and geometric record table update. The entry conditions for each step are influenced by the result of the previous takeover and the latest state of the current observation area. If the maintenance requirements, occlusion trends, or neighbor linkage status of the current observation area change after the previous takeover is completed, the order and takeover eligibility of subsequent functions can be adjusted accordingly. The original LOD is continuously retained as a backup layer, meaning that throughout the progressive process, the original LOD is not deleted, nor does it lose its mapping relationship and historical support relationship with the current observation area; rather, it remains in a ready state to resume undertaking any function. To quantify the maturity of the entire progressive takeover process, a progressive takeover maturity expression can be defined:

[0071] in, Indicates the first The first observation area The takeover maturity of a target LOD during the progressive takeover process; the larger the value, the closer the target LOD is to completing a stable handover of all allowed takeover functions. This indicates the percentage of processing functions that have been formally taken over, reflecting the progress of the takeover process; the higher the percentage, the higher this value. This indicates the stability pass rate of the takenover portion, reflecting the stability of the takenover functions during continuous review. The higher the stability, the higher this value. This indicates the consistency of the geometric record table update, reflecting the degree of consistency between the current judgment results, hierarchical responsibility distribution, and takeover progress fields after each round of takeover updates. The stronger the consistency, the higher the value. This indicates the cumulative risk level of the remaining unmanaged functions, reflecting the potential disturbance risks that may arise during the future advancement of the unmanaged parts; the higher the risk, the higher the value. , , , These represent the weighting coefficients for takeover progress, stability verification, record consistency, and remaining risk, respectively. Their values ​​are set based on whether the system prioritizes the progress rate, the stability of the taken-over portion, the integrity of records, or the degree of remaining risk control. The principle behind this expression is to uniformly convert the progress status, stability status, record status, and remaining risk during the progressive takeover process into a takeover maturity level. This maturity level is used to determine whether it is suitable to continue advancing subsequent processing functions, or whether it is necessary to temporarily suspend progress and maintain the original Level of Detail (LOD) to support more responsibilities.

[0072] Once the target LOD has completed the processing functions that allow takeover, the server performs a phase summary verification process. Upon successful verification, a formal takeover completion process is executed, transforming the target LOD into the current layer and the original LOD into a standby layer. Simultaneously, the takeover progress field, historical takeover field, and layer status field are written to the geometry record table. The phase summary verification process means that after the target LOD has completed the progressive handover of all allowed takeover functions, it no longer performs local checks on individual processing functions but instead performs a global consistency check on the overall judgment structure of the current observation area. This checks whether contour judgment, facet judgment, occlusion judgment, and neighborhood judgment remain consistent with the observation interval, the geometry record table baseline, and adjacent observation areas after combination. The formal takeover completion process means that after the phase summary verification passes, the target LOD is promoted from the formal bearer of multiple processing functions to the current layer of the current observation area, the original LOD is formally downgraded from the transitional support layer to a standby layer, and the entire process of this phased takeover is formally recorded in the geometry record table. The "Historical Takeover" field records the takeover sequence, takeover time segments for each function, review conclusions for each round, and the final completion status. The "Hierarchical Status" field records the current layer identity of the target LOD, the backup layer identity of the original LOD, and the waiting status of other candidate layers. In practice, the server first re-aggregates the four current judgment results around the current observation area, checking for contradictions between contours and faces, occlusion and neighborhood, or continuity anomalies between the current and adjacent areas. If the verification passes, the target LOD is marked as the current layer in the "Hierarchical Status" field, the original LOD is marked as the backup layer, and the backtracking mapping and historical support relationships of the original LOD are completely transferred to the backup layer status record. Subsequently, the entire takeover sequence, review conclusions for each function, update time segments of the geometry record table for each round, and the final completion time segment are written into the "Historical Takeover" field, and the takeover progress field is rewritten to the completion status. To quantify the overall reliability after the formal takeover is completed, a formal takeover reliability expression can be defined:

[0073] in, Indicates the first The first observation area The overall reliability of a target LOD after it has been officially taken over; the higher the value, the more suitable it is to be identified as the current layer. This indicates the takeover maturity level after all takeover functions have been completed, reflecting whether the progressive handover process has been sufficiently stable. This indicates the pass rate of the phase summary verification, which reflects the overall consistency level of the combination of the four current judgment results. The higher the consistency, the higher this value. This indicates the integrity of the backup layer, reflecting whether the original LOD still retains its fallback support capability after being converted into a backup layer. The higher the degree of retention, the higher this value. This indicates the residual risk level after the formal transition to the next layer, reflecting the local uncertainties that may still exist for the target LOD after it becomes the current layer. The higher the risk, the higher this value. , , , These represent the weighting coefficients for takeover maturity, phase summary verification pass rate, reserve layer integrity, and residual risk, respectively. Their values ​​are set based on the system's emphasis on handover adequacy, overall consistency, fallback assurance, or residual risk control. The principle behind this expression is to uniformly convert the adequacy of progress, structural consistency, fallback assurance, and residual risk in the final state of formal takeover completion into a comprehensive reliability score, and then use this score to confirm whether the target LOD can truly become the current layer.

[0074] S160. After the target LOD has been taken over, the target LOD is converted to the current layer and the original LOD is downgraded to the standby layer. The rendering result of the current frame is output based on the updated geometry record table.

[0075] Specifically, based on the takeover progress field, current judgment result field, and hierarchical status field in the geometry record table, a takeover completion confirmation process is performed to determine that the target LOD has completed all allowed takeover processing functions and the original LOD is retained as a backup layer. When the takeover completion confirmation result is obtained, the server first reads the takeover progress field, current judgment result field, hierarchical status field, takeover constraint field, and historical takeover field in the geometry record table around the current observation area, and checks the takeover completion status of the target LOD in contour processing, facet processing, occlusion processing, and neighborhood processing functions item by item. The takeover progress field refers to the field used to record the types of processing functions that the target LOD has completed in this round of takeover, the corresponding completion time, and the remaining uncompleted processing functions; the current judgment result field refers to the formal record field of the contour judgment result, facet judgment result, occlusion judgment result, and neighborhood judgment result corresponding to the current observation area at the current moment; the hierarchical status field refers to the field used to record the hierarchical identity, scope of responsibility, and effective status of the target LOD, the original LOD, and other candidate layers in the current observation area. The purpose of the takeover completion confirmation process is to ensure that the target LOD has completed all the processing functions allowed for takeover in this round, and that no processing functions still in a shadow participation state, a delayed takeover state, or a frozen takeover state have been mistakenly written into the official support results. It also confirms that although the original LOD has exited its current layer responsibilities, it is still fully retained as a backup layer. To quantify the takeover completion status, a takeover completion confirmation expression can be defined:

[0076] in, Indicates the first The first observation area The higher the value, the more the target LOD meets the conditions for entering the formal transition stage; This indicates the completion rate of the takeover function, reflecting whether all the processing functions that have been allowed to be taken over in the target LOD in the current observation area have been officially taken over. If all are completed, this value is higher. This indicates the coverage of the current judgment result, reflecting whether the four types of judgment results in the current judgment result field have been completely filled and whether the supporting source is clear. The more complete the coverage, the higher this value will be. This indicates the integrity of the spare layer retention, reflecting whether the original LOD has been registered as a spare layer and whether all fallback support relationships have been retained. The more complete the retention, the higher the value. This indicates the intensity of residual pending states, reflecting whether there are still unreleased delayed takeover functions, frozen takeover functions, or uncleared shadow participation states in the current observation area. The more pending states there are, the higher this value will be. , , , These represent the weighting coefficients corresponding to the allowed takeover completion rate, current judgment result coverage, backup layer retention integrity, and residual unresolved status intensity, respectively. Their values ​​are set based on the system's focus on the completeness of formal takeover, result coverage integrity, rollback guarantee integrity, and suppression of unresolved risks. The principle behind this expression is to uniformly convert whether the target LOD has completed all formal takeovers, whether the formal judgment results are complete, whether the original LOD is retained as a backup layer, and whether there are still unprocessed states in the system into a takeover completion confirmation rate, thus forming a takeover completion confirmation processing result, which serves as a direct prerequisite for subsequent level-based identity switching processing.

[0077] Based on the takeover completion confirmation process, a hierarchical identity switching process is performed to mark the target LOD as the current layer and the original LOD as a backup layer. Simultaneously, when updating the support sources for contour determination, facet determination, occlusion determination, and neighborhood determination, the server formally rewrites the hierarchical responsibility system of the current observation area based on the takeover completion confirmation process. The hierarchical identity switching process refers to elevating the target LOD from the functional bearer of the phased takeover stage to the current layer of the current observation area, while simultaneously downgrading the original LOD from the primary support role of the takeover transition stage to a backup layer role. The current layer refers to the layer that directly undertakes the responsibility of formal region determination and formal frame-level region representation generation in the current observation area. The backup layer refers to the layer that no longer directly undertakes formal output responsibilities but retains recovery capabilities in abnormal rollback, partial recovery, or subsequent comparison and verification. In practice, the server first rewrites the target LOD's identity marker to the current layer and the original LOD's identity marker to the backup layer in the layer status field. Then, it uniformly updates the support sources for contour judgment results, facet judgment results, occlusion judgment results, and neighborhood judgment results in the current judgment result field, ensuring that the formal support sources for all four types of judgment results point to the target LOD. Simultaneously, it migrates all the original formal support relationships of the original LOD in the four processing functions to the backup layer support relationships, preventing the original LOD from implicitly interfering with the formal output. To quantify the convergence degree of responsibility after the layer identity switch, a layer switch convergence expression can be defined:

[0078] in, Indicates the first The first observation area The convergence of responsibilities after the hierarchical identity switch of the target LOD is completed. The larger the value, the clearer the division of responsibilities between the current layer and the backup layer. This indicates the current level of concentration, reflecting whether the four types of formal judgment results have been uniformly undertaken by the target LOD. The higher the level of concentration, the higher this value. This indicates the isolation level of the spare layer, reflecting whether the original LOD has completely exited the formal output responsibility and only retained the fallback responsibility. The more thorough the isolation, the higher the value. This indicates the clarity of the support source, reflecting whether the support source for contour determination, facet determination, occlusion determination, and neighborhood determination is clear and unambiguous in the geometry record table. The higher the clarity, the higher this value. This indicates the degree of risk of overlapping responsibilities, reflecting whether there is still an overlap, gap, or unreleased relationship between the target LOD and the original LOD in terms of formal judgment support. The higher the risk, the higher this value. , , , These represent the weighting coefficients corresponding to the current layer concentration, backup layer isolation, support source clarity, and responsibility overlap risk, respectively. Their values ​​are set based on the system's focus on formal support concentration, backup support isolation, source clarity, and overlap risk mitigation. The principle behind this expression is to uniformly convert the clarity of responsibility boundaries between the formal current layer and the backup layer after identity switching into a layer switching convergence degree, thereby confirming that the layer identity switching has truly been completed, rather than merely a matter of rewriting fields.

[0079] When performing backup layer solidification on the original LOD to preserve its region affiliation, mapping relationship, and historical decision trajectory and write them into the rollback support field, the server does not simply remove the original LOD from the formal output link. Instead, it establishes a stable backup support structure around the original LOD. Backup layer solidification means that after the original LOD is downgraded to a backup layer, it continues to retain its region affiliation, mapping relationship, and historical decision continuity relationship with the current observation area, and systematically writes this information into a structure that can be directly invoked for subsequent rollbacks. The rollback support field is a field in the geometry record table specifically used to record the recoverable range of the backup layer, recovery trigger conditions, recovery priority, and recovery path. The region affiliation identifier refers to the identity marker that the original LOD still maintains a binding relationship with the current observation region; the mapping relationship refers to the contour mapping items, contour mapping items, occlusion mapping items, and neighborhood mapping items established in the previous stage between the original LOD and contour basis, facet basis, occlusion basis, and neighborhood basis; the historical judgment trajectory refers to the contour evolution record, facet evolution record, occlusion evolution record, and neighborhood linkage record left by the original LOD in supporting the judgment of the current observation region in multiple consecutive frames in the past. In specific implementation, the server will freeze the region affiliation identifier and mapping relationship of the original LOD, and will no longer allow it to be cleaned up by subsequent regular optimization processes; at the same time, the recovery availability range of the original LOD in contour processing function, facet processing function, occlusion processing function, and neighborhood processing function is written into the fallback support field, and the anomaly criteria and recovery trigger order required to be met when each type of function is restored are also recorded. In order to quantify the recoverability guarantee of the original LOD as a backup layer, a backup layer solidification strength expression can be defined:

[0080] in, Indicates the first The curing strength of the original LOD as a backup layer in the observation area; the larger the value, the more suitable the original LOD is as a stable backup support in subsequent anomaly handling. This indicates the region attribution retention rate, which reflects whether the binding relationship between the original LOD and the current observation region is completely preserved. The more complete the preservation, the higher the value. This indicates the degree of mapping retention, reflecting whether the mapping relationship between the original LOD and the four types of criteria can still be directly invoked in the subsequent rollback process. The more complete the retention, the higher the value. This indicates the continuity of historical judgment trajectory, reflecting whether the judgment trajectory of the original LOD supporting the current observation area in history is complete and traceable. The stronger the continuity, the higher the value. This indicates the completeness of rollback support, reflecting whether the original LOD has complete rollback support scope, triggering conditions, and recovery paths for the four types of processing functions. The higher the completeness, the higher this value. , , , These represent the weighting coefficients corresponding to the region affiliation retention rate, mapping relationship retention rate, historical judgment trajectory continuity, and rollback support completeness, respectively. Their values ​​are set according to the system's emphasis on identity retention, mapping retention, historical continuity, and rollback completeness. The principle behind this expression is to uniformly convert the value of the original LOD as a backup layer into the backup layer solidification strength from four aspects: structural binding, mapping invocation, historical tracing, and rollback implementation. This ensures that the original LOD is not just a nominal backup layer, but a truly recoverable backup layer.

[0081] Based on the updated geometry record table, the server performs current judgment result merging to form a formal region judgment result for the current observation area. Then, it performs pre-output consistency verification to check internal and neighborhood consistency. During this process, the server aggregates the formal judgment results previously scattered across various subfields into a single, directly outputtable region judgment result. Current judgment result merging refers to recombining the contour, facet, occlusion, and neighborhood judgment results—which have already been formally taken over and written into the geometry record table—to form a complete, formal region judgment result for the current observation area. The formal region judgment result is a comprehensive judgment result within the current observation area that is clearly supported by the current layer's target LOD and can directly guide the generation of the region rendering expression for this frame. Pre-output consistency verification, after the current judgment result merging is complete, performs a unified verification of this comprehensive judgment result. Internal consistency refers to the logical coordination between contour determination results and facet determination results, facet determination results and occlusion determination results, and occlusion determination results and neighborhood determination results within the same observation area. Neighborhood consistency refers to whether the current observation area and adjacent observation areas maintain closed continuity and temporal coordination in terms of contact boundaries, surface transition directions, occlusion boundaries, and linkage relationships. In specific implementation, the server first extracts the current contour determination result, current facet determination result, current occlusion determination result, and current neighborhood determination result, and then performs a unified merging based on spatial boundary descriptions, neighborhood connection descriptions, and object member indexes to form a formal region determination result. Subsequently, an internal consistency check is performed around this formal region determination result to check whether the main contour boundary is coordinated with the main facet position and whether the display relationship before and after occlusion matches the contour and facet relationship. Then, a neighborhood consistency check is performed to check whether there are contact boundary breaks, reversed surface transitions, or neighborhood linkage conflicts between the current observation area and adjacent observation areas. To quantify the overall consistency level at this stage, a pre-output consistency expression can be defined:

[0082] in, Indicates the first The overall consistency of an observation region before output; the larger the value, the more suitable the observation region is to be directly entered into the formal output path. This indicates the internal consistency, reflecting the degree of logical coordination between contour determination, facet determination, occlusion determination, and neighborhood determination within the current observation area. The stronger the coordination, the higher the value. This represents neighborhood consistency, which reflects the degree of coordination between the current observation area and adjacent observation areas in terms of boundary continuity, surface transition, and occlusion order. The stronger the coordination, the higher the value. This indicates the convergence degree of the decision merge, which reflects whether the four current decision results have been unambiguously merged in the formal region decision results. The higher the convergence degree, the higher this value. This indicates the residual risk level of the output, which reflects the risk of local mismatch, neighborhood tearing, or occlusion misalignment that may still exist before the official output. The higher the risk, the higher this value. , , , These represent the weighting coefficients for internal consistency, neighborhood consistency, decision merging convergence, and output residual risk, respectively. Their values ​​are set based on the system's emphasis on internal structural consistency, inter-regional continuity coordination, decision merging sufficiency, and residual risk suppression. The principle behind this expression is to uniformly convert the internal logical state, neighborhood connectivity state, decision merging state, and residual risk state of the currently observed region before formal output into a pre-output consistency state, thereby determining whether to allow subsequent output state write-back processing.

[0083] After the consistency review is passed, the output state write-back process is executed to generate frame-level region representations. Then, frame-level result aggregation processing is performed to form the rendering result of the current frame. At this time, the server begins to transform the confirmed and stable formal region determination results into the actual rendering result of the current frame. Output state write-back process refers to writing the current layer identity, backup layer identity, formal region determination results, and output consistency review conclusions of the current observation region into the output state field of the geometry record table. This ensures that the geometry record table not only saves procedural takeover information but also formal output effectiveness information after the end of the current frame. Frame-level region representation refers to the region-level representation unit generated based on the formal region determination results that can directly participate in the rendering of the current frame. This includes the outline rendering, facet rendering, occlusion rendering, and neighborhood continuation rendering of the current observation region in the current frame. Frame-level result aggregation processing refers to uniformly splicing and integrating the frame-level region representations of each observation region according to spatial relationships, occlusion relationships, and neighborhood relationships to form the final rendering result of the entire scene in the current frame. In practice, the server first writes the current layer identifier, the backup layer identifier, the summary of the formal region determination result, the consistency review conclusion before output, and the effective time segment into the output status field. Then, based on the formal region determination result, it generates a frame-level region representation for the current observation region. Next, according to the spatial boundary description and neighborhood connectivity description of all observation regions, it performs convergence processing on multiple frame-level region representations, ensuring that adjacent regions are closed at boundaries, continuous on surfaces, and coordinated in terms of occlusion, ultimately obtaining the current frame rendering result. To quantify the contribution of a single observation region to the representation of the current frame result, a frame-level region representation strength expression can be defined:

[0084] in, Indicates the first The intensity of the region representation of an observation region in the current frame; the larger the value, the more complete and stable the representation of the observation region in the current frame. This indicates the outline rendering level, reflecting the completeness and clarity of the outer outline of the currently observed area in this frame; It indicates the degree of fold surface presentation, which reflects the completeness of the presentation of the main and subordinate fold surfaces in the current observation area, as well as the degree of discernibility of shape transitions; This indicates the occlusion rendering degree, which reflects the foreground-background display relationship of the currently observed area in this frame and whether the occlusion boundary is accurately represented. Indicates the degree of continuity of the neighborhood, which reflects the degree of continuity between the current observation area and the adjacent observation areas in terms of boundary contact, surface transition and linkage expression; , , , These represent the weighting coefficients for contour rendering, facet rendering, occlusion rendering, and neighborhood continuity rendering, respectively. Their values ​​are set according to the system's focus on boundary visibility, surface shape perception, occlusion accuracy, and regional continuity.

[0085] This application also provides a large-scale 3D model data rendering device based on multi-level LOD organization dynamic scheduling, referencing Figure 2 , Figure 2 This is a schematic diagram of a large-scale 3D model data rendering device based on multi-level LOD organization dynamic scheduling provided in an embodiment of this application. The device is a server, which includes an acquisition module 21 and a processing module 22. The acquisition module 21 is used to acquire the original 3D model data and perform spatial continuity and visible area division processing to form an observation area. At the same time, it extracts contour basis, facet basis, occlusion basis, and neighborhood basis to construct the corresponding geometric record table. The processing module 22 is used to construct multi-level LOD candidate clusters based on the observation area, and perform functional decomposition of contour processing attributes, facet processing attributes, occlusion processing attributes, and neighborhood processing attributes for each level of LOD to establish a mapping relationship with the geometric record table. The processing module 22 is also used to calculate the LOD based on the viewpoint position, viewpoint movement trend, and occlusion change. The processing module 22 is used to generate an observation hold interval for the observation area and write the observation hold interval into the geometric record table as a functional takeover constraint. When the target LOD enters the corresponding observation area, the processing module 22 is also used to perform temporary takeover processing based on the geometric record table to generate a comparison result and compare the comparison result with the current judgment result in the geometric record table item by item and verify historical consistency. The processing module 22 is also used to perform functional step-by-step takeover of the target LOD and update the geometric record table item by item after the comparison result passes the item-by-item comparison and historical consistency verification, while retaining the original LOD as a backup layer. The processing module 22 is also used to convert the target LOD to the current layer and downgrade the original LOD to a backup layer after the target LOD has been taken over, and output the current frame rendering result based on the updated geometric record table.

[0086] It should be noted that the above embodiments of the apparatus are only illustrated by the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the apparatus and method embodiments provided in the above embodiments belong to the same concept, and the specific implementation process can be found in the method embodiments, which will not be repeated here.

[0087] This application also provides an electronic device, with reference to... Figure 3 , Figure 3This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. The electronic device may include: at least one processor 31, at least one network interface 34, a user interface 33, a memory 35, and at least one communication bus 32.

[0088] The communication bus 32 is used to enable communication between these components.

[0089] The user interface 33 may include a display screen and a camera. Optionally, the user interface 33 may also include a standard wired interface and a wireless interface.

[0090] The network interface 34 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface).

[0091] The processor 31 may include one or more processing cores. The processor 31 connects to various parts of the server via various interfaces and lines, executing instructions, programs, code sets, or instruction sets stored in the memory 35, and calling data stored in the memory 35 to perform various server functions and process data. Optionally, the processor 31 may be implemented using at least one hardware form of Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), or Programmable Logic Array (PLA). The processor 31 may integrate one or a combination of several of the following: Central Processing Unit (CPU), Graphics Processing Unit (GPU), and modem. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content to be displayed on the screen; and the modem handles wireless communication. It is understood that the modem may also not be integrated into the processor 31 and may be implemented as a separate chip.

[0092] The memory 35 may include random access memory (RAM) or read-only memory. Optionally, the memory 35 may include a non-transitory computer-readable storage medium. The memory 35 can be used to store instructions, programs, code, code sets, or instruction sets. The memory 35 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the above-described method embodiments, etc.; the data storage area may store data involved in the above-described method embodiments, etc. Optionally, the memory 35 may also be at least one storage device located remotely from the aforementioned processor 31. Figure 3 As shown, the memory 35, which serves as a computer storage medium, may include an operating system, a network communication module, a user interface module, and an application program for rendering large-scale 3D model data based on multi-level LOD organization dynamic scheduling.

[0093] exist Figure 3 In the electronic device shown, the user interface 33 is mainly used to provide an input interface for the user and obtain the user input data; while the processor 31 can be used to call the application stored in the memory 35 for rendering large-scale three-dimensional model data based on multi-level LOD organization dynamic scheduling. When executed by one or more processors, the electronic device executes one or more methods as described in the above embodiments.

[0094] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0095] This application also provides a non-transitory computer-readable storage medium storing instructions. When executed by one or more processors, these instructions cause an electronic device to perform one or more of the methods described in the above embodiments.

[0096] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0097] In the several embodiments provided in this application, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the shown or discussed mutual couplings or direct couplings or communication connections may be through some service interfaces; indirect couplings or communication connections between apparatuses or units may be electrical or other forms.

[0098] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0099] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0100] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned memory includes various media capable of storing program code, such as USB flash drives, portable hard drives, magnetic disks, or optical disks.

[0101] The foregoing description is merely an exemplary embodiment of this disclosure and should not be construed as limiting the scope of this disclosure. Any equivalent changes and modifications made in accordance with the teachings of this disclosure shall still fall within the scope of this disclosure. Those skilled in the art will readily conceive of other embodiments of this disclosure upon considering the specification and the disclosure of practical truth. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not described in this disclosure. The specification and embodiments are considered exemplary only, and the scope and spirit of this disclosure are defined by the claims.

Claims

1. A method for rendering large-scale 3D model data based on dynamic scheduling of multi-level LOD organization, characterized in that, The method includes: The original 3D model data is acquired and spatial continuity and visible area division are performed to form the observation area. At the same time, contour basis, facet basis, occlusion basis and neighborhood basis are extracted to construct the corresponding geometric record table. Based on the observation area, a multi-level LOD candidate cluster is constructed, and the LODs at each level are functionally decomposed into contour processing attributes, facet processing attributes, occlusion processing attributes, and neighborhood processing attributes to establish a mapping relationship with the geometric record table. Based on the viewpoint position, viewpoint movement trend, and occlusion change trend, an observation holding interval for the observation area is generated, and the observation holding interval is written into the geometric record table as a functional takeover constraint. When the target LOD enters the corresponding observation area, a temporary takeover process is performed based on the geometric record table to generate a comparison result. The comparison result is then compared item by item with the current judgment result in the geometric record table, and historical consistency is verified. Once the comparison results are verified to be consistent with the historical data item by item, the target LOD is taken over step by step and the geometric record table is updated item by item, while the original LOD is retained as a backup layer. Once the target LOD has been taken over, it is converted to the current layer and the original LOD is downgraded to a standby layer. The rendering result of the current frame is then output based on the updated geometry record table.

2. The method for rendering large-scale 3D model data based on dynamic scheduling of multi-level LOD organization according to claim 1, characterized in that, The process of acquiring the original 3D model data and performing spatial continuity and visible region segmentation to form an observation area, while extracting contour criteria, facet criteria, occlusion criteria, and neighborhood criteria to construct a corresponding geometric record table, specifically includes: The original 3D model data is processed to form a scene representation set, and spatial continuity recognition is performed based on the scene representation set to form a spatial continuity candidate region. The visible region segmentation process is performed on the spatially continuous candidate region to form the observation region, and the region registration process is performed on the observation region to generate region identifier, spatial boundary description, neighborhood connection description and object member index; The contour basis extraction process, the facet basis extraction process, the occlusion basis extraction process, and the neighborhood basis extraction process are performed around the observation area to form the contour basis, facet basis, occlusion basis, and neighborhood basis, which are then written into the geometry record table together with the region identifier, spatial boundary description, neighborhood connection description, and object member index.

3. The method for rendering large-scale 3D model data based on dynamic scheduling of multi-level LOD organization according to claim 1, characterized in that, The process of constructing a multi-level LOD candidate cluster based on the observation region, and functionally decomposing each level of LOD into contour processing attributes, facet processing attributes, occlusion processing attributes, and neighborhood processing attributes to establish a mapping relationship with the geometric record table, specifically includes: Perform intra-region hierarchical identification processing on the object member index within the observation region to divide it into core-preserving segments and adjustable segments; Based on the contour basis, the facet basis, the occlusion basis, and the neighborhood basis, contour preservation constraints, facet preservation constraints, occlusion preservation constraints, and neighborhood preservation constraints are constructed. Based on the core preservation segment and the adjustable segment, a multi-level progressive construction process is performed under the preservation constraints to generate multi-level LOD candidate clusters. For each level of LOD in the multi-level LOD candidate cluster, extract contour processing attributes, facet processing attributes, occlusion processing attributes, and neighborhood processing attributes, and perform attribute ranking processing. Based on the contour, facet, occlusion, and neighborhood criteria in the geometric record table, and the contour processing attributes, facet processing attributes, occlusion processing attributes, and neighborhood processing attributes in each level of LOD, a class-by-class matching process is performed to establish a mapping relationship.

4. The method for rendering large-scale 3D model data based on dynamic scheduling of multi-level LOD organization according to claim 1, characterized in that, The process of generating an observation holding interval for the observation area based on the viewpoint position, viewpoint movement trend, and occlusion change trend, and writing the observation holding interval into the geometric record table as a functional takeover constraint, specifically includes: Based on the viewpoint position, the viewpoint movement trend, and the occlusion change trend, recognition processing is performed to obtain the regional display status positioning processing result, the continuous observation trend recognition processing result, and the occlusion evolution recognition processing result. The results of the regional display status localization processing, the continuous observation trend recognition processing, and the occlusion evolution recognition processing are subjected to a retention requirement merging process to form the main retention requirement of the observation area. Based on the primary hold requirement, an observation hold interval generation process is performed to determine the starting trigger point and the duration of the observation hold interval, and the functional takeover restrictions within the observation hold interval are determined based on the mapping relationship in the geometric record table. Perform a takeover constraint decomposition process on the observation holding interval to form contour takeover constraints, facet takeover constraints, occlusion takeover constraints, and neighborhood takeover constraints, and write the observation holding interval and the corresponding takeover constraints into the geometry record table.

5. The method for rendering large-scale 3D model data based on dynamic scheduling of multi-level LOD organization according to claim 1, characterized in that, When the target LOD enters the corresponding observation area, a temporary takeover process is performed based on the geometric record table to generate a comparison result. This comparison result is then compared item by item with the current determination result in the geometric record table, and historical consistency is verified. Specifically, this includes: When the target LOD enters the corresponding observation area and passes the verification preprocessing, shadow loading processing is performed on the target LOD to form a shadow participation state, and a shadow region representation is generated based on the shadow participation state. The shadow region representation is compared with the current determination result to extract the results, so as to form contour difference items, facet difference items, occlusion difference items and neighborhood difference items; Based on the contour criteria, facet criteria, occlusion criteria, and neighborhood criteria in the geometric record table, and combined with the takeover constraints in the observation holding interval, the contour difference items, facet difference items, occlusion difference items, and neighborhood difference items are subjected to item-by-item comparison processing to obtain item-by-item comparison results. Based on the historical judgment trajectory in the geometric record table, the item-by-item comparison results are subjected to historical consistency verification processing, and verification and classification results of passed items, restricted items, and rejected items are generated; The comparison results, item-by-item comparison results, and historical consistency verification results are written into the geometric record table, and the verification and classification results provide a basis for subsequent step-by-step takeover of functions.

6. The method for rendering large-scale 3D model data based on dynamic scheduling of multi-level LOD organization according to claim 1, characterized in that, After the comparison results are verified to be consistent with historical data item by item, the target LOD is taken over step by step and the geometric record table is updated item by item, while the original LOD is retained as a backup layer. Specifically, this includes: Based on the verification and classification results in the geometric record table, takeover qualification expansion processing is performed to map the pass items, restriction items, and rejection items to takeoverable functions, delayed takeover functions, and prohibited takeover functions, respectively. Based on the observation holding interval, holding requirement field, takeover constraint field, current layer state and standby layer state in the geometry record table, the takeover order arrangement process is performed to determine the takeover order of contour processing attributes, facet processing attributes, occlusion processing attributes and neighborhood processing attributes. According to the takeover sequence, the target LOD is processed to perform the first-connection function import process so that the corresponding processing function enters the formal participation state, and the observation area is processed to perform single-function stability verification. After the single-function stability review is passed, the geometric record table is subjected to the first takeover update process to update the current judgment result and hierarchical responsibility distribution status of the corresponding processing function; The target LOD is subjected to subsequent progressive takeover processing, and after each processing function is imported, the single-function stability verification processing and geometry record table update processing are repeatedly performed, while the original LOD is retained as a backup layer to provide rollback support. After the target LOD completes the takeover permission process, the execution phase summary verification process is performed, and the formal takeover completion process is performed when the verification is passed, so as to convert the target LOD into the current layer and the original LOD into the standby layer. At the same time, the takeover progress field, historical takeover field and layer status field are written into the geometry record table.

7. The method for rendering large-scale 3D model data based on dynamic scheduling of multi-level LOD organization according to claim 1, characterized in that, Once the target LOD has been taken over, the target LOD is converted to the current layer and the original LOD is downgraded to a standby layer. The current frame rendering result is then output based on the updated geometry record table. Specifically, this includes: Based on the takeover progress field, current judgment result field, and level status field in the geometry record table, the takeover completion confirmation process is performed to determine that the target LOD has completed all the processing functions that allow takeover and the original LOD is retained as a spare layer, and the takeover completion confirmation process result is obtained. Based on the confirmation result of the takeover completion, a hierarchical identity switching process is performed to mark the target LOD as the current layer and the original LOD as the backup layer, while updating the support sources for contour determination, facet determination, occlusion determination, and neighborhood determination. The original LOD is subjected to backup layer solidification processing to retain the region ownership identifier, mapping relationship and historical judgment trajectory and write them into the rollback support field; Based on the updated geometric record table, the current judgment result is merged to form the formal region judgment result of the current observation area, and the pre-output consistency review is performed to verify internal consistency and neighborhood consistency. After the consistency check is passed, the output status is written back and a frame-level region representation is generated. Then, the frame-level result aggregation process is performed to form the rendering result of the current frame.

8. A large-scale 3D model data rendering device based on multi-level LOD organization dynamic scheduling, characterized in that, The apparatus is used to execute the large-scale 3D model data rendering method based on multi-level LOD organization dynamic scheduling as described in any one of claims 1 to 7, the apparatus comprising an acquisition module and a processing module, wherein... The acquisition module is used to acquire the original three-dimensional model data and perform spatial continuity and visible area division processing to form an observation area. At the same time, it extracts contour basis, fold basis, occlusion basis and neighborhood basis to construct the corresponding geometric record table. The processing module is used to construct a multi-level LOD candidate cluster based on the observation area, and to perform functional decomposition of each level of LOD into contour processing attributes, facet processing attributes, occlusion processing attributes and neighborhood processing attributes, so as to establish a mapping relationship with the geometric record table. The processing module is also used to generate an observation holding interval for the observation area based on the viewpoint position, viewpoint movement trend and occlusion change trend, and write the observation holding interval into the geometric record table as a function takeover constraint. The processing module is also used to perform temporary takeover processing based on the geometric record table when the target LOD enters the corresponding observation area, so as to generate a comparison result, and compare the comparison result with the current judgment result in the geometric record table item by item and verify historical consistency. The processing module is also used to take over the target LOD step by step and update the geometric record table item by item after the comparison result is verified to be consistent with the historical data item by item, while retaining the original LOD as a backup layer. The processing module is further configured to, after the target LOD has been taken over, convert the target LOD to the current layer and downgrade the original LOD to a standby layer, and output the rendering result of the current frame based on the updated geometry record table.

9. An electronic device, characterized in that, The electronic device includes a processor, a memory, a user interface, and a network interface. The memory is used to store instructions. The user interface and the network interface are both used to communicate with other devices. The processor is used to execute the instructions stored in the memory to cause the electronic device to perform the method as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium, characterized in that, The non-transitory computer-readable storage medium stores instructions that, when executed, perform the method as described in any one of claims 1 to 7.

Images