Cross-rendering engine fusion method based on nonlinear coordinate correction

Abstract

The application discloses a cross-rendering engine fusion method based on nonlinear coordinate correction, relates to the technical field of cross-platform rendering, and comprises the following steps: in response to a cross-engine scene switching instruction, performing linear coordinate conversion on original coordinates of a to-be-loaded model in a geographic space engine to obtain theoretical target coordinates of the to-be-loaded model under a coordinate system of a target rendering engine; determining a nonlinear compensation value corresponding to the theoretical target coordinates based on a nonlinear correction model; superimposing the theoretical target coordinates and the nonlinear compensation value to obtain target coordinates of the to-be-loaded model in the target rendering engine; and loading the to-be-loaded model in the target rendering engine based on the target coordinates. The application realizes high-precision spatial alignment between different rendering engines by constructing a mixed correction mechanism of linear coordinate conversion and nonlinear compensation, effectively solves the visual misalignment problem caused by the difference between coordinate systems and nonlinear distortion, and significantly improves the visual coherence and spatial positioning accuracy of cross-engine scene switching.

Description

Technical Field

[0001] This application relates to the field of cross-platform rendering technology, and in particular to a cross-rendering engine fusion method based on nonlinear coordinate correction. Background Technology

[0002] In digital twin scenarios, the integration of macroscopic geospatial data with mesoscopic and microscopic scenes typically requires the collaborative use of multiple rendering engines. Due to the differences in coordinate systems employed by each rendering engine, linear coordinate transformation models are currently commonly used for cross-engine spatial data alignment. However, due to variations in rendering precision and model material processing across different engines, the coordinate transformation process introduces nonlinear biases related to spatial location and geometry. These biases cannot be effectively compensated for by linear models, leading to spatial position shifts in the model during cross-engine switching, thereby affecting the overall consistency of the scene and the accuracy of spatial analysis.

[0003] The above content is only used to help understand the technical solution of this application and does not represent an admission that the above content is prior art. Summary of the Invention

[0004] The main purpose of this application is to provide a cross-rendering engine fusion method based on nonlinear coordinate correction, which aims to solve the technical problem of visual misalignment of models caused by the difference in coordinate systems and nonlinear distortion between different rendering engines.

[0005] To achieve the above objectives, embodiments of this application provide a cross-rendering engine fusion method based on nonlinear coordinate correction, the cross-rendering engine fusion method based on nonlinear coordinate correction comprising:

[0006] In response to the cross-engine scene switching command, the original coordinates of the model to be loaded in the geospatial engine are transformed into linear coordinates to obtain the theoretical target coordinates of the model to be loaded in the target rendering engine coordinate system.

[0007] Based on the pre-constructed nonlinear correction model, the nonlinear compensation value corresponding to the theoretical target coordinates is determined;

[0008] The theoretical target coordinates are superimposed with the nonlinear compensation value to obtain the target coordinates of the model to be loaded in the target rendering engine;

[0009] Based on the target coordinates, the model to be loaded is loaded into the target rendering engine.

[0010] In one embodiment, before the step of determining the nonlinear compensation value corresponding to the theoretical target coordinates based on a pre-built nonlinear correction model, the cross-rendering engine fusion method based on nonlinear coordinate correction further includes:

[0011] Obtain multiple feature anchor points in the model to be loaded, and determine the first coordinates of the feature anchor points in the geospatial engine and the second coordinates in the target rendering engine;

[0012] Based on the linear coordinate transformation model, the first coordinate is converted into the corresponding theoretical target coordinate, and the nonlinear deviation vector corresponding to each feature anchor point is calculated according to the deviation between the theoretical target coordinate and the second coordinate.

[0013] Based on the first coordinates corresponding to the feature anchor points and the nonlinear deviation vector, the nonlinear correction model is constructed. The nonlinear correction model is used to output the corresponding nonlinear compensation value according to the input spatial coordinates.

[0014] In one embodiment, the step of obtaining multiple feature anchor points in the model to be loaded includes:

[0015] Edge feature points are extracted from the model to be loaded using an edge detection algorithm;

[0016] Calculate the number of neighboring geometric points within a preset neighborhood radius for each edge feature point, and determine candidate feature anchor points based on the number of neighboring geometric points;

[0017] The candidate feature anchors are matched with the model structure feature library, and the feature anchors are finally determined based on the matching results.

[0018] In one embodiment, the step of constructing the nonlinear correction model based on the first coordinates corresponding to the feature anchor point and the nonlinear deviation vector, wherein the nonlinear correction model is used to output the corresponding nonlinear compensation value according to the input spatial coordinates, includes:

[0019] Using the first coordinates corresponding to each of the feature anchor points as input vectors and the corresponding nonlinear deviation vectors as output targets, a deviation training sample set is constructed.

[0020] The Gaussian process regression algorithm is used to spatially fit the bias training sample set to obtain the nonlinear correction model.

[0021] In one embodiment, prior to the steps of acquiring multiple feature anchor points in the model to be loaded and determining the first coordinates of the feature anchor points in the geospatial engine and the second coordinates in the target rendering engine, the cross-rendering engine fusion method based on nonlinear coordinate correction further includes:

[0022] An engine abstraction adaptation layer is built by providing standardized interfaces with different types of rendering engines through a standardized interface layer;

[0023] By calling the corresponding standardized interface through the engine abstraction adaptation layer, the coordinate data of the feature anchor point in the geospatial engine and the target rendering engine are obtained, and the coordinate data includes the first coordinate and the second coordinate.

[0024] In one embodiment, the cross-rendering engine fusion method based on nonlinear coordinate correction further includes:

[0025] In response to the cross-engine scene switching command, the current view parameters of the geospatial engine are obtained;

[0026] Based on the current viewpoint parameters, determine the level of detail and the range of region segments for the model to be loaded;

[0027] Based on the target coordinates, the level of detail, and the region segmentation range, the model to be loaded is divided into multiple data segments;

[0028] The data fragments are loaded into the target rendering engine according to the preset loading priority.

[0029] In one embodiment, the step of loading the data fragments in the target rendering engine according to a preset loading priority includes:

[0030] Based on the region sharding range, the multiple data shards are divided into core region shards and non-core region shards;

[0031] The core region fragment is assigned a higher loading priority than the non-core region fragment, so that the core region fragment is loaded first.

[0032] In one embodiment, after the step of assigning a higher loading priority to the core region fragment than to the non-core region fragment to prioritize loading the core region fragment, the cross-rendering engine fusion method based on nonlinear coordinate correction further includes:

[0033] Receive the loading completion status of the core region fragments from the target rendering engine;

[0034] When the loading completion rate reaches a preset threshold, a smooth transition animation is initiated to complete the cross-engine switch from the geospatial engine to the target rendering engine.

[0035] During the execution of the smooth transition animation, the non-core region fragments continue to be loaded in the background.

[0036] In one embodiment, after the step of loading the model to be loaded in the target rendering engine based on the target coordinates, the cross-rendering engine fusion method based on nonlinear coordinate correction further includes:

[0037] Obtain the actual rendering coordinates of multiple feature anchor points in the target rendering engine of the model to be loaded;

[0038] The actual rendering coordinates are compared with the target coordinates corresponding to the feature anchor points to obtain the positional deviation of each feature anchor point;

[0039] When any of the aforementioned positional deviations exceeds a preset deviation threshold, a correction recalculation process is triggered, the correction recalculation process including:

[0040] By using the engine abstraction adaptation layer, the first coordinates of the feature anchor point in the geospatial engine and the second coordinates in the target rendering engine are re-acquired.

[0041] Based on the first coordinate and the second coordinate, the nonlinear deviation vector corresponding to the feature anchor point is recalculated, and the local compensation value of the nonlinear correction model is updated.

[0042] Based on the local compensation value, incremental coordinate correction is performed on the region of deviation exceeding the limit in the model to be loaded.

[0043] In one embodiment, after the step of loading the model to be loaded in the target rendering engine based on the target coordinates, the cross-rendering engine fusion method based on nonlinear coordinate correction further includes:

[0044] Delete the uncorrected model data corresponding to the model to be loaded from the geospatial engine;

[0045] Update the engine status identifier in the state synchronization layer and mark the currently active engine as the target rendering engine.

[0046] This application embodiment also provides a cross-rendering engine fusion device based on nonlinear coordinate correction. The cross-rendering engine fusion device based on nonlinear coordinate correction includes: a memory, a processor, and a computer program stored in the memory and executable on the processor. The computer program is configured to implement the steps of the cross-rendering engine fusion method based on nonlinear coordinate correction as described above.

[0047] This application embodiment also provides a storage medium, which is a computer-readable storage medium, and stores a computer program thereon. When the computer program is executed by a processor, it implements the steps of the cross-rendering engine fusion method based on nonlinear coordinate correction as described above.

[0048] One or more technical solutions proposed in this application have at least the following technical effects:

[0049] This application constructs a hybrid correction mechanism combining linear coordinate transformation and nonlinear compensation. First, during cross-engine scene switching, the original coordinates in the geospatial engine are converted into theoretical target coordinates in the target rendering engine's coordinate system based on linear coordinate transformation, establishing a cross-engine spatial benchmark and solving the macroscopic alignment problem between different rendering engine coordinate systems. Based on this, a nonlinear correction model trained on feature anchor points is introduced to determine the nonlinear compensation value corresponding to the theoretical target coordinates, achieving precise compensation for the local distortion remaining from the linear coordinate transformation and effectively eliminating visual misalignment caused by projection distortion and differences in engine precision. Furthermore, the target coordinates are obtained by superimposing the theoretical target coordinates with the nonlinear compensation value, and the model to be loaded is loaded into the target rendering engine based on this, achieving precise spatial mapping from the geospatial engine to the high-resolution rendering engine. Ultimately, this achieves high-precision visual coherence and spatial consistency during cross-engine scene fusion. Attached Figure Description

[0050] Figure 1 This is a flowchart illustrating the first embodiment of the cross-rendering engine fusion method based on nonlinear coordinate correction involved in the embodiments of this application;

[0051] Figure 2 This is a flowchart illustrating the second embodiment of the cross-rendering engine fusion method based on nonlinear coordinate correction involved in the embodiments of this application;

[0052] Figure 3 This is a schematic diagram of the overall architecture of the cross-rendering engine fusion method based on nonlinear coordinate correction involved in the embodiments of this application;

[0053] Figure 4 This is a flowchart illustrating the third embodiment of the cross-rendering engine fusion method based on nonlinear coordinate correction involved in the embodiments of this application;

[0054] Figure 5 This is a flowchart illustrating the fourth embodiment of the cross-rendering engine fusion method based on nonlinear coordinate correction involved in the embodiments of this application;

[0055] Figure 6 This is a schematic diagram of the cross-rendering engine fusion device based on nonlinear coordinate correction involved in the embodiments of this application.

[0056] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0057] It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit this application.

[0058] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0059] In digital twin scenarios, the integration of macroscopic geospatial data with mesoscopic and microscopic scenes typically requires the collaborative use of multiple rendering engines. Due to the differences in coordinate systems employed by each rendering engine, linear coordinate transformation models are currently commonly used for cross-engine spatial data alignment. However, due to variations in rendering precision and model material processing across different engines, the coordinate transformation process introduces nonlinear biases related to spatial location and geometry. These biases cannot be effectively compensated for by linear models, leading to spatial position shifts in the model during cross-engine switching, thereby affecting the overall consistency of the scene and the accuracy of spatial analysis.

[0060] In view of the above problems, this application proposes a cross-rendering engine fusion method based on nonlinear coordinate correction. In response to a cross-engine scene switching command, the original coordinates of the model to be loaded in the geospatial engine are linearly transformed to obtain the theoretical target coordinates of the model in the target rendering engine's coordinate system. Based on a pre-built nonlinear correction model, a nonlinear compensation value corresponding to the theoretical target coordinates is determined. The theoretical target coordinates are then superimposed with the nonlinear compensation value to obtain the target coordinates of the model to be loaded in the target rendering engine. Based on the target coordinates, the model to be loaded is loaded in the target rendering engine.

[0061] This application provides a solution that constructs a hybrid correction mechanism combining linear coordinate transformation and nonlinear compensation. First, during cross-engine scene switching, the original coordinates in the geospatial engine are converted into theoretical target coordinates in the target rendering engine's coordinate system based on linear coordinate transformation, establishing a cross-engine spatial benchmark and solving the macroscopic alignment problem between different rendering engine coordinate systems. Based on this, a nonlinear correction model trained on feature anchor points is introduced to determine the nonlinear compensation value corresponding to the theoretical target coordinates, achieving precise compensation for the local distortion remaining from the linear coordinate transformation and effectively eliminating visual misalignment caused by projection distortion and differences in engine precision. Furthermore, the target coordinates are obtained by superimposing the theoretical target coordinates with the nonlinear compensation value, and the model to be loaded is loaded into the target rendering engine based on this, achieving precise spatial mapping from the geospatial engine to the high-resolution rendering engine. Ultimately, this achieves high-precision visual coherence and spatial consistency during cross-engine scene fusion.

[0062] It should be noted that the executing entity in this embodiment can be a computing service device with capabilities for three-dimensional spatial data processing, multi-engine communication coordination, and real-time rendering scheduling, such as a local terminal, graphics workstation, cloud computing platform, graphics processing server, etc., or an electronic device or cross-platform rendering system capable of implementing the above functions. The following description uses a cross-platform rendering system as an example to illustrate this embodiment and the subsequent embodiments.

[0063] The first embodiment of the cross-rendering engine fusion method based on nonlinear coordinate correction proposed in this application is described in reference [link to relevant documentation]. Figure 1 The method includes steps S10 to S40:

[0064] Step S10: In response to the cross-engine scene switching command, perform linear coordinate transformation on the original coordinates of the model to be loaded in the geospatial engine to obtain the theoretical target coordinates of the model to be loaded in the target rendering engine coordinate system.

[0065] It should be noted that the cross-engine scene switching command is a signal that triggers the switching of the visualization scene from the current geospatial engine to the target rendering engine. It can be triggered automatically by the system (such as in the current view parameters of the geospatial engine, when the view distance meets the preset distance) or manually by the user (such as clicking the switch button).

[0066] Geospatial engines are engines that support spatial data processing and visualization in geographic coordinate systems, such as Cesium and three.js, and can output standard geographic coordinates such as WGS84.

[0067] Target rendering engines refer to rendering platforms that focus on high-fidelity visual effects and physical simulation, such as UE (Unreal Engine) or Unity, which typically use a local orthogonal coordinate system.

[0068] As a feasible implementation method, linear coordinate transformation can be carried out using a seven-parameter coordinate transformation method, a four-parameter transformation method, or a custom transformation matrix. The specific choice can be adapted according to the differences in coordinate systems across engine combinations. For example, when switching between Cesium and UE, the seven-parameter method is used to ensure the basic accuracy of the transformation, while when switching between three.js and Unity, a simplified four-parameter transformation method can be used to improve the transformation efficiency.

[0069] Understandably, due to the fundamental differences between cross-engine coordinate systems (geographic coordinate system and local orthogonal coordinate system), direct loading would cause a serious shift in the model's position. Therefore, by first performing linear coordinate transformation, we can avoid excessive basic deviations in subsequent nonlinear corrections, lay the initial coordinate foundation for centimeter-level alignment, and thus improve the overall accuracy and stability of cross-engine coordinate transformation.

[0070] Step S20: Based on the pre-built nonlinear correction model, determine the nonlinear compensation value corresponding to the theoretical target coordinates.

[0071] It should be noted that the nonlinear correction model is a mathematical model trained using the deviation data of feature anchor points to fit the nonlinear deviation pattern across engines. Its core function is to output the corresponding nonlinear deviation compensation amount based on the input coordinate position. The nonlinear compensation value is a correction amount used for the theoretical target coordinates to offset the coordinate offset that cannot be eliminated by linear transformation, and is usually a three-dimensional vector.

[0072] Optionally, a coordinate transformation library based on preset compensation rules can be used to replace the nonlinear correction model. Specifically, a linear transformation from the WGS84 coordinate system to the local orthogonal coordinate system of the target rendering engine is achieved by integrating the Proj.4 coordinate transformation library, while a manual preset compensation mechanism is established to handle nonlinear deviations. This solution pre-determines the nonlinear deviation patterns of different scene types (such as urban parks, roads and bridges, and indoor scenes) through manual measurement and statistical analysis, and stores the corresponding compensation values ​​in the scene configuration file in the form of three-dimensional vectors. In practical applications, the corresponding scene configuration file is automatically called according to the current scene type, and the preset compensation values ​​are superimposed with the linear transformation results to obtain the final coordinate output.

[0073] It should be noted that Proj.4 is a mature open-source coordinate transformation library that supports transformation calculations between various geodetic coordinate systems and projected coordinate systems. In this scheme, it is mainly used to achieve basic spatial benchmark unification. The manually preset compensation value is a fixed compensation amount determined in advance based on historical engineering experience or sampled measurement data. Its accuracy depends on the comprehensiveness of the previous measurements and the rationality of the scene division.

[0074] In this embodiment, a nonlinear correction model can be constructed using the Gaussian Process Regression (GPR) algorithm, with the theoretical target coordinates of the feature anchor points as input samples and the measured deviation vector as output labels for training.

[0075] Specifically, steps S201 to S203 precede step S20:

[0076] Step S201: Obtain multiple feature anchor points in the model to be loaded, and determine the first coordinates of the feature anchor points in the geospatial engine and the second coordinates in the target rendering engine.

[0077] It should be noted that feature anchor points are key geometric points with spatial uniqueness and structural stability extracted from the model to be loaded. Examples include building corner vertices, road intersection center points, lane line intersections, road edge endpoints, and building ridge vertices. These serve as the core benchmarks for constructing deviation relationships and nonlinear correction models. The first coordinate is the geographic coordinate of the feature anchor point in the geospatial engine (such as latitude, longitude, and elevation in the WGS84 coordinate system), possessing global spatial reference attributes. The second coordinate is the actual coordinate of the model to be loaded in the target engine, obtained through measurement or the engine interface, under the local orthogonal coordinates of the same feature anchor point.

[0078] As one possible implementation, step S201 includes steps S2010 to S2012:

[0079] Step S2010: Extract edge feature points from the model to be loaded using an edge detection algorithm.

[0080] Step S2011: Calculate the number of neighboring geometric points within a preset neighborhood radius for each edge feature point, and determine candidate feature anchor points based on the number of neighboring geometric points.

[0081] Step S2012: Match the candidate feature anchors with the model structure feature library, and finally determine the feature anchors based on the matching results.

[0082] In this embodiment, the Canny edge detection algorithm is first used to extract edge features, obtaining geometrically significant edge feature points. Then, for each edge feature point, the number of neighboring geometric points within its preset neighborhood radius is counted, and a minimum density threshold is set. When the number of neighboring geometric points is not lower than this minimum density threshold, the edge feature point is determined to be located in a densely structured region, possessing sufficient geometric stability, and is thus retained as a candidate feature anchor point to exclude isolated or noise-sensitive edge points. Finally, the candidate feature anchor points are matched with a pre-built model structure feature library. This library is constructed based on statistical analysis of a large number of similar scene samples and includes various structural templates such as corner, intersection, endpoint, and vertex types. Each template defines a corresponding geometric parameter threshold range and associated feature descriptors.

[0083] During the matching process, the similarity of feature descriptors and the conformity of geometric parameters are calculated comprehensively, and based on the matching score, feature anchor points with spatial uniqueness and structural stability are finally determined.

[0084] Specifically, based on the anchor type of the candidate feature anchor, the corresponding target template is determined from the model structure feature library; the SIFT descriptor of the candidate feature anchor in a preset local neighborhood is obtained, and the Euclidean distance or cosine similarity between the SIFT descriptor and the template descriptor associated with the target template is calculated as the feature matching similarity; if the feature matching similarity meets the preset feature matching threshold, the geometric parameters of the candidate feature anchor are further extracted, including the edge angle, the distance ratio between the two sides of the intersection point, and the number of intersecting lines, and it is determined whether the geometric parameters fall within the corresponding parameter threshold range defined by the target template, thereby determining the geometric parameter conformity; when both the feature matching similarity and the geometric parameter conformity meet their respective preset conditions, the candidate feature anchor is determined to be successfully matched with the target template and is identified as a feature anchor.

[0085] For example, anchor point types include building corner vertices, road intersection center points, building ridge vertices, and road edge endpoints. Corner point templates include building corner vertices and door / window opening corner points; intersection point templates include road intersection center points and lane line intersections; endpoint templates include bridge expansion joint endpoints and road edge endpoints; and vertex templates include building ridge vertices and bridge pier vertices.

[0086] Step S202: Based on the linear coordinate transformation model, the first coordinate is converted into the corresponding theoretical target coordinate, and the nonlinear deviation vector corresponding to each feature anchor point is calculated according to the deviation between the theoretical target coordinate and the second coordinate.

[0087] It should be noted that the nonlinear deviation vector is the difference vector between the second coordinate and the theoretical target coordinate. It can accurately quantify the degree of nonlinear distortion that cannot be covered by linear transformation. Its three-dimensional components correspond to the deviation values ​​of the X, Y, and Z axes, respectively.

[0088] In this embodiment, the theoretical target coordinates obtained by transforming the first coordinates corresponding to the feature anchor point through a linear coordinate transformation model are subtracted from the second coordinates actually measured in the target rendering engine for the same feature anchor point. That is, the nonlinear deviation vector = local orthogonal coordinates - target theoretical coordinates. This nonlinear deviation vector quantitatively characterizes the coordinate deviation caused by nonlinear factors that still exists after the linear transformation.

[0089] For example, the WGS84 coordinates (X1, Y1, Z1) of a feature anchor point in the geospatial engine, and the local orthogonal coordinates (X2, Y2, Z2) of the feature anchor point in the target rendering engine are obtained. After linear coordinate transformation, the theoretical target coordinates (X2', Y2', Z2') in the target rendering engine coordinate system are obtained. Based on this, the nonlinear deviation vector of the feature anchor point is calculated as: Nonlinear deviation vector = (X2 - X2', Y2 - Y2', Z2 - Z2').

[0090] Step S203: Based on the first coordinates corresponding to the feature anchor points and the nonlinear deviation vector, construct the nonlinear correction model. The nonlinear correction model is used to output the corresponding nonlinear compensation value according to the input spatial coordinates.

[0091] In this embodiment, the nonlinear correction model can output the corresponding nonlinear deviation compensation value in real time according to the input spatial location.

[0092] As one possible implementation, step S203 includes steps S2030 to S2031:

[0093] Step S2030: Use the first coordinates corresponding to each feature anchor point as the input vector and the corresponding nonlinear deviation vector as the output target to construct a deviation training sample set.

[0094] Step S2031: Use the Gaussian process regression algorithm to spatially fit the bias training sample set to obtain the nonlinear correction model.

[0095] In this embodiment, after calculating the nonlinear deviation vector corresponding to each set of feature anchor points, the first coordinates corresponding to each feature anchor point are used as the input vector, and the corresponding nonlinear deviation vector is used as the output target to construct a deviation training sample set containing spatial location and nonlinear deviation vector. Subsequently, the Gaussian process regression algorithm is used to spatially fit the deviation training sample set to construct a nonlinear correction model.

[0096] Gaussian process regression, as a non-parametric Bayesian learning method, can effectively handle spatial interpolation problems and provide a measure of prediction uncertainty. Specifically, it takes the first coordinate of the feature anchor point as input and the corresponding nonlinear deviation vector as the output target, and completes model training by optimizing the kernel function hyperparameters. The resulting nonlinear correction model has spatial reasoning capabilities; when new spatial coordinates are input, the model can output corresponding nonlinear compensation values ​​based on the principle of spatial correlation.

[0097] Alternatively, radial basis function (RBF) interpolation, Kriging spatial interpolation, or lightweight neural networks (such as MLPs) can be used as alternative modeling methods for nonlinear correction models. To optimize model storage and computational efficiency, the continuous correction field generated by the above algorithms can be discretized into a multi-resolution grid structure (such as an octree structure) for storage and retrieval. Specifically, firstly, a multi-resolution grid is dynamically constructed based on the spatial distribution density of feature anchor points: grid subdivision is implemented in densely populated areas to improve local representation accuracy, while a coarser grid granularity is maintained in sparse areas to save resources. Next, the nonlinear deviation vector of each feature anchor point is directly mapped to the correction parameter of its corresponding grid vertex. For grid vertices without feature anchor point coverage, the mapping is completed by interpolating the deviation data of neighboring anchor points, thereby establishing a complete mapping relationship between grid vertices and correction parameters.

[0098] When it is necessary to obtain the nonlinear compensation value at any spatial location, the corresponding nonlinear compensation value can be quickly output by locating the grid cell where the target location is located, calling the correction parameters stored in the vertices of its neighboring grid cells, and calculating through linear interpolation.

[0099] This structured discrete representation method based on multi-resolution grids effectively reduces the model storage size while maintaining spatial interpolation accuracy and significantly improves the computational efficiency of nonlinear compensation value lookup, achieving a good balance between correction accuracy and runtime performance.

[0100] Step S30: Superimpose the theoretical target coordinates with the nonlinear compensation value to obtain the target coordinates of the model to be loaded in the target rendering engine.

[0101] It should be noted that the target coordinates are the final coordinates of the model to be loaded in the target rendering engine used for visualization. They are the coordinate results after eliminating linear and nonlinear deviations, and can achieve height alignment with the model's position in the geospatial engine.

[0102] This embodiment generates a final, hybrid-corrected coordinate system by vector superimposing the macroscopic alignment result of linear coordinate transformation with the microscopic correction of nonlinear correction. This final coordinate system can be directly used by the target rendering engine. The target coordinate system contains the comprehensive correction amount after eliminating nonlinear bias and random distortion, and serves as the direct basis for ensuring high-precision spatial positioning of the model to be loaded in the target rendering engine.

[0103] Step S40: Based on the target coordinates, load the model to be loaded into the target rendering engine.

[0104] It should be noted that the model to be loaded is a 3D model that needs to be displayed in the target rendering engine after switching between engines, and includes complete data such as geometry, material properties, and level of detail (LOD).

[0105] This embodiment aims to spatially locate and load resources into the target rendering engine based on the target coordinates of the model to be loaded after coordinate correction, ultimately achieving accurate reconstruction and visualization of the 3D scene.

[0106] As a feasible implementation method, the loading process employs an intelligent scheduling strategy based on viewpoint adaptation. Specifically, the current viewpoint parameters are acquired in real time, and the level of detail and fragmentation data of the model are dynamically scheduled according to these parameters. When the viewpoint distance is greater than a preset threshold, a low-detail model is automatically loaded to optimize transmission and rendering performance; when the viewpoint distance is less than or equal to the preset threshold, a high-detail model is switched to ensure visual realism. Simultaneously, during the loading process, core region fragments within the user's field of view are prioritized to ensure the immediate presentation of critical content.

[0107] Optionally, loading efficiency can be further improved by constructing a multi-level caching mechanism. Memory caching stores frequently used model fragments and material resources, allowing them to be read directly from memory when the same content is requested again, effectively avoiding repeated disk I / O operations and network transmission overhead. Furthermore, a preloading strategy can be combined to predict the next scene area a user might enter based on their navigation behavior and preload the corresponding model data in the background, achieving a smooth scene transition that is imperceptible to the user.

[0108] Understandably, through the aforementioned intelligent scheduling and caching optimization mechanisms, step S40, while ensuring the accuracy of the model's spatial location, further addresses the performance bottleneck issue during the loading of large-scale 3D scenes. This implementation not only guarantees the integrity and accuracy of scene content after cross-engine switching but also achieves a smooth user interaction experience through efficient resource scheduling, thereby realizing efficient collaboration in cross-rendering engine fusion at both the accuracy and performance levels.

[0109] In this implementation, a hybrid coordinate correction mechanism combining linear transformation and nonlinear compensation achieves high-precision spatial alignment across rendering engines. Specifically, a macroscopic coordinate baseline is first established during the linear coordinate transformation stage, and then a nonlinear correction field is used to capture and compensate for local distortions, effectively solving the visual misalignment problem caused by coordinate system differences and engine characteristics. Simultaneously, by combining a viewpoint-adaptive intelligent loading strategy with a multi-level caching mechanism, resource scheduling efficiency is significantly improved while ensuring spatial positioning accuracy. This achieves a balance between accuracy and performance during cross-engine scene switching, providing reliable technical support for multi-engine collaborative operations.

[0110] Based on the above embodiments of this application, in the second embodiment of this application, the content that is the same as or similar to that in the first embodiment can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 2 Before step S201, the cross-rendering engine fusion method based on nonlinear coordinate correction further includes steps S2001-S2002:

[0111] Step S2001: Construct an engine abstraction adaptation layer through the standardized interfaces provided by the standardized interface layer for different types of rendering engines.

[0112] It should be noted that the standardized interface layer is a predefined set of unified interfaces used to adapt to the communication needs of different types of rendering engines. The engine abstraction and adaptation layer is built on this standardized interface layer. By encapsulating and abstracting the native interfaces of the geospatial engine and the target rendering engine, it maps the capabilities of heterogeneous engines into a consistent calling method, thereby achieving unified access and centralized management of multiple engines, significantly improving the system's scalability and maintenance efficiency.

[0113] In this embodiment, when constructing the engine abstraction adaptation layer, the standardized interface for geospatial engines explicitly supports WGS84 coordinate output, 3D Tiles loading, and OGC standard calls, making it compatible with mainstream geospatial engines such as Cesium and Three.js. The interface output parameters include coordinate values ​​and precision level information. For the standardized interface for high-performance rendering engines, it supports local orthogonal coordinate input, material attribute parsing, and pre-loading status feedback, making it compatible with mainstream high-performance rendering engines such as UE and Unity. The interface input parameters include corrected target coordinates and model LOD level information. The parameter formats and calling protocols for both types of engine interfaces are standardized. When adding any type of engine, only the adaptation module for that engine and the standardized interface needs to be developed. There is no need to redevelop the adaptation interface and linear transformation parameters, nor to modify the core logic of the engine abstraction adaptation layer and subsequent correction processes, achieving a one-time adaptation and multi-engine reuse extension effect.

[0114] Step S2002: By calling the corresponding standardized interface through the engine abstraction adaptation layer, the coordinate data of the feature anchor point in the geospatial engine and the target rendering engine are obtained, and the coordinate data includes the first coordinate and the second coordinate.

[0115] In this embodiment, a parallel call mechanism is used to acquire coordinate data. The engine abstraction and adaptation layer simultaneously sends coordinate requests to both the geospatial engine and the target rendering engine, and synchronously collects the first and second coordinates through a standardized interface, avoiding the accumulation of time caused by serial calls and significantly improving the efficiency of coordinate acquisition.

[0116] This embodiment solves the problems of low adaptation efficiency and protocol incompatibility in traditional multi-engine fusion by designing a standardized interface and engine abstraction adaptation layer, and ensures the efficiency of coordinate data acquisition through a parallel calling mechanism.

[0117] For example, to help understand the implementation process of the cross-rendering engine fusion method based on nonlinear coordinate correction obtained by combining this embodiment with the above embodiments, please refer to... Figure 3 , Figure 3 A schematic diagram of the overall architecture of a cross-rendering engine fusion method based on nonlinear coordinate correction is provided, specifically:

[0118] The overall architecture of this application adopts a layered design, including an engine abstraction and adaptation layer, a nonlinear correction core layer, and a server management center.

[0119] The engine abstraction and adaptation layer, located at the top level of the architecture, defines standardized interfaces for two types of engines through its included standardized interface layer. Specifically, the geospatial engine interface adapts to geospatial visualization engines such as Cesium and three.js, supporting WGS84 coordinate output and 3D Tiles model loading; the high-fidelity rendering engine interface adapts to high-fidelity rendering platforms such as UE and Unity, supporting local orthogonal coordinate input and material attribute parsing. The engine abstraction and adaptation layer achieves unified access and protocol conversion for multiple engines.

[0120] The nonlinear correction core layer, serving as the middle layer of the architecture, comprises a feature anchor point extraction module, a deviation calculation module, and a nonlinear correction field construction module. Together, they complete the entire coordinate correction calculation process from feature recognition to nonlinear correction field construction. Specifically, the feature anchor point extraction module automatically identifies structurally stable feature anchor points (such as building corners and road intersections) from the model to be loaded. Subsequently, the deviation calculation module calls the engine abstraction and adaptation layer to obtain the coordinate pairs of the same feature anchor point in the dual engines and calculates its nonlinear deviation vector from the linearly transformed target theoretical coordinates. Finally, the nonlinear correction field construction module uses these deviation data as samples and employs algorithms such as Gaussian process regression or RBF interpolation to generate a spatially continuous compensation field, achieving dynamic correction at any spatial location.

[0121] The server-side management center is located at the bottom layer of the architecture, comprising a data storage layer, a state synchronization layer, and a scheduling service layer. The data storage layer persistently stores feature anchor data, correction parameters, and historical deviation records. The state synchronization layer monitors and coordinates the running status of the two engines in real time, ensuring consistency with the correction results. The scheduling service layer links the correction process and cross-engine switching actions, dynamically determining the LOD level and piece loading priority based on the current viewpoint parameters to ensure efficient rendering while maintaining accuracy. Furthermore, the scheduling service layer can trigger a closed-loop verification mechanism, actively comparing the actual rendered coordinates with the target coordinates after model loading. When the deviation exceeds the limit, it automatically initiates the recalibration process, forming a complete closed loop of "collection—modeling—calibration—verification—feedback."

[0122] Based on the above embodiments of this application, in the third embodiment of this application, the same or similar content as the above embodiments can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 4 The cross-rendering engine fusion method based on nonlinear coordinate correction further includes steps S50~S80:

[0123] Step S50: In response to the cross-engine scene switching command, obtain the current view parameters of the geospatial engine.

[0124] It should be noted that the current viewpoint parameters refer to the spatial state data when the user observes the scene in the geospatial engine, including key information such as the viewpoint center position, viewpoint distance, and field of view angle (pitch angle).

[0125] Step S60: Determine the detail level and region segmentation range of the model to be loaded based on the current viewpoint parameters.

[0126] It should be noted that the level of detail is a model precision level dynamically selected based on the view distance, while the region segmentation range defines the model space region that needs to be loaded under the current view.

[0127] In this embodiment, the current viewpoint parameters of the geospatial engine are obtained in real time through the engine abstraction adaptation layer. Based on the comparison results of the current viewpoint parameters (such as viewpoint distance) and preset thresholds, the level of detail of the model to be loaded is automatically determined.

[0128] For example, when the viewing distance is greater than 500 meters, a simplified LOD2 model is loaded; when the viewing distance is between 100 and 500 meters, a standard LOD3 model (ultra-simplified model, with ≤50,000 faces) is loaded; and when the distance is less than 100 meters, a refined LOD4 model is loaded. Simultaneously, the visible space range is calculated based on the center position of the viewing angle and the field of view, and the area to be loaded is determined by combining this with the target coordinates.

[0129] Step S70: Based on the target coordinates, the level of detail, and the region segmentation range, the model to be loaded is divided into multiple data segments.

[0130] During the data partitioning stage, the model data is divided into multiple independent data blocks according to the determined regional partitioning range, and each data block is labeled with the corresponding level of detail and spatial location information.

[0131] Step S80: Load the data fragments in the target rendering engine according to the preset loading priority.

[0132] During the loading and execution phase, data chunks are loaded according to a preset priority strategy. Chunks located in the center of the view are given the highest priority to ensure that content in the user's focus area is loaded and presented first.

[0133] Specifically, step S80 includes steps S810 to S820:

[0134] Step S810: Based on the region sharding range, divide the multiple data shards into core region shards and non-core region shards.

[0135] It should be noted that core region slices refer to model data blocks located in the center of the user's current field of view that have the greatest impact on the visual experience, typically corresponding to the visible area within a certain radius of the center of the field of view; non-core region slices refer to data blocks located at the edge or secondary areas of the field of view. Differentiated allocation of loading priorities is a key mechanism for optimizing resource scheduling efficiency, improving the user experience by ensuring the timely presentation of critical content.

[0136] Step S820: Assign a higher loading priority to the core region fragment than to the non-core region fragment, so as to load the core region fragment first.

[0137] In this embodiment, based on the spatial relationship between the field of view cone and the model in the current viewpoint parameters, data fragments are divided into core regions and non-core regions. Specifically, using the viewpoint center as a reference, the area within a preset radius of the viewpoint center is defined as the core region, and data fragments within this core region are marked as core region fragments; while other areas within the field of view cone are marked as non-core region fragments. During the priority allocation phase, the core region fragments are assigned the highest loading priority to ensure that core region fragments enter the loading queue first and obtain computing and network resources first.

[0138] Optionally, the division of the core area can employ a dynamic adjustment strategy. Based on the speed and direction of viewpoint movement, changes in user attention can be predicted, and areas likely to enter the center of the field of view can be pre-defined as core areas. Simultaneously, a priority decay mechanism can be established to appropriately increase the priority of non-core area segments that fail to load for an extended period, preventing peripheral areas from remaining unloaded.

[0139] Optionally, step S820 is followed by steps S821 to S822:

[0140] Step S821: Receive the loading completion rate of the core region fragment from the target rendering engine.

[0141] Step S822: When the loading completion rate reaches a preset threshold, a smooth transition animation is started to complete the cross-engine switch from the geospatial engine to the target rendering engine; wherein, during the execution of the smooth transition animation, the non-core region fragments continue to be loaded in the background.

[0142] It should be noted that loading completion rate refers to the proportion of core region fragments that have been fully loaded in the target rendering engine out of the total number of core region fragments. This parameter is fed back in real time through the engine's callback mechanism. Smooth transition animations are visual continuity effects achieved through interpolation algorithms, used to eliminate visual jarring during scene transitions.

[0143] In this embodiment, the loading progress of core region fragments is continuously monitored by an event listening mechanism. When the loading completion rate reaches a preset threshold (e.g., 95%), a frame-synchronized smooth transition animation is triggered. This frame-synchronized smooth transition animation establishes a gradual trajectory of visual parameters such as viewpoint and lighting between the geospatial engine and the target rendering engine, completing a smooth visual transition within a short time window. During the execution of the transition animation, the system maintains a background thread to continue loading the remaining non-core region fragments, ensuring that the smoothness of animation playback is not affected through resource isolation. Simultaneously, a loading status synchronization mechanism is established to update the loading progress of non-core regions to the status synchronization layer in real time, providing a status basis for possible user interactions.

[0144] This embodiment further shortens the interactive waiting time by organically combining core area priority loading with visual transition animations, greatly improving the response speed and smoothness of cross-engine scene switching, and providing users with a seamless visual experience transition.

[0145] Optionally, when loading data chunks in the target rendering engine, parallel downloading of multiple data chunks can be supported. By establishing multiple download threads to synchronously acquire data from different chunks, the overall loading efficiency can be significantly improved.

[0146] Optionally, the process of loading data fragments can adopt a memory cache priority mechanism, which checks whether the required data fragments exist in the memory cache, and if the required data fragments exist, they are loaded directly.

[0147] This embodiment achieves intelligent resource scheduling based on the user's perspective through a viewpoint-adaptive loading mechanism. By dynamically adjusting the model's detail level and fragmentation range, it effectively reduces unnecessary data loading while ensuring visual quality. Furthermore, the priority-based partitioning loading mechanism effectively solves the resource contention problem during large-scale scene loading. By identifying and prioritizing the core content most critical to user experience, this embodiment ensures the rapid placement of key visual elements during scene transitions, significantly shortening the perceived waiting time for the user.

[0148] Based on the above embodiments of this application, in the fourth embodiment of this application, the same or similar content as the above embodiments can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 5 After step S40, the cross-rendering engine fusion method based on nonlinear coordinate correction further includes steps S41 to S43:

[0149] Step S41: Obtain the actual rendering coordinates of multiple feature anchor points in the target rendering engine of the model to be loaded.

[0150] Step S42: Compare the actual rendered coordinates with the target coordinates corresponding to the feature anchor points to obtain the positional deviation of each feature anchor point.

[0151] Step S43: When any of the positional deviations exceeds a preset deviation threshold, a correction recalculation process is triggered.

[0152] It should be noted that actual rendering coordinates refer to the spatial coordinates of the feature anchor points of the model after it has been loaded and rendered in the target rendering engine. Position deviation is the spatial offset obtained by comparing the actual rendering coordinates with the expected target coordinates, and is used to quantitatively evaluate the actual effect of coordinate correction.

[0153] In this embodiment, the actual rendering coordinates of feature anchor points in the loaded model are obtained through the query interface of the target rendering engine. Then, the actual coordinates of each feature anchor point are compared point-by-point with its corresponding expected target coordinates, and its linear offset in the X, Y, and Z directions is calculated to form a position deviation dataset. The system presets a deviation threshold; when the position deviation of any feature anchor point exceeds this threshold, it is determined that the current coordinate correction result is inaccurate in that area, and the correction recalculation process is automatically triggered.

[0154] The correction and recalculation process includes steps S431 to S433:

[0155] Step S431: Through the engine abstraction adaptation layer, re-acquire the first coordinates of the feature anchor point in the geospatial engine and the second coordinates in the target rendering engine;

[0156] Step S432: Based on the first coordinate and the second coordinate, recalculate the nonlinear deviation vector corresponding to the feature anchor point, and update the local compensation value of the nonlinear correction model;

[0157] Step S433: Based on the local compensation value, perform incremental coordinate correction on the region of deviation exceeding the limit in the model to be loaded.

[0158] It should be noted that the calibration recalculation process is a local optimization process targeting areas where coordinate deviations exceed limits. By re-collecting data and updating the nonlinear calibration model, the accuracy of a specific area is improved. The local compensation value refers to the nonlinear correction amount recalculated for areas with excessive deviations. Incremental coordinate correction is a coordinate adjustment operation performed only on the problem area without reloading the entire model.

[0159] In this embodiment, upon triggering the correction recalculation process, the coordinate query interfaces of the geospatial engine and the target rendering engine are first invoked in parallel through the engine abstraction adaptation layer to reacquire the latest coordinate data of the feature anchor points in both engines. Based on the reacquired coordinate data, an updated deviation dataset is generated according to the calculation formula of the nonlinear deviation vector (nonlinear deviation vector = local orthogonal coordinates - target theoretical coordinates obtained by linear transformation). Subsequently, an incremental learning algorithm is used to locally update the nonlinear correction model. Specifically, through the conditional update mechanism of Gaussian process regression, the new deviation data is integrated into the original model, and only the model parameters related to the deviation exceeding the limit are adjusted to generate local compensation values ​​for that region. Finally, based on the updated local compensation values, incremental coordinate correction is performed on the deviation exceeding the limit region of the model already loaded in the target rendering engine. For example, real-time correction is achieved by directly modifying the vertex coordinate data in the GPU buffer without reloading the entire model.

[0160] Optionally, when updating the nonlinear correction model, a sliding window mechanism can be used to retain recently collected sample data and remove outdated historical data to ensure that the model can adapt to dynamic changes in the scene.

[0161] Optionally, a model version management mechanism can be established to save the model parameters of each revision, so that if the correction effect is not ideal, it can be quickly rolled back to a historical stable version.

[0162] Optionally, a sampling polling mechanism can be used to monitor position deviation data and periodically perform deviation detection on key areas in the scene to avoid the performance overhead of full detection.

[0163] Optionally, a deviation trend prediction model can be established. By analyzing historical deviation data, potential areas of accuracy decay can be identified in advance, enabling preventative model maintenance.

[0164] As a possible implementation, step S40 is followed by steps S44-S45:

[0165] Step S44: Delete the uncorrected model data corresponding to the model to be loaded in the geospatial engine.

[0166] Step S45: Update the engine status identifier in the state synchronization layer and mark the currently active engine as the target rendering engine.

[0167] It should be noted that uncorrected model data refers to model instances in the geospatial engine that retain their original coordinates and have not undergone nonlinear correction. These become releasable redundant resources after cross-engine switching. The state synchronization layer is a middleware component that maintains the global state of the system. The engine state identifiers it stores are used to indicate the type of rendering engine currently in operation.

[0168] In this embodiment, once the model to be loaded is loaded into the target rendering engine and the calibration accuracy is confirmed to meet the requirements, the system deletes the uncalibrated model data that has completed the switching task through the geospatial engine interface. Simultaneously, the engine status identifier in the state synchronization layer is updated to the type identifier of the target rendering engine, ensuring that subsequent interaction commands and rendering tasks are correctly routed to the target rendering engine.

[0169] Optionally, a resource recycling mechanism can be established to temporarily back up deleted model data and set a lifespan, supporting rapid recovery and rollback within a specific time window.

[0170] This embodiment constructs a closed-loop control for cross-engine coordinate correction through accuracy verification and adaptive correction mechanisms. This embodiment not only promptly detects and corrects accuracy degradation caused by environmental changes, engine updates, and other factors, but also ensures that the coordinate correction effect maintains a high level of accuracy over the long term through continuous self-optimization capabilities, significantly improving the robustness and reliability of the entire cross-rendering engine fusion system.

[0171] This application provides a cross-rendering engine fusion device based on nonlinear coordinate correction. The cross-rendering engine fusion device based on nonlinear coordinate correction includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the cross-rendering engine fusion method based on nonlinear coordinate correction in the above embodiment 1.

[0172] The following is for reference. Figure 6The diagram illustrates a structural schematic of a cross-rendering engine fusion device suitable for implementing embodiments of this application based on nonlinear coordinate correction. The cross-rendering engine fusion device based on nonlinear coordinate correction in the embodiments of this application may include various hardware and software components for implementing the cross-rendering engine fusion method based on nonlinear coordinate correction. Figure 6 The cross-rendering engine fusion device based on nonlinear coordinate correction shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0173] like Figure 6 As shown, the cross-rendering engine fusion device based on nonlinear coordinate correction may include a processing unit 1001 (e.g., a central processing unit, graphics processing unit, etc.) that can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 1002 or a program loaded from storage device 1003 into random access memory (RAM) 1004. The RAM 1004 also stores various programs and data required for the operation of the cross-rendering engine fusion device based on nonlinear coordinate correction. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to the I / O interface 1006: input devices 1007 including, for example, touchscreens, touchpads, keyboards, etc.; output devices 1008 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 1003 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1009. Communication device 1009 allows the cross-rendering engine fusion device based on nonlinear coordinate correction to communicate wirelessly or wiredly with other devices to exchange data. While the figures show cross-rendering engine fusion devices based on nonlinear coordinate correction with various systems, it should be understood that implementation or possession of all the systems shown is not required. More or fewer systems may be implemented alternatively.

[0174] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from read-only memory 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0175] The cross-rendering engine fusion device based on nonlinear coordinate correction provided in this application, employing the cross-rendering engine fusion method based on nonlinear coordinate correction in the above embodiments, can solve the technical problem of model visual misalignment caused by coordinate system differences and nonlinear distortion between different rendering engines. Compared with the prior art, the beneficial effects of the cross-rendering engine fusion device based on nonlinear coordinate correction provided in this application are the same as those of the cross-rendering engine fusion method based on nonlinear coordinate correction provided in the above embodiments, and other technical features in this cross-rendering engine fusion device based on nonlinear coordinate correction are the same as those disclosed in the previous embodiment method, and will not be repeated here.

[0176] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0177] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0178] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, which are used to execute the cross-rendering engine fusion method based on nonlinear coordinate correction in the above embodiments.

[0179] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, radio frequency (RF), etc., or any suitable combination thereof.

[0180] The aforementioned computer-readable storage medium may be included in a cross-rendering engine fusion device based on nonlinear coordinate correction; or it may exist independently and not be assembled into a cross-rendering engine fusion device based on nonlinear coordinate correction.

[0181] The aforementioned computer-readable storage medium carries one or more programs that, when executed by a cross-rendering engine fusion device based on nonlinear coordinate correction, cause the cross-rendering engine fusion device based on nonlinear coordinate correction to: respond to a cross-engine scene switching command, perform linear coordinate transformation on the original coordinates of the model to be loaded in the geospatial engine to obtain the theoretical target coordinates of the model to be loaded in the coordinate system of the target rendering engine; determine a nonlinear compensation value corresponding to the theoretical target coordinates based on a pre-built nonlinear correction model; superimpose the theoretical target coordinates and the nonlinear compensation value to obtain the target coordinates of the model to be loaded in the target rendering engine; and load the model to be loaded in the target rendering engine based on the target coordinates.

[0182] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0183] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0184] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0185] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described cross-rendering engine fusion method based on nonlinear coordinate correction. This method can solve the technical problem of visual misalignment of models caused by differences in coordinate systems and nonlinear distortion between different rendering engines. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the cross-rendering engine fusion method based on nonlinear coordinate correction provided in the above embodiments, and will not be repeated here.

[0186] This application provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the cross-rendering engine fusion method based on nonlinear coordinate correction as described above.

[0187] The computer program product provided in this application can solve the technical problem of visual misalignment of models caused by differences in coordinate systems and nonlinear distortion between different rendering engines. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as the beneficial effects of the cross-rendering engine fusion method based on nonlinear coordinate correction provided in the above embodiments, and will not be repeated here.

[0188] The above are merely preferred embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent scope of this application.

[0189] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0190] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method.

[0191] The above are merely preferred embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. A cross-rendering engine fusion method based on nonlinear coordinate correction, characterized in that, The cross-rendering engine fusion method based on nonlinear coordinate correction includes: In response to the cross-engine scene switching command, the original coordinates of the model to be loaded in the geospatial engine are transformed into linear coordinates to obtain the theoretical target coordinates of the model to be loaded in the target rendering engine coordinate system. Based on a pre-constructed nonlinear correction model, the nonlinear compensation value corresponding to the theoretical target coordinates is determined. The nonlinear correction model is constructed using a Gaussian process regression algorithm. The theoretical target coordinates are superimposed with the nonlinear compensation value to obtain the target coordinates of the model to be loaded in the target rendering engine; Based on the target coordinates, load the model to be loaded into the target rendering engine; Before the step of determining the nonlinear compensation value corresponding to the theoretical target coordinates based on the pre-built nonlinear correction model, the cross-rendering engine fusion method based on nonlinear coordinate correction further includes: Obtain multiple feature anchor points in the model to be loaded, and determine the first coordinates of the feature anchor points in the geospatial engine and the second coordinates in the target rendering engine; Based on the linear coordinate transformation model, the first coordinate is converted into the corresponding theoretical target coordinate, and the nonlinear deviation vector corresponding to each feature anchor point is calculated according to the deviation between the theoretical target coordinate and the second coordinate. Based on the first coordinates corresponding to the feature anchor points and the nonlinear deviation vector, the nonlinear correction model is constructed. The nonlinear correction model is used to output the corresponding nonlinear compensation value according to the input spatial coordinates.

2. The cross-rendering engine fusion method based on nonlinear coordinate correction as described in claim 1, characterized in that, The step of obtaining multiple feature anchor points in the model to be loaded includes: Edge feature points are extracted from the model to be loaded using an edge detection algorithm; Calculate the number of neighboring geometric points within a preset neighborhood radius for each edge feature point, and determine candidate feature anchor points based on the number of neighboring geometric points; The candidate feature anchors are matched with the model structure feature library, and the feature anchors are finally determined based on the matching results.

3. The cross-rendering engine fusion method based on nonlinear coordinate correction as described in claim 1, characterized in that, The step of constructing the nonlinear correction model based on the first coordinates corresponding to the feature anchor points and the nonlinear deviation vector, wherein the nonlinear correction model outputs the corresponding nonlinear compensation value according to the input spatial coordinates, includes: Using the first coordinates corresponding to each of the feature anchor points as input vectors and the corresponding nonlinear deviation vectors as output targets, a deviation training sample set is constructed. The Gaussian process regression algorithm is used to spatially fit the bias training sample set to obtain the nonlinear correction model.

4. The cross-rendering engine fusion method based on nonlinear coordinate correction as described in claim 1, characterized in that, Before the step of acquiring multiple feature anchor points in the model to be loaded and determining the first coordinates of the feature anchor points in the geospatial engine and the second coordinates of the feature anchor points in the target rendering engine, the cross-rendering engine fusion method based on nonlinear coordinate correction further includes: An engine abstraction adaptation layer is built by providing standardized interfaces with different types of rendering engines through a standardized interface layer; By calling the corresponding standardized interface through the engine abstraction adaptation layer, the coordinate data of the feature anchor point in the geospatial engine and the target rendering engine are obtained, and the coordinate data includes the first coordinate and the second coordinate.

5. The cross-rendering engine fusion method based on nonlinear coordinate correction as described in claim 1, characterized in that, The cross-rendering engine fusion method based on nonlinear coordinate correction also includes: In response to the cross-engine scene switching command, the current view parameters of the geospatial engine are obtained; Based on the current viewpoint parameters, determine the level of detail and the range of region segments for the model to be loaded; Based on the target coordinates, the level of detail, and the region segmentation range, the model to be loaded is divided into multiple data segments; The data fragments are loaded into the target rendering engine according to the preset loading priority.

6. The cross-rendering engine fusion method based on nonlinear coordinate correction as described in claim 5, characterized in that, The step of loading the data fragments in the target rendering engine according to a preset loading priority includes: Based on the region sharding range, the multiple data shards are divided into core region shards and non-core region shards; The core region fragment is assigned a higher loading priority than the non-core region fragment, so that the core region fragment is loaded first.

7. The cross-rendering engine fusion method based on nonlinear coordinate correction as described in claim 6, characterized in that, Following the step of assigning a higher loading priority to the core region fragments than to the non-core region fragments, so as to prioritize loading the core region fragments, the cross-rendering engine fusion method based on nonlinear coordinate correction further includes: Receive the loading completion status of the core region fragments from the target rendering engine; When the loading completion rate reaches a preset threshold, a smooth transition animation is initiated to complete the cross-engine switch from the geospatial engine to the target rendering engine. During the execution of the smooth transition animation, the non-core region fragments continue to be loaded in the background.

8. The cross-rendering engine fusion method based on nonlinear coordinate correction as described in claim 1, characterized in that, Following the step of loading the model to be loaded into the target rendering engine based on the target coordinates, the cross-rendering engine fusion method based on nonlinear coordinate correction further includes: Obtain the actual rendering coordinates of multiple feature anchor points in the target rendering engine of the model to be loaded; The actual rendering coordinates are compared with the target coordinates corresponding to the feature anchor points to obtain the positional deviation of each feature anchor point; When any of the aforementioned positional deviations exceeds a preset deviation threshold, a correction recalculation process is triggered, the correction recalculation process including: By using the engine abstraction adaptation layer, the first coordinates of the feature anchor point in the geospatial engine and the second coordinates in the target rendering engine are re-acquired. Based on the first coordinate and the second coordinate, the nonlinear deviation vector corresponding to the feature anchor point is recalculated, and the local compensation value of the nonlinear correction model is updated. Based on the local compensation value, incremental coordinate correction is performed on the region of deviation exceeding the limit in the model to be loaded.

9. The cross-rendering engine fusion method based on nonlinear coordinate correction as described in claim 1, characterized in that, Following the step of loading the model to be loaded into the target rendering engine based on the target coordinates, the cross-rendering engine fusion method based on nonlinear coordinate correction further includes: Delete the uncorrected model data corresponding to the model to be loaded from the geospatial engine; Update the engine status identifier in the state synchronization layer and mark the currently active engine as the target rendering engine.

Images