Coordinate deflection method and device based on scene geometric features, medium and product

Abstract

Embodiments of the present application provide a coordinate deflection method and device based on scene geometric features, a medium and a product. The method comprises: performing topological analysis on scene data of a target scene according to a scene type of the target scene, determining a deflection constraint region and a topological feature parameter of the target scene; generating a deflection point in the deflection constraint region according to the topological feature parameter and a deflection strategy; the deflection strategy is determined according to the scene type of the target scene; and outputting and displaying the deflection point in the target scene. The method is used to improve the effectiveness of the generated offset point.

Description

Technical Field

[0001] This application relates to the field of image processing technology, and in particular to a coordinate deflection method, device, medium and product based on scene geometric features. Background Technology

[0002] In cutting-edge technologies such as intelligent transportation, high-precision map construction, and autonomous driving, coordinate deflection methods are widely used in key aspects such as path planning, lane-level localization, and virtual sensor modeling. This technology generates a derived set of points for decision support or environmental perception by applying a specific direction and magnitude of offset to the original coordinate points, making it an important means of achieving refined spatial modeling and dynamic scene understanding. With the increasing complexity of urban road structures (such as multi-level interchanges and dense ramps), higher demands are being placed on the accuracy and boundary adaptability of coordinate deflection.

[0003] Currently, common coordinate deflection methods typically generate points based on a fixed direction (such as perpendicular to the road centerline) or a preset offset distance. The execution process generally includes: first, extracting the reference trajectory or road geometric features; then, calculating the normal or tangent vector in the local coordinate system; and finally, translating the original point along the specified direction to obtain the offset point.

[0004] However, existing methods often result in offset points frequently exceeding scene boundaries, leading to a high proportion of invalid points and failing to meet user needs. Summary of the Invention

[0005] The coordinate deflection method, device, medium, and product based on scene geometric features provided in this application are used to improve the effectiveness of the generated offset points.

[0006] In a first aspect, embodiments of this application provide a coordinate deflection method based on scene geometric features, including:

[0007] Based on the scene type of the target scene, perform topological analysis on the scene data of the target scene to determine the deflection constraint region and topological feature parameters of the target scene;

[0008] Based on the topological feature parameters and the deflection strategy, deflection points are generated within the deflection constraint region; the deflection strategy is determined according to the scene type of the target scene.

[0009] Output and display the deflection points in the target scene.

[0010] In one possible implementation, when the target scene is a linear scene, deflection points are generated within the deflection constraint region based on topological feature parameters and a deflection strategy, including:

[0011] The deflection direction and road curvature are determined based on the topological characteristic parameters and the deflection strategy.

[0012] Based on the deflection direction, generate the first initial deflection point within the target scene;

[0013] Adjust the point density of the first initial deflection point within the target scene based on the road curvature;

[0014] Based on the deflection constraint region, the first initial deflection point within the target scene is filtered to determine the deflection point within the deflection constraint region.

[0015] In one possible implementation, when the target scene is a regional scene, deflection points are generated within the deflection constraint region based on topological feature parameters and a deflection strategy, including:

[0016] Based on the topological characteristic parameters and the deflection strategy, the planar partitioning region in the deflection constraint region is determined;

[0017] A second initial deflection point is generated within the planar partitioned region; wherein the density and / or number of the second initial deflection points differ depending on the position of the planar partitioned region within the deflection constraint region.

[0018] The second initial deflection points between different planar regions are deduplicated to determine the deflection points within the deflection constraint region.

[0019] In one possible implementation, when the target scene is a stereoscopic scene, deflection points are generated within the deflection constraint region based on topological feature parameters and a deflection strategy, including:

[0020] Based on topological feature parameters and deflection strategy, the layered regions in the deflection constraint region are determined; different layered regions have different heights in the stereo scene.

[0021] A third initial deflection point is generated within the layered region; the larger the area of ​​the layered region, the more third initial deflection points are generated within the layered region.

[0022] Based on the spatial coordinates of the third initial deflection point, the third initial deflection point is filtered to determine the deflection points within the deflection constraint region.

[0023] In one possible implementation, when the target scene is a point-like scene, deflection points are generated within the deflection constraint region based on topological feature parameters and a deflection strategy, including:

[0024] Based on the topological feature parameters and the deflection strategy, a fourth initial deflection point is generated on the annular surface with different radius intervals, with the center point of the deflection constraint region as the annular center.

[0025] When the number of fourth initial deflection points on the annular surface is greater than the preset number, the distribution position of the fourth initial deflection points on the annular surface is adjusted to determine the deflection points within the deflection constraint area.

[0026] In one possible implementation, before performing topological analysis on the scene data of the target scene according to the scene type of the target scene to determine the deflection constraint region and topological feature parameters of the target scene, the method includes:

[0027] Acquire geospatial data of the target scene;

[0028] Geospatial data is cleaned and processed to obtain scene data;

[0029] The scene type of the target scene is determined based on the topological relationship between the coordinate data in the scene data.

[0030] In one possible implementation, the deflection point in the target scene is output and displayed, including:

[0031] According to the preset verification rules, the deflection points in the target scene are verified to determine the number of valid points in the target scene; the verification rules are determined according to the scene type of the target scene, and the number of valid points is the number of deflection points that are not in the boundary part of the deflection constraint region.

[0032] If the number of valid points in the target scene meets the preset requirements, then the deflection points in the target scene will be output and displayed.

[0033] If the number of valid points in the target scene does not meet the preset requirements, the deflection points will be supplemented until the number of valid points in the target scene meets the preset requirements. Then, the deflection points in the target scene will be output and displayed.

[0034] In one possible implementation, when there are at least two target scenarios, the method further includes:

[0035] Based on GPU multi-threaded parallel tasks, the deflection points of the target scene are sent to the corresponding threads for verification according to different scene types to determine the number of valid points in different target scenes.

[0036] Based on the number of valid points in different target scenes, output and display the deflection points in the target scene.

[0037] In one possible implementation, after outputting and displaying the deflection points in the target scene, the method further includes:

[0038] Determine the sensitive deflection points and the number of sensitive deflection points among the deflection points;

[0039] Based on the sensitive deflection points and the number of sensitive deflection points, determine the proportion and / or distribution of sensitive deflection points among the deflection points;

[0040] Adjust the deflection strategy based on the proportion and / or distribution of sensitive deflection points among the deflection points.

[0041] Secondly, embodiments of this application provide a coordinate deflection device based on scene geometric features, comprising:

[0042] The topology parsing module is used to perform topology parsing on the scene data of the target scene based on the scene type of the target scene, and to determine the deflection constraint region and topology feature parameters of the target scene.

[0043] The generation module is used to generate deflection points within the deflection constraint region based on topological feature parameters and deflection strategy; the deflection strategy is determined according to the scene type of the target scene.

[0044] The display module is used to output and display the deflection points in the target scene.

[0045] Thirdly, embodiments of this application provide an electronic device, including: a memory and a processor;

[0046] The memory stores the instructions that the computer executes;

[0047] The processor executes computer execution instructions stored in memory, causing the processor to perform the first aspect and / or various possible implementations of the first aspect as described above.

[0048] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the first aspect and / or various possible implementations of the first aspect.

[0049] Fifthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the first aspect and / or various possible implementations of the first aspect.

[0050] The coordinate deflection method, device, medium, and product based on scene geometric features provided in this application embodiment perform topological analysis on the scene data of the target scene according to the scene type, thereby transforming the abstract scene structure into a computable deflection constraint region. Then, combined with the topological feature parameters obtained from the analysis, a deflection strategy adapted to the scene type is matched and applied, thereby ensuring that the generated deflection points fall within the effective range of the scene. Thus, by constructing differentiated constraint regions and deflection rules for different scene types, the generated offset points closely conform to the geometric and topological features of the actual scene, thereby effectively avoiding boundary crossing phenomena, reducing the proportion of invalid points, and improving the effectiveness of the generated offset points. Attached Figure Description

[0051] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0052] Figure 1 A scene diagram illustrating the coordinate deflection method based on scene geometric features provided in this application;

[0053] Figure 2 A flowchart illustrating the coordinate deflection method based on scene geometric features provided in this application;

[0054] Figure 3a A schematic diagram illustrating the process of generating deflection points in a linear scene as provided in this application;

[0055] Figure 3b A schematic diagram illustrating the process of generating deflection points in a regional scene as provided in this application;

[0056] Figure 3c A flowchart illustrating the process of generating deflection points in a stereoscopic scene, as provided in this application;

[0057] Figure 3d A schematic diagram illustrating the process of generating deflection points in a point-like scene as provided in this application;

[0058] Figure 4 A flowchart illustrating the process of determining the scenario type of the target scenario provided in this application;

[0059] Figure 5 This is a schematic diagram illustrating the process of completing the deflection point according to an embodiment of this application.

[0060] Figure 6a A schematic diagram of the sensitivity testing process provided in an embodiment of this application;

[0061] Figure 6b A schematic diagram of the sensitivity determination process provided in this application embodiment;

[0062] Figure 7 A schematic diagram of the coordinate deflection device based on scene geometric features provided in this application;

[0063] Figure 8 A schematic diagram of the structure of the electronic device provided in this application.

[0064] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0065] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0066] First, let me explain the terms used in this application:

[0067] Coordinate offset refers to the process in geospatial data processing of generating new offset points based on original coordinate points according to specific directions, distances, or rules. This method is commonly used in scenarios such as high-precision map construction, autonomous driving simulation, and path planning. By offsetting laterally, longitudinally, or circularly based on geometric structures such as road centerlines, boundaries, or polygonal regions, it generates a derived set of points for testing, modeling, or perception. Its core objective is to ensure that the offset points fall within the effective area while maintaining consistency with the geometric features of the original scene, avoiding out-of-bounds or invalid outputs.

[0068] In the field of geographic information data processing and application, coordinate deflection technology is crucial for adapting to different scenarios and meeting diverse business needs. With the booming development of geographic information-related industries, the demand for accurate utilization of coordinate data in various scenarios is increasing, such as the precise processing of road scene coordinates in intelligent transportation and the reasonable optimization of regional scene coordinates in urban planning. However, a key issue must be addressed in the coordinate deflection process: the significant differences in geometric features between different scenarios place extremely stringent requirements on the range, direction, and effectiveness of coordinate deflection.

[0069] If the coordinate deflection method fails to fully consider the geometric characteristics of these scenarios, it is easy for a large number of invalid points to emerge. For example, in linear scenarios, the offset points frequently exceed the strip area, and in three-dimensional scenarios, the offset points fall into non-travelable areas across layers, which seriously affects data quality and business execution. At the same time, the traditional inefficient boundary verification method will greatly reduce processing efficiency, which will seriously hinder the actual business operation and efficient use of data.

[0070] Currently, how to achieve accurate and efficient coordinate deflection based on the geometric characteristics of different scenarios, ensure the effectiveness of the deflection point set, and significantly improve processing efficiency has become a thorny problem that urgently needs to be solved in the industry. At present, there is a lack of mature and comprehensive processing technologies and methods to effectively address this issue.

[0071] In most past and present related practices, the most common approach closest to this application is the traditional coordinate offset strategy, which employs a "random offset + fixed range" model. This strategy typically does not differentiate between specific scenarios; instead, it directly performs random offset operations on the coordinates based on a pre-defined fixed range. After the offset is completed, conventional methods such as point-by-point verification are used to determine whether the offset points meet the requirements. For example, when dealing with different scenarios such as roads, overpasses, and parking lots, a uniform random offset rule and fixed range setting are used, completely disregarding the unique geometric characteristics of the scenario itself, such as the direction and width limitations of linear roads, the multi-layered structure of overpasses, and the polygonal boundaries of parking lots.

[0072] The coordinate deflection method based on scene geometric features provided in this application can extract the spatial topology of different scenes, construct targeted deflection constraint regions (such as the ±25-meter strip area of ​​road buffer50, the layered three-dimensional area of ​​overpass), and clarify the offset boundary from the source, thus solving the problem of mismatch between traditional fixed rules and scene characteristics.

[0073] Meanwhile, a differentiated deflection strategy based on scene topology design ensures that the offset points conform to the actual shape of the scene, avoiding invalid points caused by traditional random offsets and improving the effectiveness of the point set. Verification efficiency is significantly improved through "pre-filtering (rapid removal of invalid points using circumscribed rectangles) + GPU parallel precise verification (ray casting / layered judgment)"; at the same time, invalid points are automatically supplemented to ensure that the number of test points meets the requirements, solving the problem that traditional point-by-point verification is too time-consuming and cannot support high-frequency testing.

[0074] Therefore, through the whole process design of "analyzing topology → dynamic offset → efficient verification", the precise matching between offset points and scene features is achieved, which not only improves the realism of the test, but also ensures the efficiency of the test, and effectively solves the core technical defects of traditional methods.

[0075] Figure 1 A scene illustration for the coordinate deflection method based on scene geometric features provided in this application, as shown below. Figure 1As shown, this application can be applied to sensitivity detection testing in cloud-based geofencing whitelist scenarios. Typical scenarios may include gas stations, CNG stations, charging stations, service areas, parking lots, overpasses, tunnels, etc. In intelligent transportation systems, geofencing is used to define business rules for specific areas (such as vehicle access restrictions and charging pile usage permissions), and whitelist scenarios require testing to verify the accuracy of these rules. The execution entity of this application can be a coordinate deflection system based on scene geometric features. This system may include a cloud-based data processing module (responsible for scene parsing, deflection point generation and verification), a GPU-accelerated computing module (for parallel verification), and a sensitivity testing interface module (for batch verification of whether deflection points trigger sensitive rules). Thus, large-scale test data can be rapidly generated and verified through cloud processing to adapt to the complex geometric features of multiple scenarios.

[0076] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0077] Figure 2 A flowchart illustrating the coordinate deflection method based on scene geometric features provided in this application is shown below. Figure 2 As shown, the method includes:

[0078] S201. Based on the scene type of the target scene, perform topological analysis on the scene data of the target scene to determine the deflection constraint region and topological feature parameters of the target scene.

[0079] The target scenario can refer to a geofence whitelist scenario that requires coordinate transformation (such as a 50-meter road buffer, parking lot, overpass, etc.). A geofence whitelist scenario can refer to a predefined set of trusted geographical areas that are authorized to perform specific location processing (such as coordinate transformation, data anonymization, or trajectory publishing). In contrast to a blacklist (prohibited areas), a whitelist scenario indicates that within these areas, scenario data can be securely transformed or shared, provided that preset rules are met, without causing privacy leaks or compliance risks.

[0080] For example, urban main roads and their buffer zones (such as within 50 meters on both sides of the road), public parking lots, highway entrances and exits, overpasses, logistics parks, and other areas with clearly defined boundaries and functional attributes.

[0081] The scene type of a target scene can refer to the classification and identification of the target scene based on geographic semantics, functional purpose, or spatial structure. For example, scene types can include linear scenes, regional scenes, three-dimensional scenes, and point scenes. Among them, linear scenes can refer to one-dimensional spatial structures dominated by paths or trajectories, such as roads and railways; regional scenes can refer to two-dimensional planar spaces with closed boundaries and internal areas, such as parking lots and squares; three-dimensional scenes can refer to three-dimensional spatial structures containing elevation or vertical layering information, such as overpasses, tunnel layers, or building blocks; and point scenes can refer to zero-dimensional geographic elements with discrete location identifiers as the core, such as bus stops, charging piles, or points of interest (POIs).

[0082] Scene data for the target scene can refer to structured or semi-structured data used to describe the spatial geometry and semantic information within the geofence whitelist area. Specific forms include, but are not limited to: boundary coordinate strings composed of latitude and longitude (such as polygon vertex sequences or centerline trajectories), standard GIS files (such as Shapefile, GeoJSON, KML, etc.), road network topology maps, high-precision map layers, and metadata containing attribute information (such as area type, travel direction, and hierarchical relationship).

[0083] Topological parsing of scene data of a target scene refers to the process of extracting its inherent topological attributes, such as spatial connectivity, adjacency, hierarchical structure, and path reachability, from the original scene data.

[0084] The deflection constraint region of the target scene can refer to a spatial sub-region within the target scene that allows coordinate deflection, in order to ensure that the coordinates after deflection still maintain geographical rationality and functional consistency.

[0085] For example, in a linear scene, the deflection constraint region can be a buffer zone extending along the centerline normal; in a regional scene, the deflection constraint region can be the area within the inner boundary of the closed planar region; in a three-dimensional scene, the deflection constraint region can be each layered region; and in a point-like scene (such as POI, device site), the deflection constraint region can be a ring-shaped region centered on the center point or a region containing other shapes of each point in the point-like scene.

[0086] The topological feature parameters of the target scene can refer to the key indicators extracted from the topological analysis results and used to guide the coordinate deflection strategy, including but not limited to: node degree (intersection complexity), edge connectivity, region concavity and convexity, minimum enclosing rectangle direction, trunk direction vector, radius of curvature, layer depth (such as the number of interchange layers), number of connected components, etc.

[0087] In this embodiment of the application, the topological feature parameters of the target scene can be determined based on the scene type of the target scene. For example:

[0088] When the scene type is a linear scene, the topological feature parameters include, but are not limited to: the endpoints, midpoint coordinates, length, slope and curvature of the line segments, and the connection relationship of the line segments.

[0089] When the scene type is a region scene, the topological feature parameters include, but are not limited to: the vertex coordinates, number of edges, area, center point, minimum bounding rectangle of the region, and parameters characterizing the concavity / convexity and the existence of inner loops of the region;

[0090] When the scene type is a three-dimensional scene, the topological feature parameters include, but are not limited to: the volume, surface area, height range of the three-dimensional object, and the three-dimensional coordinate data of the interfaces of each two-dimensional interface when the three-dimensional object is decomposed into multiple two-dimensional cross-sections.

[0091] When the scene type is a point scene, the topological feature parameters include, but are not limited to: the coordinates of each point, the distance between points, and the attributes of the points.

[0092] S202. Generate deflection points within the deflection constraint region based on topological feature parameters and deflection strategy; the deflection strategy is determined according to the scene type of the target scene.

[0093] The deflection strategy refers to a set of differentiated coordinate perturbation rules formulated according to the type of the target scene. It aims to ensure location privacy or meet data anonymization compliance requirements while maximizing the usability and rationality of geographic semantics. In some embodiments, the strategy may include rules for setting the offset direction, offset range, and point density allocation. By deeply coupling these parameters with scene topology features and functional attributes, deflection points can be generated within the deflection constraint region.

[0094] For example, in linear scenes, deflection points conforming to the road direction are generated within a ±10-meter buffer along the normal direction based on the centerline direction vector and curvature parameters; in regional scenes, deflection points are generated at a uniform density distribution inside the boundary based on the concavity and convexity of the polygon and the inner ring information to avoid falling into the hole area; in three-dimensional scenes, deflection points are generated only within the current Z-layer projection plane by combining the elevation layer identifier and the intra-layer connectivity to prevent cross-layer errors; in point-like scenes, each original point generates a circular constraint area with a fixed radius; for dense point sets, adjacent circular areas are merged to avoid overlap; thus, deflection points are randomly generated within the constraint area.

[0095] S203. Output and display the deflection points in the target scene.

[0096] This involves presenting the deflection points in the target scene on a map or spatial visualization interface. For example, the deflected coordinates can be overlaid on the base map as a vector layer and visualized using point markers (such as circles or icons), color coding (to distinguish scene types or deflection confidence), transparency adjustment (to reflect point density), or dynamic trajectory lines. For linear or regional scenes, the original geometric contours (such as road centerlines or parking lot boundaries) can also be rendered synchronously to intuitively show the spatial relationship between the deflection points and the constraint areas, thereby ensuring that users or the system can clearly identify the rationality and compliance of the deflection results.

[0097] The coordinate deflection method based on scene geometric features provided in this application can extract the core topological parameters of the target scene to transform the abstract scene structure into a computable deflection constraint region. Simultaneously, differentiated deflection rules can be designed according to the scene type to ensure that the offset direction matches the scene characteristics. This allows deflection points to be generated within the deflection constraint region according to the deflection strategy, thereby ensuring that the offset points strictly fall within the effective range of the scene. Therefore, the method in this application improves the effectiveness of deflection points, avoids the invalid point problem caused by fixed rules in traditional methods, and ensures that the distribution of deflection points conforms to actual business needs, reducing resource waste.

[0098] Optionally, Figure 3a A flowchart illustrating the process of generating deflection points in a linear scene provided in this application is shown below. Figure 3a As shown, when the target scene is a linear scene, deflection points are generated within the deflection constraint region based on topological feature parameters and the deflection strategy, including:

[0099] The deflection direction and road curvature are determined based on the topological characteristic parameters and the deflection strategy.

[0100] Based on the deflection direction, generate the first initial deflection point within the target scene;

[0101] Adjust the point density of the first initial deflection point within the target scene based on the road curvature;

[0102] Based on the deflection constraint region, the first initial deflection point within the target scene is filtered to determine the deflection point within the deflection constraint region.

[0103] The deflection direction refers to the direction of offset within the target scene, based on the linear scene orientation, along the normal to the centerline of the linear scene (i.e., perpendicular to the direction of travel). For example, the deflection direction can be set to ±30° along the road orientation, indicating that the offset direction is not absolutely vertical, but allows for dynamic adjustment within a certain angle range to adapt to complex road structures (such as curves and ramp connections). This avoids the point from moving directly out of the road boundary or into an obstacle area due to vertical offset.

[0104] After determining the deflection direction, the first initial deflection point can be generated along the offset direction that represents the actual direction of the linear scene.

[0105] Road curvature can be a quantitative indicator of the degree of local bending of the centerline of a linear scene, used to dynamically adjust the distribution density of deflection points. For example, point density can be positively correlated with the curvature of the linear scene; that is, increasing point density in high-curvature areas (such as sharp curves and ramps) to capture geometric changes more precisely, and decreasing density in low-curvature areas (such as straight roads) to improve computational efficiency. Thus, the initial deflection point can be adjusted.

[0106] After obtaining the first initial deflection point after point density adjustment, this initial deflection point is compared with a deflection constraint region constructed through topological analysis, which reflects the effective spatial range of the scene. Only points located within this constraint region are retained, while deflection points that cross the boundary or fall into invalid areas (such as outside roads, inside buildings, or other elevation layers) are discarded. This ensures that the final output deflection points conform to the actual road geometry and topology, effectively avoiding invalid data caused by blind offsetting, thereby improving the spatial rationality of the offset points and the reliability of downstream tasks.

[0107] In this embodiment, the core of the linear scenario is directional constraint and curvature adaptation, deflecting along the road direction to avoid going out of bounds, and the high-risk judgment logic is covered by the densified point set of curves.

[0108] In linear scenarios, the deflection direction needs to be constrained along the road direction to avoid vertical offset exceeding the strip area (e.g., buffer 50 meters). This is achieved by generating the coordinates of the first initial deflection point through "direction angle controlling the offset direction" and "road width limiting the offset range." The coordinates of this first initial deflection point satisfy the following conditions:

[0109]

[0110]

[0111] The above formula is used to calculate the new coordinates of the deflection point in a linear scene. The core is in the original coordinates ( Based on this, combined with the road direction angle Deflection angle of deflection direction and deflection distance New coordinate points are generated along a reasonable direction (constrained by the road direction) to avoid deflection beyond the strip area (such as the range representing a buffer of 50 meters).

[0112] in,( ) represents the coordinates of the original center point or endpoint of the line segment in the linear scene. These coordinates are derived from the "centerline coordinate sequence" obtained through topological analysis and serve as the basic starting position for deflection calculation.

[0113] The direction angle is derived from topological analysis. For example, 30° needs to be converted to radians during calculation, as it determines the basic direction of the road's extension.

[0114] The deviation angle is the direction of deflection, and its value ranges from ±30° (i.e., (Converted to radians) is used to control the degree of deviation between the deflection direction and the road direction, avoiding deviation perpendicular to the road direction, thereby preventing deviation beyond the strip area.

[0115] This is the deflection distance, and its value does not exceed 1 / 2 of the road width. For example, when the buffer is 50 meters, ≤25 meters, this parameter comes from the "road width parameter" of topology analysis, which limits the maximum range of deflection.

[0116] ( () is the first initial deflection point generated at the end.

[0117] The deflection point density of linear scene points is positively correlated with road curvature (denser density on curves, sparser density on straight sections), meaning higher density on curves (higher curvature) and lower density on straight sections (lower curvature). This ensures test coverage of high-risk curve areas. The deflection point density of linear scene points satisfies the following:

[0118]

[0119] in, This represents the deflection point density of the current road segment (unit: points / meter, i.e., the number of deflection points generated per meter).

[0120] This is the base density for straight road sections, which is the default value (e.g., it can be set to 0.1 units / meter), and this value can be configured according to actual needs;

[0121] The curvature of the current road segment is derived from the "curvature distribution parameter" obtained through topological analysis. The curvature of the straight section... The curvature of sharp bends The greater the curvature, the more curved the road.

[0122] This is the curvature influence coefficient, with a default value of 2, but it can also be configured. It ensures that the effect of curvature on density increases more significantly with increasing curvature, thus amplifying the influence of curvature on density.

[0123] Therefore, through basic density Combining curvature curvature influence coefficient To dynamically adjust the final density .

[0124] Optionally, Figure 3b This application provides a flowchart illustrating the process of generating deflection points in a regional scene, as shown below. Figure 3b As shown, when the target scene type is a regional scene, deflection points are generated within the deflection constraint region based on topological feature parameters and deflection strategy, including:

[0125] Based on the topological characteristic parameters and the deflection strategy, the planar partitioning region in the deflection constraint region is determined;

[0126] A second initial deflection point is generated within the planar partitioned region; wherein the density and / or number of the second initial deflection points differ depending on the position of the planar partitioned region within the deflection constraint region.

[0127] The second initial deflection points between different planar regions are deduplicated to determine the deflection points within the deflection constraint region.

[0128] The planar partitioned region refers to the region after dividing the scene into sub-regions, which can be further divided into several sub-regions based on orientation, functional semantics, or geometric features. In this embodiment, the planar partitioned region can be divided into an internal low-risk region and an edge high-risk region based on risk control.

[0129] After determining the planar division regions, the density or number of deflection points can be dynamically adjusted based on the specific location of each planar division region within the deflection constraint region and its corresponding scene semantic or geometric importance. For example, the point density can be increased in structurally complex or safety-sensitive areas (such as the outside of curves or merging zones), while the points can be distributed more sparsely in structurally simple areas. Subsequently, since adjacent planar division regions may overlap or have adjacent boundaries, their respective second initial deflection points may have spatial repetitions. Therefore, deduplication processing is required (such as merging similar points based on coordinate distance thresholds). Thus, a set of effective deflection points that are non-redundant, reasonably distributed, and completely located within the deflection constraint region can be output, balancing accuracy, efficiency, and scene adaptability.

[0130] For example, the deflection constraint region can be divided into an "internal region" (>2 meters from the boundary) and an "edge region" (<2 meters from the boundary). Differentiated point density allocations are then applied based on region type: a 60% uniform distribution in the internal region and a 40% dense distribution in the edge region to achieve high-precision coverage. Subsequently, offset operations are performed in each region. The internal region undergoes uniform mesh offset, while the edge region dynamically adjusts its offset range (e.g., ±3 meters) and incorporates scene-based judgments to avoid boundary violations, thus generating a second initial deflection point. Since the internal and edge regions overlap near the boundary, potentially leading to duplicate points, the point sets of the two regions need to be merged, and redundant or duplicate points caused by dense distribution need to be removed. This outputs a valid, non-overlapping deflection point within the target scene that conforms to geometric constraints.

[0131] In this embodiment of the application, the number of deflection points generated by dividing the region into different planes in the regional scene can satisfy the following:

[0132]

[0133]

[0134] in, This can refer to calculating the number of points needed to generate in the internal area. By multiplying the total number of points Ntotal by 0.6 and then rounding to the nearest integer, the result is guaranteed to be an integer number of points.

[0135] This can refer to calculating the number of points that need to be generated in the edge region, subtracting the number of points in the inner region from the total number of points, to ensure that the total number of points meets the input value required for testing.

[0136] The total number of deflection points to be generated for the area scene is a value input from the test requirements, which determines the total number of deflection points for the entire area scene;

[0137] The number of points to be generated in the internal area is rounded to ensure that the number of points is an integer, and is used to generate deflection points in the low-risk internal area;

[0138] The number of points to be generated in the edge area is obtained by subtracting the number of points in the inner area from the total number of points to ensure that the total number of points meets the target. This is used to generate deflection points in high-risk edge areas.

[0139] Therefore, in regional scenarios, the total number of deflection points can be allocated according to the ratio of "60% for the internal area and 40% for the edge area" to focus on covering high-risk edge areas.

[0140] Optionally, Figure 3c This application provides a flowchart illustrating the process of generating deflection points in a 3D scene, as shown below. Figure 3c As shown, when the target scene is a stereoscopic scene, deflection points are generated within the deflection constraint region based on topological feature parameters and deflection strategy, including:

[0141] Based on topological feature parameters and deflection strategy, the layered regions in the deflection constraint region are determined; different layered regions have different heights in the stereo scene.

[0142] A third initial deflection point is generated within the layered region; the larger the area of ​​the layered region, the more third initial deflection points are generated within the layered region.

[0143] Based on the spatial coordinates of the third initial deflection point, the third initial deflection point is filtered to determine the deflection points within the deflection constraint region.

[0144] In this context, a layered region refers to multiple horizontally layered sub-regions formed by dividing the deflection constraint area in a 3D scene according to height (Z-axis) information. For example, based on the vertical hierarchy of road structures (such as ground level, upper level of overpasses, underground tunnels, etc.), each layer can be divided at fixed height differences (e.g., 2 meters) to form several "layers" with independent spatial ranges (e.g., layers 1 to 8). Each layered region corresponds to a specific height range to support layered deflection processing in 3D space, ensuring that offset points are generated independently in different height levels and avoiding cross-layer confusion, thereby achieving accurate modeling and effective coverage of complex 3D traffic structures (such as multi-level overpasses and overpasses).

[0145] The third initial deflection point can refer to the preliminary set of offset points generated within each layered region based on the planar division result of that layer and the corresponding deflection strategy. For example, each layer performs a "layer-by-layer scene offset" operation, which generates points on the two-dimensional plane of the current layer according to a preset deflection direction and rules, and assigns them three-dimensional coordinates by combining the Z-axis height information of that layer. These points already have a clear height attribute when generated, and the number of their distribution points is related to the area of ​​the layered region, that is, the larger the area, the higher the number of corresponding points, to ensure sufficient sampling of a large area; while the number of points is appropriately reduced for small areas or edge layers, balancing efficiency and accuracy, forming a differentiated point layout pattern that adapts to the structure of the three-dimensional scene.

[0146] Based on the spatial coordinates of the third initial deflection point, filtering of the third initial deflection point can be performed by first determining whether the point falls within the effective Z-axis range of the corresponding layer region (e.g., within ±1~2 meters), preventing layer skipping due to Z-axis offset exceeding the actual layer height; secondly, by filtering invalid cross-layer points, identifying and removing points that have fallen into other layers due to misjudgment or calculation errors (e.g., a point from an upper layer mistakenly falling into a lower layer); and finally, by combining the judgment of ineffective missing layers, excluding points located in inactive or blank layers without actual scene coverage. Thus, valid deflection points located within each layer region that conform to geometric and topological constraints can be retained, outputting a set of three-dimensional valid deflection points that can be used for high-precision map construction or autonomous driving perception tasks.

[0147] In this embodiment, the core of the stereoscopic scene is Z-axis layering + two-dimensional adaptation, that is, first splitting the two-dimensional plane according to altitude, and then reusing the regional scene strategy to ensure that the point set falls on the effective layer height.

[0148] The three-dimensional coordinates of the deflection point in the 3D scene satisfy:

[0149]

[0150]

[0151]

[0152] in, Characterizing the three-dimensional deflection point The coordinates are directly taken from the two-dimensional deflection points of a certain layer. Coordinates, reusing the two-dimensional deflection results of the region scene;

[0153] Characterizing the three-dimensional deflection point The coordinates are directly taken from the two-dimensional deflection points of a certain layer. The coordinates also reuse the two-dimensional deflection results of the area scene;

[0154] Characterizing the three-dimensional deflection point Coordinates are based on the current layer reference. Shaft height plus The axial offset is obtained to ensure that the point falls within the effective height range of the current layer.

[0155] in, The coordinates of a two-dimensional deflection point on a certain layer can be generated by the "regional scene coordinate formula" and are based on the polygon coordinates of each layer of the bridge surface obtained by topological analysis. It is the deflection basis of the three-dimensional scene on the two-dimensional plane of that layer.

[0156] As the current layer benchmark The axis height, derived from the layer height parameter in the topology analysis, such as 8 meters for the upper layer and 3 meters for the lower layer, determines the reference position of the layer in the vertical direction.

[0157] for The axial offset, for example, if the offset in this direction meets ±0.5 meters, ensures that the point falls within the effective height range of the current layer and avoids crossing layers.

[0158] The coordinates of the three-dimensional deflection points in the generated 3D scene are determined by combining two-dimensional planar deflection and vertical offset to ensure they fall accurately within the effective area of ​​the 3D scene.

[0159] In a 3D scene, the number of points within each layer's area is positively correlated with the layer's size. This avoids wasting resources by over-placing points in smaller layers. Specifically, the number of points within each layer's area... satisfy:

[0160]

[0161] The number of deflection points required for each layer is positively correlated with the area of ​​the bridge deck at that layer; that is, the larger the area of ​​the layer, the more deflection points are allocated. This avoids wasting resources by over-distributing points in smaller layers, thus achieving a reasonable allocation of points. The total number of deflection points is then used to determine the optimal allocation. Multiply by the ratio of the current layer's area to the total area of ​​all layers, and then round to the nearest integer to obtain the number of deflection points for the current layer.

[0162] Optionally, Figure 3d This application provides a flowchart illustrating the process of generating deflection points in a point-like scene, as shown below. Figure 3d As shown, when the target scene is a point-like scene, deflection points are generated within the deflection constraint region based on topological feature parameters and the deflection strategy, including:

[0163] Based on the topological feature parameters and the deflection strategy, a fourth initial deflection point is generated on the annular surface with different radius intervals, with the center point of the deflection constraint region as the annular center.

[0164] When the number of fourth initial deflection points on the annular surface is greater than the preset number, the distribution position of the fourth initial deflection points on the annular surface is adjusted to determine the deflection points within the deflection constraint area.

[0165] Specifically, based on topological feature parameters and deflection strategies, generating a fourth initial deflection point on annular surfaces with different radius intervals (e.g., 3 meters / 5 meters) with the center point of the deflection constraint region as the ring center can mean dividing the deflection constraint region into multiple concentric rings, each ring corresponding to a specific radius range, and uniformly generating a fixed number of points (e.g., 20 per ring) on ​​each ring layer to achieve layered coverage of the annular region. By combining this with the uniform generation of offset points along the annular trajectory, the distribution of points is ensured to conform to the geometric characteristics of the annular road, improving spatial fit.

[0166] When the number of fourth initial deflection points generated on a certain annular surface exceeds a preset threshold, the distribution of these points on the annular surface can be dynamically adjusted. For example, the point spacing can be optimized, similar points can be merged, or resampling can be performed to avoid local over-density leading to redundancy. At the same time, an invalid point filtering mechanism that exceeds the maximum radius (such as 5 meters) can be combined to remove out-of-bounds points and finally retain a reasonable distribution set of points located within the effective annular area.

[0167] This allows for control over the number and effectiveness of points while maintaining coverage integrity, outputting an efficient and non-redundant set of deflection points suitable for ring-shaped scenes, and improving modeling accuracy and computational efficiency under complex topology structures.

[0168] In this embodiment of the application, the point scene can generate a multi-radius ring deflection point with the center point as the center, so as to cover the full range represented by "center point → surrounding radiation area".

[0169] Among them, the deflection point of the point-like scene satisfies:

[0170]

[0171]

[0172] in, It can refer to calculating the deflection point. Through the center point It is obtained by adding the product of the annular radius and the cosine of the annular angle.

[0173] It can refer to calculating the deflection point. Through the center point It is obtained by adding the product of the annular radius and the sine of the annular angle.

[0174] in, The coordinates of the center point of the point scene are derived from the center point parameters of the point scene obtained from topological analysis, and serve as the reference center for the generation of the ring deflection point.

[0175] The radius of the ring is 1 meter, 3 meters, or 5 meters. It is derived from the surrounding risk radius parameter in the topology analysis and is used to determine the size range of the ring.

[0176] It is a ring-shaped angle, and its distribution is uniform in. Within the range. When each ring generates 20 points, This allows for the generation of uniformly distributed points on a ring using different angle values.

[0177] The coordinates of the deflection point for the generated point scene are new position points calculated based on the center point, the ring radius, and the ring angle.

[0178] In this embodiment of the application, when there are multiple target scenes, deflection points in different types of target scenes can be generated through multi-threaded processing.

[0179] Optionally, before performing topological analysis on the scene data of the target scene according to the scene type of the target scene, and determining the deflection constraint region and topological feature parameters of the target scene, the method includes:

[0180] Acquire geospatial data of the target scene;

[0181] Geospatial data is cleaned and processed to obtain scene data;

[0182] The scene type of the target scene is determined based on the topological relationship between the coordinate data in the scene data.

[0183] Among them, the geospatial data of the target scene can refer to various data related to a specific geographical location, including but not limited to coordinate points, terrain information, building distribution, road network, etc.

[0184] Cleaning geospatial data to obtain scene data can refer to cleaning and optimizing the raw data through a series of data preprocessing steps: First, invalid coordinates, such as points with incorrect formats or latitude and longitudes outside the reasonable geographical range, are removed to eliminate obviously incorrect or unreasonable information; at the same time, it can also include operations such as filling missing values, eliminating duplicate records, unifying coordinate systems and standardizing data formats, so as to ensure that the output scene data meets the requirements of accuracy, completeness and consistency.

[0185] Determining the scene type of a target scene based on the topological relationships between coordinate data in the scene data can be achieved by analyzing the connectivity, closure, and dimensional characteristics between coordinate points to identify its spatial structure category.

[0186] If the coordinate string forms a non-closed continuous line segment (such as "lon1, lat1; lon2, lat2; lon3, lat3"), it is determined to be a linear scene, which is commonly seen in road centerlines or trajectory paths.

[0187] If the coordinate strings are connected end to end to form a closed polygon, it is classified as a region scene, which is suitable for representing plots of land, intersections, or drivable areas;

[0188] If the coordinates contain Z-axis elevation information (such as "lon, lat, altitude"), it is identified as a 3D scene, used to describe three-dimensional structures such as overpasses, tunnels, or multi-level roads.

[0189] If the data consists of only a single coordinate or a discrete set of points without topological connections, it is considered a point-like scene, which is often used to locate objects such as landmarks and points of interest.

[0190] Therefore, core features can be extracted for different scenarios, thus providing data support for subsequent steps; and abstract topological features can be transformed into "computable geometric ranges". When generating dynamic deflection points, points are generated only within the constraint area, ensuring the initial validity of the deflection points.

[0191] Figure 4 The flowchart illustrating the process of determining the scenario type of the target scenario provided in this application is as follows: Figure 4 As shown, this process can take the set of coordinates to be deflected as input, and after preprocessing steps such as whitelist verification, format standardization, and coordinate system unification, it enters the core scene type detection and classification module. Among them:

[0192] During the classification phase, the system first determines the scene type by detecting whether the coordinate string is a continuous line segment, a closed polygon, contains Z-axis information, or is a discrete point set: if it is a continuous line segment, it is classified as a linear scene; if it is a closed polygon, it is classified as a region scene; if it contains 3D coordinates, it is identified as a 3D scene; if it is only an isolated point or point set, it is considered a point scene. For each type, the system calls the corresponding deflection processing sub-process and performs subsequent calculations based on topological feature parameters and deflection strategies.

[0193] For linear scenes, a buffer zone can be generated centered on the line segment, the relationship between curvature and length can be calculated, a dynamic offset domain can be constructed, and deflection points distributed along the normal direction can be generated in conjunction with the road orientation. For regional scenes, the center point and boundary geometric features are extracted to generate inner / outer ring zones or uniform grid offsets, supporting layered or multi-level offsets. For 3D scenes, they can be processed in layers along the Z-axis, generating 2D offset points layer by layer and superimposing elevation information to achieve accurate 3D spatial modeling. For point-like scenes, a circular or square offset domain is generated centered on a single point, and the distance between points is calculated to avoid overlap, forming a reasonable point set expansion. Thus, all processing results can be summarized after unordered feature extraction, and a comprehensive output containing offset points for each scene type can be generated by merging the parsing results for subsequent processing.

[0194] Optionally, output and display the deflection points in the target scene, including:

[0195] According to the preset verification rules, the deflection points in the target scene are verified to determine the number of valid points in the target scene; the verification rules are determined according to the scene type of the target scene, and the number of valid points is the number of deflection points that are not in the boundary part of the deflection constraint region.

[0196] If the number of valid points in the target scene meets the preset requirements, then the deflection points in the target scene will be output and displayed.

[0197] If the number of valid points in the target scene does not meet the preset requirements, the deflection points will be supplemented until the number of valid points in the target scene meets the preset requirements. Then, the deflection points in the target scene will be output and displayed.

[0198] The verification rules refer to the verification rules set according to the type of the target scene, which are used to evaluate whether the deflection point is located within the allowed area and meets the specific requirements of that scene type. For example, in a linear scene, the verification rules may include the distance limit between the deflection point and the original line segment, the range of angle changes, etc.; in a region scene, the verification rules may be the distance threshold from the deflection point to the boundary, the internal grid distribution density, etc.

[0199] The number of valid points refers to the number of deflection points that, after verification, are not located at the boundary of the deflection constraint region, i.e., the total number of deflection points inside the deflection constraint region. These points neither violate any spatial constraints nor are located at special positions at the boundary of the deflection constraint region; they are deflection points completely located within the region where operation is permitted.

[0200] When the number of valid points meets the preset requirements, the number of deflection points that have passed the verification has reached the expected standard, indicating that the deflection results are of high quality and can accurately reflect the spatial characteristics and topological relationships of the original scene. At this time, these deflection points can be directly output and displayed for subsequent map building, path planning or other geographic information applications.

[0201] If the number of valid points does not meet the preset requirements, it indicates that the current set of deflection points cannot fully cover or represent the key features of the original scene, and there may be problems such as data loss or excessive deviation. Therefore, further data processing and optimization are required to ensure the accuracy and completeness of the final results.

[0202] In this embodiment of the application, data processing can be performed by completing the deflection points. The completion process can refer to increasing the number of qualified deflection points to achieve the preset number of valid points.

[0203] In some embodiments, the method may include: generating new points based on existing deflection points through interpolation, expanding using a point set from neighboring regions, or recalculating the positions of deflection points within a specific region. The completion process must ensure that the newly added points also satisfy the scene type validation rules and reflect the spatial distribution characteristics of the original scene as realistically as possible.

[0204] In this embodiment of the application, when there are at least two target scenarios, the method further includes:

[0205] Based on GPU multi-threaded parallel tasks, the deflection points of the target scene are sent to the corresponding threads for verification according to different scene types to determine the number of valid points in different target scenes.

[0206] Based on the number of valid points in different target scenes, output and display the deflection points in the target scene.

[0207] Based on GPU multi-threaded parallel tasks, all deflection points to be verified in the target scene can be dynamically allocated to different GPU thread blocks for parallel processing according to their scene type (such as linear, regional, 3D, or point). Each thread block loads the verification rules corresponding to the scene type (such as boundary distance thresholds, curvature constraints, elevation validity, etc.), independently performs spatial compliance judgment on the allocated deflection points, and counts the number of valid points in the sub-scene. Thus, by fully utilizing the large-scale concurrent computing power of GPUs through parallel design, the verification efficiency of large-scale deflection point sets can be improved, which is especially suitable for the processing of massive spatial data involved in high-precision maps or city-level autonomous driving simulations.

[0208] After each thread completes the verification and summarizes the number of valid points in different target scenarios, the system judges based on a preset validity threshold: if the number of valid points in a scenario meets the threshold, the deflection points of that scenario are directly output and visualized; if the threshold is not met, a completion mechanism (such as local resampling or density enhancement) is triggered, and verification is performed again after completion until the requirements are met. Finally, all the verified deflection points are categorized and integrated according to scenario type, and output and displayed uniformly. This ensures that the results conform to geometric and topological constraints and have sufficient spatial coverage and task applicability.

[0209] Figure 5 This is a flowchart illustrating the process of completing the deflection point according to an embodiment of this application, as shown below. Figure 5 As shown, the method includes:

[0210] Part 1: Data Preparation Stage

[0211] 1. Input reception: Receive the deflection point set;

[0212] 2. Data initialization: This includes processing the deflection points in the deflection point set into a unified coordinate format (such as WGS84 latitude and longitude); and labeling each point with a "scene type label" (linear / regional / three-dimensional / point).

[0213] 3. Parameter loading: Load the key parameters of the current scene (coordinates of the minimum bounding rectangle, boundary data, Z-axis range of the 3D scene, center point and radius of the point scene).

[0214] Second: Level 1 Validation (Pre-filtering):

[0215] 1. Verification logic: Check point by point whether it is within the "minimum bounding rectangle / 3D bounding box" (different bounding ranges correspond to different scenarios);

[0216] 2. Result routing: If the result meets the criteria, it enters the "Precise Verification Queue"; if it does not meet the criteria, it is marked as "Invalid Pre-filter", the reason for invalidity is recorded (such as "outside the bounding rectangle of the region") and stored in the log.

[0217] 3. Statistical feedback: Calculate "Pre-filter pass rate = number of candidate points / total number of input points".

[0218] Three: Second-level verification (GPU parallel precise verification):

[0219] 1. Task Startup: Split the "Precise Verification Queue" according to scenario type and start GPU multi-threaded parallel tasks;

[0220] 2. Scene Differentiation Validation: Linear / Region Scenes: GPU Raycasting → Fire a ray in the positive X-axis direction; if the number of intersections is odd, it is valid (within the boundary); 3D Scenes: GPU Layered Validation → First determine if the Z-axis is within the current layer range (e.g., the upper layer is 8 ± 0.5 meters), then use raycasting to determine the planar position; Point Scenes: GPU Distance Calculation → Use the Euclidean distance formula to determine if the distance between the point and the center point is ≤ the maximum radius (e.g., 5 meters);

[0221] 3. Results summary: Mark each point as "valid / invalid", and record the specific reason for invalid points (such as "boundary exceeded" or "Z-axis exceeded").

[0222] Step 4: Complete the valid points:

[0223] 1. Quantity judgment: Check whether the number of valid points has reached the target number (N_target);

[0224] 2. Completion Logic (if valid points < N_target): Calculate the completion gap: N_gap = N_target - number of current valid points; Completion point generation: Reuse the original scene strategy of the "Dynamic Deflection Point Generation Module" (such as linear deflection along the direction) to generate only N_gap points; Completion point verification: The completion points skip pre-filtering and directly enter "GPU precise verification"; Retry limit: If the completion point verification is still invalid, it can be retried a maximum of 5 times;

[0225] 3. Anomaly Handling: If the target number cannot be reached after 5 retries, output an alarm log (it is recommended to adjust the deflection parameters, such as expanding the offset range), and output the current maximum valid point set.

[0226] Step 5: Output the results:

[0227] 1. Final output content: Set of effective deflection points for each scene (including coordinates and scene labels);

[0228] Validation and completion report (pre-filter pass rate, final effectiveness rate, invalidity reason classification, number of completions);

[0229] 2. Module exit: The results are passed to the "Geofence Whitelist Scenario Test" stage to complete the verification and completion process.

[0230] Optionally, after outputting and displaying the deflection points in the target scene, the method further includes:

[0231] Determine the sensitive deflection points and the number of sensitive deflection points among the deflection points;

[0232] Based on the sensitive deflection points and the number of sensitive deflection points, determine the proportion and / or distribution of sensitive deflection points among the deflection points;

[0233] Adjust the deflection strategy based on the proportion and / or distribution of sensitive deflection points among the deflection points.

[0234] Sensitive deflection points refer to those points among the generated deflection points that are located in sensitive areas or have specific security, privacy, or regulatory restrictions. For example, deflection points near protected geographical entities such as military facilities, government offices, residential areas, schools, and hospitals, or points falling into prohibited data collection / mapping areas (such as classified zones or no-fly zones). These points may be marked as "sensitive" due to national security, personal privacy, or compliance risks, and require special treatment in subsequent processing.

[0235] The proportion and / or distribution of sensitive deflection points among deflection points can refer to the proportion of sensitive deflection points among all deflection points (e.g., 5%), as well as their spatial clustering characteristics or location patterns (e.g., concentrated in a certain road segment, distributed in a strip in the city center, or scattered in multiple areas).

[0236] Adjusting the deflection strategy can refer to dynamically modifying deflection parameters or rules based on the proportion and distribution characteristics of sensitive deflection points to reduce the risk of sensitive exposure. For example, reducing the offset distance in areas with a high density of sensitive points, disabling certain deflection directions, switching to a more conservative deflection mode (such as only offsetting towards the inside of the road), or directly blocking deflection operations within specific geofences; at the same time, it can be combined with map desensitization mechanisms to perturb, blur, or remove the coordinates of already generated sensitive points to ensure that the output results meet data security and compliance requirements.

[0237] Figure 6a This is a schematic diagram of the sensitivity testing process provided in the embodiments of this application, such as... Figure 6a As shown, the method includes:

[0238] 1. Data preprocessing: Convert valid coordinate points into the format required by the test interface and associate them with corresponding scene information to ensure a clear test context;

[0239] 2. Group testing: Call different geofencing test interfaces according to the scenario type to match the business judgment logic of various scenarios;

[0240] The process involves batch inputting valid coordinate points, then calling corresponding judgment interfaces based on the test scenario type: For linear scenarios, the geofence boundary judgment interface is called to determine if the point is within a linear area boundary; for regional scenarios, the area-inside-outside judgment interface is called to determine if the point is within a polygonal area; for 3D scenarios, the layered space judgment interface is called to determine if the floor where the point is located meets the set elevation range; and for point-based scenarios, the surrounding area judgment interface is called to determine if the point falls within a specified radius. By calling different interfaces in a categorized manner, accurate location determination for various spatial forms is achieved.

[0241] 3. Result Analysis: By comparing the results returned by the interface with the expected values, identify abnormal judgments (such as being judged as outside when it should be inside the fence), and record the details of the abnormal points;

[0242] 4. Output report: Summarize test coverage, pass rate and anomaly distribution to provide a basis for optimizing geofencing whitelist rules.

[0243] Figure 6b This is a flowchart illustrating the sensitivity determination process provided in an embodiment of this application, as shown below. Figure 6b As shown, the method includes:

[0244] 1. Data Input and Preprocessing: Receive sensitive test results and scene topology data, group them by scene type and associate coordinate attributes to lay the foundation for subsequent analysis;

[0245] 2. Calculation of scene-specific indicators: Design exclusive sensitive indicators for four types of scenes (such as focusing on calculating the proportion of sensitive indicators in the edge zone for regional scenes, and analyzing the layered sensitive indicator distribution for three-dimensional scenes), which are in line with the topological characteristics of different scenes;

[0246] 3. Threshold comparison and risk identification: Screen abnormal sensitive areas by preset thresholds (which can be adjusted according to business needs) and mark high-risk sub-regions with a continuous concentration of sensitive points (such as road bends and area edge zones).

[0247] 4. Comprehensive Judgment and Output: Based on all indicators, the overall sensitivity level of the scenario is given, and an assessment report containing quantitative data, high-risk coordinates and judgment criteria is generated to provide decision support for the optimization of whitelist rules.

[0248] For linear scenarios, three key indicators can be calculated in parallel: first, the percentage of sensitive points along curves (the ratio of sensitive points within a curve area to the total number of test points along curves); second, the percentage of sensitive points along straight sections (the ratio of sensitive points within a straight section to the total number of test points along straight sections); and third, the overall sensitivity rate (the ratio of the total number of sensitive points in the linear scenario to the total number of test points in that scenario). These three indicators reflect the distribution of sensitivity risks from both local (curves and straight sections) and overall perspectives. Based on these indicator results, high-risk road sections are marked to identify road sections with a high risk of sensitivity incidents.

[0249] For regional scenarios, three key indicators can be calculated in parallel: first, the edge zone sensitivity ratio, which is the ratio of the number of sensitive points in the edge zone to the total number of test points in the edge zone; second, the interior zone sensitivity ratio, which is the ratio of the number of sensitive points in the interior zone to the total number of test points in the interior zone; and third, the overall sensitivity rate, which is the ratio of the total number of sensitive points in the entire regional scenario to the total number of test points in that region. These three indicators reflect the distribution characteristics of sensitivity risks from the edge, interior, and overall levels, respectively. Based on the results of these indicators, a high-risk edge zone marking operation is performed to identify edge regions with high sensitivity risks.

[0250] For 3D scenes, three key indicators can be calculated in parallel: first, layered sensitivity distribution, which is the proportion of sensitive points in each floor to the total number of test points on that floor, reflecting the sensitivity situation on different floors; second, vertical sensitivity continuity, which is the number of floors with sensitive points appearing in three or more consecutive vertical layers, used to determine whether the sensitivity risk has a vertical extension characteristic; and third, overall sensitivity rate, which is the ratio of the total number of sensitive points in the entire 3D scene to the total number of test points in the scene, reflecting the overall risk level. These three indicators assess sensitivity risk from layered, vertical continuity, and global perspectives, respectively. Based on the results of these indicators, a high-risk floor marking operation is performed to identify floors with a high sensitivity risk.

[0251] For point-like scenarios, three key indicators can be calculated in parallel: first, the ring-shaped sensitive distribution, which is the proportion of sensitive points in each ring area of ​​different radii to the total number of test points in that ring area, reflecting the spatial ring-shaped distribution characteristics of sensitive points; second, the core area sensitive rate, which is the ratio of the number of sensitive points to the total number of test points in a 1-meter radius centered on the target point, used to assess the sensitivity risk level in the core area; and third, the overall sensitivity rate, which is the ratio of the total number of sensitive points in the entire point-like scenario to the total number of test points in the scenario, reflecting the overall risk level. These three indicators assess sensitivity risk from the perspectives of ring-shaped distribution, core area, and global perspective. Based on the results of these indicators, a high-risk ring area marking operation is performed to identify ring areas with high sensitivity risk, providing a basis for subsequent risk assessment and handling.

[0252] After calculating the sensitivity indicators for all scenarios, a unified risk assessment process is initiated: first, the sensitivity thresholds are compared (configurable); then, the sensitivity level of the scenario is comprehensively determined (high / medium / low / no sensitivity); finally, a sensitivity assessment report is generated, and the report and high-risk coordinate information are output through the module's output. This enables differentiated sensitivity analysis and hierarchical management for scenarios with different geometric shapes.

[0253] The coordinate deflection method based on scene geometric features provided in this application can significantly improve the effectiveness of test cases and reduce invalid resource consumption. In traditional linear scenes (such as road buffer50), the vertical offset easily exceeds the strip area, resulting in an invalid point ratio of over 30%. By using the "regional shape adaptive deflection algorithm," combined with scene geometric features to design differentiated deflection rules, and then using ray casting to verify the validity of invalid points and automatically complete them, the effective test case ratio is increased to ≥95%, completely solving the problem of "deflection coordinates exceeding scene boundaries."

[0254] To address the issue of traditional 3D scenarios failing to consider elevation layering, which can cause offset points to fall on non-accessible layers, test points are generated based on layer height to ensure that offset points align with actual business access ranges. This eliminates test cases that are "geometrically valid but business-invalid" and reduces the waste of test resources due to invalid interface requests.

[0255] To address the differences in point / line / plane / three-dimensional spatial forms across eight major scenarios, a customized deflection strategy was designed to avoid the incomplete coverage of special scenarios (such as tunnels and toll stations) by traditional "general deflection" methods, thus achieving comprehensive testing coverage across all scenarios without any blind spots.

[0256] Traditional uniform point distribution leads to excessive point placement in low-risk areas, resulting in wasted computing resources. Dynamically allocating point resources according to risk distribution ensures testing depth in high-risk areas while reducing redundant computing in low-risk areas, achieving "precise resource allocation" and improving overall testing efficiency.

[0257] By parsing the unique geometric parameters of eight major scenarios through MIF files and dynamically calling classification and deflection rules based on the parameters, this application can flexibly adapt to different types of geofencing whitelist scenarios without having to repeatedly develop algorithms for a single scenario, thus reducing the technical adaptation cost of cross-scenario testing.

[0258] Figure 7 A schematic diagram of the coordinate deflection device based on scene geometric features provided in this application is shown below. Figure 7 As shown, the coordinate deflection device 70 based on scene geometric features provided in this embodiment includes:

[0259] The topology parsing module 701 is used to perform topology parsing on the scene data of the target scene according to the scene type of the target scene, and to determine the deflection constraint region and topology feature parameters of the target scene.

[0260] The generation module 702 is used to generate deflection points within the deflection constraint region based on topological feature parameters and deflection strategy; the deflection strategy is determined according to the scene type of the target scene.

[0261] Display module 703 is used to output and display the deflection points in the target scene.

[0262] In one possible implementation, when the target scene is a linear scene, the generation module 702 can also be specifically used for:

[0263] The deflection direction and road curvature are determined based on the topological characteristic parameters and the deflection strategy.

[0264] Based on the deflection direction, generate the first initial deflection point within the target scene;

[0265] Adjust the point density of the first initial deflection point within the target scene based on the road curvature;

[0266] Based on the deflection constraint region, the first initial deflection point within the target scene is filtered to determine the deflection point within the deflection constraint region.

[0267] In one possible implementation, when the target scene's scene type is a region scene, the generation module 702 can also be specifically used for:

[0268] Based on the topological characteristic parameters and the deflection strategy, the planar partitioning region in the deflection constraint region is determined;

[0269] A second initial deflection point is generated within the planar partitioned region; wherein the density and / or number of the second initial deflection points differ depending on the position of the planar partitioned region within the deflection constraint region.

[0270] The second initial deflection points between different planar regions are deduplicated to determine the deflection points within the deflection constraint region.

[0271] In one possible implementation, when the target scene type is a stereoscopic scene, and when the target scene type is a region scene, the generation module 702 can also be specifically used for:

[0272] Based on topological feature parameters and deflection strategy, the layered regions in the deflection constraint region are determined; different layered regions have different heights in the stereo scene.

[0273] A third initial deflection point is generated within the layered region; the larger the area of ​​the layered region, the more third initial deflection points are generated within the layered region.

[0274] Based on the spatial coordinates of the third initial deflection point, the third initial deflection point is filtered to determine the deflection points within the deflection constraint region.

[0275] In one possible implementation, when the target scene is a point-like scene, the generation module 702 can also be specifically used for: based on the topological feature parameters and the deflection strategy.

[0276] Based on the topological feature parameters and the deflection strategy, a fourth initial deflection point is generated on the annular surface with different radius intervals, with the center point of the deflection constraint region as the annular center.

[0277] When the number of fourth initial deflection points on the annular surface is greater than the preset number, the distribution position of the fourth initial deflection points on the annular surface is adjusted to determine the deflection points within the deflection constraint area.

[0278] In one possible implementation, the topology resolution module 701 can also be specifically used for:

[0279] Acquire geospatial data of the target scene;

[0280] Geospatial data is cleaned and processed to obtain scene data;

[0281] The scene type of the target scene is determined based on the topological relationship between the coordinate data in the scene data.

[0282] In one possible implementation, the display module 703 can also be specifically used for:

[0283] According to the preset verification rules, the deflection points in the target scene are verified to determine the number of valid points in the target scene; the verification rules are determined according to the scene type of the target scene, and the number of valid points is the number of deflection points that are not in the boundary part of the deflection constraint region.

[0284] If the number of valid points in the target scene meets the preset requirements, then the deflection points in the target scene will be output and displayed.

[0285] If the number of valid points in the target scene does not meet the preset requirements, the deflection points will be supplemented until the number of valid points in the target scene meets the preset requirements. Then, the deflection points in the target scene will be output and displayed.

[0286] In one possible implementation, when there are at least two target scenarios, the display module 703 can also be specifically used for:

[0287] Based on GPU multi-threaded parallel tasks, the deflection points of the target scene are sent to the corresponding threads for verification according to different scene types to determine the number of valid points in different target scenes.

[0288] Based on the number of valid points in different target scenes, output and display the deflection points in the target scene.

[0289] In one possible implementation, the display module 703 can also be specifically used for:

[0290] Determine the sensitive deflection points and the number of sensitive deflection points among the deflection points;

[0291] Based on the sensitive deflection points and the number of sensitive deflection points, determine the proportion and / or distribution of sensitive deflection points among the deflection points;

[0292] Adjust the deflection strategy based on the proportion and / or distribution of sensitive deflection points among the deflection points.

[0293] The coordinate deflection device based on scene geometric features provided in this embodiment can execute the method provided in the above method embodiment. Its implementation principle and technical effect are similar, and will not be described in detail here.

[0294] Figure 8 A schematic diagram of the structure of the electronic device provided in this application. Figure 8 As shown, the electronic device 80 provided in this embodiment includes at least one processor 801 and a memory 802. Optionally, the device 80 further includes a communication component 803. The processor 801, memory 802, and communication component 803 are connected via a bus 804.

[0295] In a specific implementation, at least one processor 801 executes computer execution instructions stored in memory 802, causing at least one processor 801 to perform the above-described method.

[0296] The specific implementation process of processor 801 can be found in the above method embodiments, and its implementation principle and technical effect are similar. It will not be repeated here.

[0297] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0298] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0299] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0300] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0301] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-described method.

[0302] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0303] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.

[0304] The division of units is merely a logical functional division; 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 coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0305] 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.

[0306] In addition, the functional units in the various embodiments of the present invention 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.

[0307] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium 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 invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0308] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0309] Finally, it should be noted that other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein, and is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A coordinate deflection method based on scene geometric features, characterized in that, include: Based on the scene type of the target scene, perform topology analysis on the scene data of the target scene to determine the deflection constraint region and topology feature parameters of the target scene; Based on the topological feature parameters and the deflection strategy, deflection points are generated within the deflection constraint region; the deflection strategy is determined according to the scene type of the target scene. Output and display the deflection points in the target scene.

2. The method according to claim 1, characterized in that, When the target scene is a linear scene, generating deflection points within the deflection constraint region based on the topological feature parameters and the deflection strategy includes: Based on the topological feature parameters and deflection strategy, the deflection direction and road curvature are determined; Based on the deflection direction, a first initial deflection point is generated within the target scene; Adjust the point density of the first initial deflection point within the target scene according to the road curvature; Based on the deflection constraint region, the first initial deflection point within the target scene is filtered to determine the deflection point within the deflection constraint region.

3. The method according to claim 1, characterized in that, When the target scene is a regional scene, generating deflection points within the deflection constraint region based on the topological feature parameters and the deflection strategy includes: Based on the topological feature parameters and the deflection strategy, the planar partitioning region in the deflection constraint region is determined; A second initial deflection point is generated within the plane division region; wherein, the density and / or number of the second initial deflection points differ depending on the location of the plane division region within the deflection constraint region. The second initial deflection points between different planar division regions are deduplicated to determine the deflection points within the deflection constraint region.

4. The method according to claim 1, characterized in that, When the target scene is a stereoscopic scene, generating deflection points within the deflection constraint region based on the topological feature parameters and the deflection strategy includes: Based on the topological feature parameters and the deflection strategy, the layered regions in the deflection constraint region are determined; different layered regions have different heights in the stereoscopic scene. A third initial deflection point is generated within the layered region; the larger the area of ​​the layered region, the more points the third initial deflection point will have within the layered region. Based on the spatial coordinates of the third initial deflection point, the third initial deflection point is filtered to determine the deflection points within the deflection constraint region.

5. The method according to claim 1, characterized in that, When the target scene is a point-like scene, generating deflection points within the deflection constraint region based on the topological feature parameters and the deflection strategy includes: Based on the topological feature parameters and the deflection strategy, a fourth initial deflection point is generated on the annular surface with different radius intervals, taking the center point of the deflection constraint region as the ring center. When the number of fourth initial deflection points on the annular surface is greater than the preset number, the distribution position of the fourth initial deflection points on the annular surface is adjusted to determine the deflection points within the deflection constraint area.

6. The method according to any one of claims 1 to 5, characterized in that, Before performing topological parsing on the scene data of the target scene according to the scene type of the target scene to determine the deflection constraint region and the topological feature parameters of the target scene, the method includes: Obtain the geospatial data of the target scene; The geospatial data is cleaned to obtain scene data; The scene type of the target scene is determined based on the topological relationship between the coordinate data in the scene data.

7. The method according to claim 1, characterized in that, The output, which displays the deflection points in the target scene, includes: According to the preset verification rules, the deflection points in the target scene are verified to determine the number of valid points in the target scene; the verification rules are determined according to the scene type of the target scene, and the number of valid points represents the number of deflection points that are not located in the boundary part of the deflection constraint region. If the number of valid points in the target scene meets the preset requirements, then the deflection points in the target scene are output and displayed; If the number of valid points in the target scene does not meet the preset requirement, the deflection points are supplemented until the number of valid points in the target scene meets the preset requirement after supplementation. Then, the deflection points in the target scene are output and displayed.

8. The method according to claim 7, characterized in that, When there are at least two target scenarios, the method further includes: Based on GPU multi-threaded parallel tasks, the deflection points of the target scene are sent to the corresponding threads for verification according to different scene types to determine the number of valid points in different target scenes. Based on the number of valid points in different target scenes, output and display the deflection points in the target scene.

9. The method according to any one of claims 1 to 8, characterized in that, After outputting and displaying the deflection points in the target scene, the method further includes: Determine the sensitive deflection points and the number of sensitive deflection points among the deflection points; Based on the sensitive deflection points and the number of sensitive deflection points, determine the proportion and / or distribution of the sensitive deflection points among the deflection points; The deflection strategy is adjusted based on the proportion and / or distribution of the sensitive deflection points among the deflection points.

10. A coordinate deflection device based on scene geometric features, characterized in that, include: The topology parsing module is used to perform topology parsing on the scene data of the target scene based on the scene type of the target scene, and to determine the deflection constraint region and topology feature parameters of the target scene. The generation module is used to generate deflection points within the deflection constraint region based on topological feature parameters and deflection strategy. The deflection strategy is determined based on the scenario type of the target scene; The display module is used to output and display the deflection points in the target scene.

11. An electronic device, characterized in that, include: Memory, processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory, causing the processor to perform the method as described in any one of claims 1-9.

12. A computer-readable storage medium / computer program product, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any one of claims 1-9; The computer program product includes a computer program that, when executed by a processor, implements the method described in any one of claims 1-9.

Images