Vector road network generation method and device, equipment and medium

Abstract

The invention discloses a vector road network generation method and device, equipment and a medium, and relates to the technical field of road network generation, and the method comprises the steps: obtaining a plurality of road segmentation vectorized images of a target range; identifying a central road and an edge road of any road segmentation vectorized image; performing cross-mapsheet splicing and road optimization on each edge road to obtain a boundary road network; and integrating and optimizing each center road and the boundary road network to obtain a target vector road network image corresponding to the target range. According to the method, a complicated global splicing problem is decomposed into a local processing stage and a boundary processing stage through a strategy of'division and treatment and first inside and then outside ', so that the calculation complexity is reduced, the efficiency of processing a large-scale region is improved, and full-process automatic processing from an image segmentation result to a high-quality vector road network is further realized, so that the cost is reduced, and the efficiency is improved. And the optimization efficiency is improved.

Description

Technical Field

[0001] This application relates to the field of road network generation technology, and in particular to a vector road network generation method, apparatus, device and medium. Background Technology

[0002] In recent years, with the advancement of the rural revitalization strategy, the demand for digital surveying and mapping of vast natural villages has become increasingly urgent. Semantic segmentation technology, represented by deep learning, can automatically identify road areas from high-resolution remote sensing images, significantly improving efficiency.

[0003] However, the results directly generated by this method are usually raster images or discrete road patches, which have technical defects such as poor geometric shape, missing topological relationships, data redundancy and noise, and difficulty in stitching multiple map sheets. These defects result in the generated road network failing to form a connected topological network, lacking practical value for navigation and analysis, and requiring huge computational load and low efficiency when processing large-scale, multi-map sheet data.

[0004] Therefore, how to transform the semantic segmentation results of rural roads in remote sensing images into topologically correct, geometrically accurate, and seamlessly stitched rural vector road networks in an economical and efficient manner is not only a technical bottleneck in the current field of automated rural mapping, but also an urgent need to achieve large-scale and rapid mapping. Summary of the Invention

[0005] The main objective of this application is to provide a method, apparatus, device, and medium for generating vector road networks, which aims to transform the semantic segmentation results of rural roads in remote sensing images into topologically correct, geometrically accurate, and seamlessly stitched rural vector road networks in an economical and efficient manner.

[0006] To achieve the above objectives, this application proposes a vector road network generation method, the method comprising: Obtain multiple road segmentation vectorized images within the target area; Identify the central road and edge roads in any of the road segmentation vectorized images; The edge roads are spliced ​​across map sheets and optimized to obtain the boundary road network. The central roads and the boundary road network are integrated and optimized to obtain the target vector road network image corresponding to the target range.

[0007] In one embodiment, identifying the center road and edge roads of any of the road segmentation vectorized images includes: For any of the road segmentation vectorized images, geometric smoothing and centerline extraction are performed on the road segmentation vectorized images to obtain a first optimized image; Based on the first optimized image, new path segments are constructed and integrated for optimization to obtain a second optimized image; By weighting the global path segments of the second optimized image and merging the main path segments in the second optimized image, a third optimized image is obtained. Based on the third optimized image, the central road and edge roads of the road segmentation vectorized image are identified.

[0008] In one embodiment, the step of constructing and integrating new path segments based on the first optimized image to obtain a second optimized image includes: Construct a first road topology map corresponding to the first optimized image, and clean up invalid line segments in the first road topology map to obtain a fourth optimized image; A second road topology map corresponding to the fourth optimized image is constructed, and new path segments are added to the second road topology map through spatial indexing to obtain the second optimized image.

[0009] In one embodiment, the process of determining the invalid line segment includes: Obtain the degree value corresponding to each line segment in the first road topology map; Line segments whose degree values ​​conform to the preset isolated line segment degree values ​​are considered invalid line segments. Obtain the length of the line segment whose degree value meets the preset hanging line segment degree value, and take the line segment whose length meets the preset pruning length as the invalid line segment.

[0010] In one embodiment, identifying the central road and edge roads of the road segmentation vectorized image based on the third optimized image includes: After cleaning up the invalid line segments in the third optimized image, the newly added path line segments in the third optimized image are constructed and integrated through spatial indexing and path extension processing to obtain the fifth optimized image. Obtain the boundary lines of the fifth optimized image, and determine the boundary influence region in the fifth optimized image based on the preset distance and each boundary line; Determine the positional relationship between the global path segment of the fifth optimized image and the boundary influence region; Based on the aforementioned positional relationships, the central road and edge roads of the road segmentation vectorized image are determined.

[0011] In one embodiment, the step of weighting the global path segments of the second optimized image and merging the main path segments in the second optimized image to obtain the third optimized image includes: The global path segments of the second optimized image are weighted to determine the segment weights corresponding to the global path segments of the second optimized image; For any weighted path segment in the global path segments, determine the longest path corresponding to the weighted path segment; Based on the longest paths described, path merging is performed to determine the main path segments in the second optimized image, thus obtaining the third optimized image.

[0012] In one embodiment, the step of performing cross-map sheet stitching and road optimization on each of the edge roads to obtain a boundary road network includes: Based on the location coordinates of each edge road, the edge roads are stitched together across map sheets to obtain a global edge road stitched image; The global edge stitching image is subjected to endpoint connection and path merging to obtain the boundary road network.

[0013] Furthermore, to achieve the above objectives, this application also proposes a vector road network generation device, which includes: The acquisition module is used to acquire multiple road segmentation vectorized images within the target area; The recognition module is used to identify the center road and edge roads in any of the road segmentation vectorized images. The optimization module is used to perform cross-map sheet stitching and road optimization on the edge roads to obtain the boundary road network; The integration module is used to integrate and optimize each of the central roads and the boundary road network to obtain the target vector road network image corresponding to the target range.

[0014] In addition, to achieve the above objectives, this application also proposes a vector road network generation device, the device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the vector road network generation method as described above.

[0015] In addition, to achieve the above objectives, this application also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of the vector road network generation method described above.

[0016] In addition, to achieve the above objectives, this application also provides a computer program product, which includes a computer program that, when executed by a processor, implements the steps of the vector road network generation method described above.

[0017] This application provides a method, apparatus, device, and medium for generating vector road networks. The method acquires multiple road segmentation vectorized images of a target area, identifies the central and edge roads in any of these images, performs cross-map sheet stitching and road optimization on the edge roads to obtain a boundary road network, and then integrates and optimizes the central roads with the boundary road network to obtain the target vector road network image corresponding to the target area. This approach, through a "divide and conquer, inside-out" strategy, decomposes the complex global stitching problem into two stages: local processing and boundary processing. This reduces computational complexity, improves efficiency in processing large-scale areas, and achieves fully automated processing from image segmentation results to a high-quality vector road network, reducing manual intervention, lowering costs, and improving optimization efficiency. Furthermore, central roads can be processed in parallel across map sheets, fully utilizing computing resources and further improving processing speed. Attached Figure Description

[0018] 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.

[0019] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating an embodiment of the vector road network generation method of this application. Figure 2 Example scene diagram of the original remote sensing image of the target area provided for the vector road network generation method of this application; Figure 3 Example scene diagram of semantic segmentation image of remote sensing image provided for the vector road network generation method of this application; Figure 4 Example scene diagrams of road segmentation vectorized images provided for the vector road network generation method of this application; Figure 5 Example diagrams of scenarios for achieving geometric smoothing are provided for the vector road network generation method of this application; Figure 6 Example diagrams of road centerline extraction provided for the vector road network generation method of this application; Figure 7 One of the example diagrams illustrating the scenario of clearing invalid line segments provided for the vector road network generation method of this application; Figure 8Example diagram of a scenario for connecting "dead ends" provided for the vector road network generation method of this application; Figure 9 The second example diagram illustrating the scenario of clearing invalid line segments provided for the vector road network generation method of this application; Figure 10 Example diagrams of path extension processing provided for the vector road network generation method of this application; Figure 11 Example diagrams of scenarios with cracks between different map sheets provided for the vector road network generation method of this application; Figure 12 Example diagrams of scenarios involving cross-map boundary road network connections provided for the vector road network generation method of this application; Figure 13 A simplified flowchart illustrating the vector road network generation method of this application; Figure 14 Example scene diagram of the central road provided for the vector road network generation method of this application; Figure 15 Example scene diagrams of edge roads provided for the vector road network generation method of this application; Figure 16 This is a schematic diagram of the module structure of the vector road network generation device according to an embodiment of this application; Figure 17 This is a schematic diagram of the hardware operating environment involved in the vector road network generation method in this application embodiment.

[0021] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0022] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0023] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific embodiments.

[0024] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or an electronic device, big data service platform, or vector road network generation system capable of realizing the above functions. The following description uses a vector road network generation system as an example to illustrate this embodiment and the subsequent embodiments.

[0025] Based on this, embodiments of this application provide a vector road network generation method, referring to... Figure 1 , Figure 1This is a flowchart illustrating an embodiment of the vector road network generation method of this application.

[0026] In this embodiment, the vector road network generation method includes steps S11 to S14: Step S11: Obtain multiple road segmentation vectorized images of the target area; It should be noted that the target area refers to the geographic region where road network extraction and optimization are needed, such as a village, a town, or a larger administrative division. Road segmentation vectorized images are vector data images representing the location and shape of roads, obtained through remote sensing image segmentation technology. Stored in vector format, they contain geometric information about roads, such as start and end points and path shape. Vector format images offer higher precision and editability. For example, SHP files refer to geospatial data files stored in Shapefile format. Shapefile is a widely used geospatial data format for storing the geometric and attribute information of geographic objects. A Shapefile typically consists of multiple files, including a main file (.shp), an index file (.shx), and an attribute table file (.dbf). These files together constitute a complete geospatial dataset that can be read and processed by various GIS software and tools.

[0027] Specifically, this can be achieved in various ways. For example, satellite remote sensing images, images taken by drones, or images collected by ground vehicles can be used as the original images. (See reference...) Figure 2 Then, these original images undergo preprocessing, such as noise reduction and contrast enhancement, before a deep learning model is used to perform semantic segmentation on the remote sensing images. (See reference...) Figure 3 Then, the segmentation results are vectorized to obtain a road segmentation vectorized image, which can be referenced. Figure 4 The vectorization process can employ techniques such as edge detection and contour extraction to convert the road's geometry into vector data for subsequent processing.

[0028] Furthermore, the acquired road segmentation vectorized images may come from different data sources (such as satellite remote sensing images, drone images, or ground vehicle images), which may have different resolutions and accuracies. Therefore, after acquiring the images, it is also necessary to standardize them to ensure consistency in resolution, coordinate system, etc., for subsequent stitching and optimization operations.

[0029] For example, assuming the target area is a natural village, high-resolution images of the village are obtained through satellite remote sensing imagery. A deep learning model is used for semantic segmentation to identify road areas, and these areas are vectorized to obtain multiple road segmentation vectorized images.

[0030] Step S12: Identify the center road and edge roads of any of the road segmentation vectorized images; It should be noted that central roads refer to roads located within the map sheet that do not need to be connected across map sheets, while edge roads refer to roads located at the edge of the map sheet that may need to be connected across map sheets. This divides the road network into two parts for separate processing, improving processing efficiency and quality.

[0031] Specifically, for any of the road segmentation vectorized images, geometric smoothing and centerline extraction are performed on the road segmentation vectorized images to obtain a first optimized image. Then, based on the first optimized image, new path segments are constructed and integrated for optimization to obtain a second optimized image. By weighting the global path segments in the second optimized image and merging the main path segments in the second optimized image, a third optimized image is obtained. Then, based on the third optimized image, the central road and edge roads of the road segmentation vectorized image are identified.

[0032] For example, for each road segmentation vectorized image, Gaussian filtering is first used to smooth the boundaries of the road polygons, then buffering operations are used to repair topological errors inside the polygons, and finally the centerline of the road is extracted to clearly identify the central road and edge roads, providing accurate data for subsequent processing steps.

[0033] Step S13: Perform cross-map splicing and road optimization on each of the edge roads to obtain the boundary road network; It should be noted that cross-map sheet splicing refers to connecting edge roads in different map sheets to form a continuous road network. Road optimization refers to topological and geometric optimization of the spliced ​​roads to ensure the connectivity and accuracy of the road network. The boundary road network refers to the spliced ​​and optimized edge road network, which covers the boundary areas of each map sheet within the entire target area. The purpose of cross-map sheet splicing and road optimization is to form a seamless boundary road network and solve the problems of road breaks and misalignments at map sheet joints.

[0034] Specifically, based on the location coordinates of each edge road, the edge roads are stitched together across map sheets to obtain a global edge road stitched image. Then, the endpoints of the global edge road stitched image are connected and paths are merged to obtain the boundary road network.

[0035] For example, suppose the target area is a town, containing road segmentation vectorized images from multiple map sheets. First, based on the location coordinates of each edge road, these edge roads are stitched together across map sheets to obtain a global edge road stitched image. Then, endpoint connections and path merging are performed on the global edge road stitched image to obtain the boundary road network. This ensures the connectivity and accuracy of the boundary road network, providing high-quality data for subsequent integration and optimization.

[0036] Step S14: Integrate and optimize each of the central roads and the boundary road network to obtain the target vector road network image corresponding to the target range.

[0037] It should be noted that integration optimization refers to merging the central roads and boundary road networks to form a complete and seamless vector road network. The geometric and attribute information in the SHP file provides the data foundation for integration optimization. By reading the geometric data in the SHP file, the location and shape of each road can be determined, allowing for merging and optimization. The target vector road network image refers to the final road network image covering the entire target area, obtained by transforming the vector road network formed by merging the central roads and boundary road networks. Both the target vector road network image and the vector road network possess complete topological structure and geometric information, which can be used for practical applications such as navigation and analysis.

[0038] Specifically, a high-quality vector road network image is generated, ensuring the connectivity and consistency of the road network throughout the target area. In one possible implementation, spatial merging and attribute update operations can be used to merge the central roads and boundary road networks to form a complete vector road network, and the result is saved as an SHP file. Since the "center-boundary" separation mechanism in the above steps fundamentally ensures that the two datasets of central roads and boundary roads are geographically complementary and non-overlapping, the merging requires only a simple data aggregation operation, without the need for complex spatial conflict detection and processing. After merging, a final vector road network result covering the entire study area and seamlessly stitched together is obtained.

[0039] For example, assuming the target area is a natural village, and the central road and boundary road network have already been obtained, firstly, the central road and boundary road network are spatially merged under a unified geographic coordinate system. Then, the attribute information of the merged dataset is updated to ensure that the attribute information of all roads is complete and consistent. Finally, a seamless vector road network image covering the entire target area is obtained, and the result is saved as an SHP file. Through these processes, a high-quality vector road network image can be generated, meeting the needs of practical applications.

[0040] Understandably, this invention proposes a novel paradigm for large-scale road network processing to solve the problem of road breaks and misalignments when splicing cross-map data. Its core lies in a "divide and conquer" approach: First, through a unique "boundary influence zone" and map connectivity analysis, the road network within each independent original data unit (map sheet) is intelligently divided into a "central road network" that does not require cross-regional connections and a "peripheral road network" located at the edge. Then, processing is carried out according to the principle of "inside first, then outside": First, internal: Perform parallel and independent topology refinement on the "central road network" of all map sheets, efficiently completing the internal structure optimization of most road networks.

[0041] After all internal processing is completed, the previously separated "boundary road networks" from all map sheets are gathered together as a completely new, single whole, and subjected to a global, high-precision topology reconstruction.

[0042] This method employs an innovative "center-boundary" separation strategy, limiting the complex global stitching problem to a smaller number of boundary roads, while a large number of central roads can be processed in parallel within each map sheet. This significantly reduces computational complexity, improves efficiency in processing large-scale areas, and fundamentally guarantees the topological correctness and geometric continuity of the final road network at the stitching points, achieving true seamlessness. Furthermore, this strategy possesses excellent parallel processing capabilities and scalability.

[0043] This embodiment acquires multiple road segmentation vectorized images of a target area, then identifies the central and edge roads in any of these images. The edge roads are then stitched across map sheets and optimized to obtain a boundary road network. Finally, the central roads are integrated and optimized with the boundary road network to obtain the target vector road network image corresponding to the target area. This "divide and conquer, inside-out" strategy decomposes the complex global stitching problem into two stages: local processing and boundary processing. This reduces computational complexity, improves efficiency in processing large-scale areas, and achieves fully automated processing from image segmentation results to a high-quality vector road network, reducing manual intervention, lowering costs, and improving optimization efficiency. Furthermore, central roads can be processed in parallel across map sheets, fully utilizing computing resources and further improving processing speed.

[0044] In one feasible implementation, identifying the central road and edge roads of any of the road segmentation vectorized images includes: Step S21: For any of the road segmentation vectorized images, perform geometric smoothing and centerline extraction processing on the road segmentation vectorized images to obtain a first optimized image; It should be noted that geometric smoothing refers to processing the boundaries of road polygons in the road segmentation vectorized image to eliminate jagged edges and high-frequency noise, making the road boundaries more regular. Centerline extraction refers to extracting the line segment representing the center position of the road from the smoothed road polygons. The road segmentation vectorized image is obtained through remote sensing image segmentation technology and is a vector data image representing the location and shape of the road, typically containing geometric and attribute information. The first optimized image is the image after geometric smoothing and centerline extraction processing, containing a regular road centerline.

[0045] Specifically, in one embodiment, firstly, a geometric smoothing algorithm is applied to the input road segmentation vectorized image to eliminate edge jaggedness and high-frequency noise. For example, a Gaussian filtering method is applied to smooth the boundaries of the road polygons, eliminating high-frequency noise and jaggedness caused by semantic segmentation, making its shape closer to that of a real road. That is, a positive buffer with a small radius is applied to the smoothed polygon, followed by a negative buffer with an equal radius (e.g., geom.buffer(1).buffer(-1)). See reference [link to relevant documentation]. Figure 5 This effectively eliminates tiny holes, self-intersections, and narrow "bottleneck" areas inside polygons, thus playing a role in topology repair and geometric simplification.

[0046] Furthermore, based on the geometrically smoothed polygons, a centerline extraction algorithm (such as pygeoops.centerline) is called to calculate the centerline of the polygons, converting the planar data into linear road centerline data, thus obtaining the first optimized image and completing the data conversion from "plane" to "line". (See reference...) Figure 6 This transforms the original, irregular planar patterns generated by semantic segmentation into linear geometries with regular shapes and correct topology.

[0047] Step S22: Based on the first optimized image, construct and integrate the newly added path segments to obtain the second optimized image; It should be noted that the construction of new path segments refers to identifying and adding road segments that might have been missed in the original image, based on the first optimized image. Integration optimization refers to merging and optimizing the newly added path segments and existing road centerlines to form a more complete and coherent road network. The second optimized image refers to further optimizing the image after geometric smoothing and centerline extraction, thereby repairing broken sections in the road network, filling in missing road connections, and making the road network more complete and coherent.

[0048] Specifically, a first road topology map corresponding to the first optimized image is constructed, and invalid line segments in the first road topology map are cleaned up to obtain a fourth optimized image. Then, a second road topology map corresponding to the fourth optimized image is constructed, and new path segments in the second road topology map are constructed through spatial indexing. The new path segments in the second road topology map are then integrated and optimized with the fourth optimized image to obtain the second optimized image.

[0049] Step S23: By weighting the global path segments of the second optimized image and merging the main path segments in the second optimized image, a third optimized image is obtained. It should be noted that the global path segment refers to the set of all road segments in the second optimized image. Weighting refers to assigning different weight values ​​to path segments based on their length or other attributes, so as to prioritize the retention of important path segments during the merging process. The trunk path segment refers to the segment that plays a major connecting role in the road network, typically having a longer length and higher connectivity. The third optimized image is the image after weighting the path segments and merging the trunk path segments, containing the simplified and optimized road network.

[0050] Specifically, the global path segments of the second optimized image are weighted to determine the segment weights corresponding to the global path segments of the second optimized image. Then, for any weighted path segment in the global path segments, the longest path corresponding to the weighted path segment is determined. Path merging is then performed based on each of the longest paths to determine the trunk path segments in the second optimized image, thus obtaining the third optimized image.

[0051] Step S24: Based on the third optimized image, identify the central road and edge roads of the road segmentation vectorized image.

[0052] Specifically, after cleaning up invalid line segments in the third optimized image, the newly added path line segments in the third optimized image are constructed and integrated through spatial indexing and path extension processing to obtain the fifth optimized image. Then, the boundary lines of the fifth optimized image are obtained, and the boundary influence area in the fifth optimized image is determined based on the preset distance and each boundary line. This determines the positional relationship between the global path line segments of the fifth optimized image and the boundary influence area. Based on each positional relationship, the central road and edge road of the road segmentation vectorized image are determined.

[0053] This embodiment performs geometric smoothing and centerline extraction on any of the road segmentation vectorized images to obtain a first optimized image. Then, based on the first optimized image, new path segments are constructed and integrated to obtain a second optimized image. Next, by weighting the global path segments in the second optimized image and merging the main path segments, a third optimized image is obtained. Based on the third optimized image, the central and edge roads of the road segmentation vectorized image are identified. Geometric smoothing eliminates jagged edges and high-frequency noise on the road polygon boundaries, making the road boundaries more regular and closer to the real road shape, thereby improving the accuracy of road recognition. Centerline extraction converts the planar road polygons into linear centerlines, simplifying the process. The geometric representation of roads is used to construct and integrate new path segments, repairing broken parts of the road network and filling missing road connections, making the road network more complete and coherent. At the same time, global path segments are weighted and main path segments are merged to remove redundant small segments, simplify the road network structure, highlight the main roads, and make the road network clearer and more hierarchical. Finally, based on the optimized image, the topological relationships between roads, such as connection points and intersections, are more accurately determined by identifying the central and edge roads. The complex road network processing is decomposed into multiple orderly steps, each with a clear goal and optimization focus, making the entire processing more organized and efficient. This method is applicable to road networks of various complexities, from simple small village roads to complex urban road networks, and can achieve high-quality results through step-by-step optimization.

[0054] In one feasible implementation, the step of constructing and integrating new path segments based on the first optimized image to obtain a second optimized image includes: Step S31: Construct a first road topology map corresponding to the first optimized image, and clean up invalid line segments in the first road topology map to obtain a fourth optimized image; It should be noted that the first road topology map refers to the road network topology map containing "node-edge" (or "line-line") elements constructed based on the first optimized image. Invalid line segments refer to line segments in the topology map that do not meet the preset conditions, such as isolated line segments or excessively short line segments. These line segments may be generated due to data noise or segmentation errors. The fourth optimized image refers to the road topology map obtained after cleaning up invalid line segments.

[0055] Specifically, each road centerline segment is abstracted as a "node". If two line segments have a geometric intersection in space (i.e., they have a common point), then an "edge" is established between the nodes representing these two line segments to construct the first road topology graph. After the graph is constructed, each line segment (node) has a key attribute - "degree", which indicates how many other line segments are directly connected to it.

[0056] Furthermore, to remove noise and redundant data and optimize the road network topology, invalid line segments are identified by analyzing their connectivity (Degree). (See reference...) Figure 7 For example, a line segment with a degree of 0 is an isolated line segment. If its length is less than a preset threshold, it can be considered an invalid line segment and removed to make the fourth optimized image have a clearer and more regular road network topology.

[0057] Step S32: Construct a second road topology map corresponding to the fourth optimized image, and construct new path segments in the second road topology map through spatial indexing to obtain the second optimized image.

[0058] It should be noted that the second road topology map refers to a more refined topology map containing "endpoint-segment" relationships, constructed based on the fourth optimized image. Spatial indexing is a technique for quickly querying spatial data. By building spatial indexes (such as R-Trees), line segments related to specific areas can be quickly located and processed. New path segments refer to line segments identified and constructed in the second road topology map to fill in gaps in the road network.

[0059] Specifically, based on the fourth optimized image, a more refined "endpoint-segment" topology graph is constructed, where the start and end points of each segment are abstracted as "nodes" in the graph; and the segment itself becomes an "edge" connecting its corresponding two endpoint nodes, i.e., the second road topology graph. Simultaneously, the system establishes an efficient spatial index (such as an R-Tree) for all endpoints (nodes) to instantly query neighboring points around any point. This allows for the rapid identification of potentially disconnected roads or isolated segments using the spatial index, and the construction of new path segments to fill these gaps, resulting in a more complete and coherent road network in the second road topology graph.

[0060] Furthermore, new path segments are constructed in the second road topology map through spatial indexing. In one embodiment, firstly, using the "endpoint-segment" graph, i.e., the second road topology map, all nodes with a degree of 1 are identified; these are the "dead ends" of the road network. Then, for each dead end, spatial indexing is used to quickly find the nearest other dead end within a set search radius (e.g., 50 meters) to connect the two "dead ends." (See reference...) Figure 8 .

[0061] Furthermore, before performing the connection, a new line segment is generated connecting the two dead ends. It is then checked whether this virtual line segment will cross any existing road in the second road topology graph from the "middle" section. If a crossing occurs, it means the connection would create an non-compliant, non-node intersection, and therefore the connection is deemed invalid and abandoned. After completing the path crossing detection for every two "dead ends," the generated valid new path segment is added as a new edge to the second road topology graph, resulting in the second optimized image.

[0062] Understandably, this invention implements the application of multiple graph models, that is, abandoning single geometric operations and innovatively using two graph models alternately to drive the repair process. Among them, the macroscopic "line-to-line connectivity graph" is used to quickly filter noise and evaluate overall connectivity; the microscopic "endpoint-to-segment topology graph" is used to perform precise operations on the road network's breakpoints and simulates human expert thinking, that is, it designs a set of repair sequences with clear priorities, sequentially executing "direct connection between neighboring breakpoints" and "extension of isolated breakpoints to the nearest road segment", and uses core mechanisms such as "virtual path crossing detection" to prevent the generation of erroneous intersections and ensure the rationality of the connection.

[0063] This embodiment constructs a first road topology map corresponding to the first optimized image and cleans up invalid line segments in the first road topology map to obtain a fourth optimized image. Then, it constructs a second road topology map corresponding to the fourth optimized image and uses spatial indexing to construct new path segments in the second road topology map to obtain the second optimized image. By constructing the first road topology map and cleaning up invalid line segments to remove noise and redundant data, the road network becomes clearer and more organized. At the same time, the construction and integration optimization of new path segments can fill in the broken parts in the road network, repair missing connections, and ensure the continuity of the road network. The construction of the topology map can accurately represent the connection relationship between roads. Therefore, by cleaning up invalid line segments and integrating and optimizing new path segments, the topology structure of the road network can be optimized, reducing topology errors and anomalies. The road network processed by this method has an accurate geometric shape and a clear topology structure, which can better meet the needs of subsequent applications such as navigation, route planning, and traffic flow analysis.

[0064] In one feasible implementation, the process of determining the invalid line segment includes: Step S41: Obtain the degree value corresponding to each line segment in the first road topology map; It's important to note that the degree value refers to the connectivity of each line segment in the topology graph, that is, the number of edges connected to that line segment. In a road network, the degree value of a line segment reflects how many other line segments it is directly connected to.

[0065] Specifically, by analyzing the degree value of each line segment, it can be determined whether the line segment is an isolated line segment or a suspended line segment. In one possible implementation, a graph theory algorithm can be used to traverse the topology graph and calculate the degree value of each line segment in the first road topology graph, without limitation.

[0066] Step S42: The line segments whose degree values ​​meet the preset isolated line segment degree values ​​are designated as invalid line segments; It should be noted that an isolated line segment refers to a line segment that has no connection points in the topology map, usually represented as a dead end or an isolated road segment. The preset isolated line segment degree value is a threshold used to identify isolated line segments, and is usually set to 0.

[0067] Specifically, to remove line segments that have no practical significance in the road network, line segments with a degree of 0 (i.e., the preset degree value of isolated line segments) can be marked as invalid line segments, meaning that they do not intersect with any other line segments. At the same time, if their length is within a preset length range (e.g., <5 meters), they are judged as invalid line segments, that is, meaningless noise, and removed, so as to clean up noise and redundant data in the road network and optimize the topology.

[0068] Step S43: Obtain the length of the line segment corresponding to the line segment whose degree value meets the preset hanging line segment degree value, and take the line segment whose length meets the preset pruning length as the invalid line segment.

[0069] It should be noted that dangling segments refer to short line segments with only one connection point in the topology map, typically represented as suspended road segments. The preset dangling segment degree value is a threshold used to identify dangling segments, usually set to 1. The preset pruning length is a length threshold used to determine whether a dangling segment is an invalid segment, usually set according to actual application requirements.

[0070] Specifically, to further clean up noise and redundant data in the road network, in one possible implementation, line segments with a degree of 1 and a length less than a preset pruning length can be marked as invalid segments and removed from the topology graph to eliminate dangling segments that have no practical significance in the road network and optimize the topology structure.

[0071] For example, for each line segment in the first road topology map, its degree value is checked. If the degree value is 1, it means that it has only one connection point, forming a "dead end". If its length is also less than a certain threshold (usually longer than that required for isolated line segments, such as 30 meters), it will also be regarded as a short branch that needs to be pruned and removed, that is, it will be removed from the topology map as an invalid line segment.

[0072] This embodiment obtains the degree values ​​corresponding to each line segment in the first road topology map, and then identifies line segments whose degree values ​​meet the preset isolated line segment degree values ​​as invalid line segments. It also obtains the line segment lengths corresponding to line segments whose degree values ​​meet the preset suspended line segment degree values, and identifies line segments whose lengths meet the preset pruning lengths as invalid line segments. By using preset degree and length thresholds, invalid line segments are quickly identified, noise and redundant data are removed, improving the quality and accuracy of road network data. Furthermore, by identifying and removing isolated and suspended line segments, the topology of the road network is optimized, reducing topology errors and anomalies, making the road network clearer and more coherent.

[0073] In one feasible implementation, identifying the central road and edge roads of the road segmentation vectorized image based on the third optimized image includes: Step S51: After cleaning up the invalid line segments in the third optimized image, construct and integrate the newly added path line segments in the third optimized image through spatial indexing and path extension processing to obtain the fifth optimized image; It should be noted that the third optimized image is the image after weighting path segments and merging main path segments, containing a simplified and optimized road network. Spatial indexing is a technique for quickly querying spatial data. By building a spatial index (such as an R-Tree), line segments related to a specific area can be quickly located and processed. Path extension processing refers to extending dead-end points to connect them to the nearest road segments, thus filling gaps in the road network.

[0074] Specifically, after cleaning up invalid line segments in the third optimized image, spatial indexing is used to quickly identify dead ends or isolated line segments that may need to be connected, in order to further clean up some isolated short line segments and dangling line segments. The operation method is the same as steps S41-S43, but the threshold for line segment length can be appropriately increased. For example, the length threshold for isolated line segments can be set to 150 meters, and the length threshold for dangling line segments can be set to 100 meters. (Refer to...) Figure 9 .

[0075] Furthermore, for those dead ends where no matching object could be found, the system initiates a path extension strategy to construct new path segments through path extension processing. The path extension strategy is as follows: First, the system calculates the vertical projection distance from the dead-end point to all other road segments (edges in the topology graph) within the search range (e.g., 100 meters), finds the target road segment with the shortest distance, and after determining the target road segment to be connected, checks the geometry of the target road segment itself. If there is an existing vertex of the road segment near the projection point (e.g., within 15 meters), the system will prioritize "capturing" the connecting line to this existing vertex, that is, connecting the dead-end point and the vertex by generating an extension line, so as to maintain the original geometry of the road network to the greatest extent and avoid generating unnecessary small line segments and new vertices.

[0076] Similarly, before generating extension lines, path crossing detection must be performed to ensure that the extension lines do not illegally cross other roads.

[0077] Furthermore, this strategy incorporates conflict resolution logic. For example, when dealing with separated roads, there might be situations where two parallel dead-end roads attempt to connect to the same transverse road. The system will evaluate both connection options and typically only execute the shorter one to avoid generating unnatural parallel connections. (See reference...) Figure 10 .

[0078] Furthermore, the main path identification and merging function is invoked again (see steps S61-S63, which will not be elaborated here) to merge these newly generated connecting segments with the roads they connect to, ensuring the logical continuity of the road network.

[0079] After completing the above steps, a local network with maximized topological connectivity within the current map area can be obtained, which is the fifth optimized image.

[0080] Step S52: Obtain the boundary lines of the fifth optimized image, and determine the boundary influence area in the fifth optimized image based on the preset distance and each boundary line; It should be noted that the boundary line refers to the map sheet boundary of the fifth optimized image, that is, the edge line of the image. The boundary influence area refers to the area within a certain preset distance around the boundary line, where road segments may need to be stitched across map sheets.

[0081] Specifically, after obtaining the boundary lines of the fifth optimized image, the boundary influence area is determined based on a preset distance. In one embodiment, the top, bottom, left, and right boundary lines in the fifth optimized image are obtained. Based on these four boundary lines, a buffer with a specified distance (e.g., buffer_distance=100 meters) is generated inside the polygon. Then, the four strip-shaped buffer polygons generated by the boundary lines are spatially merged (Union) to form a single, complete "boundary influence area" polygon that surrounds the edge of the entire data area. This identifies which road segments are located at the edge of the map sheet and need to be spliced ​​across the map sheet.

[0082] Step S53: Determine the positional relationship between the global path segment of the fifth optimized image and the boundary influence region; Step S54: Based on the aforementioned positional relationships, determine the central road and edge roads of the road segmentation vectorized image.

[0083] It should be noted that the global path segments in step S53 refer to the set of all road segments in the fifth optimized image. Positional relationship refers to the spatial relationship between the global path segments and the boundary influence area, such as whether they intersect or contain each other.

[0084] Specifically, to further refine and classify road segments, all global path segments (i.e., road centerlines) within the fifth optimized image are traversed, and their positional relationship with the "boundary influence area" polygon is determined through spatial relationship calculations, and they are then classified: Boundary roads: Roads that intersect with the "boundary influence zone" polygon, as well as other roads that intersect with it. Other roads are included to ensure that road segments have complete topological relationships when connecting road networks across map boundaries in the future. They do not actually participate in the actual road network connection calculation process.

[0085] Central roads: Roads other than boundary roads.

[0086] Understandably, after this step, the separated "central roads" have undergone a complete local road network topology reconstruction and refinement process (i.e., steps S31-S43 and S61-S63), and their topology is in its final form, so they can be directly archived. The "boundary roads," on the other hand, are passed to the next stage for cross-map splicing.

[0087] Therefore, the road network can be divided into two parts and processed separately. The central roads can be processed independently and in parallel within each map sheet, while the peripheral roads need to be spliced ​​across map sheets, thereby optimizing the processing flow and improving efficiency and quality.

[0088] This embodiment cleans up invalid line segments in the third optimized image, constructs and integrates newly added path segments in the third optimized image through spatial indexing and path extension processing to obtain a fifth optimized image. Then, it obtains the boundary lines of the fifth optimized image and, based on preset distances and the boundary lines, determines the boundary influence areas in the fifth optimized image. This determines the positional relationship between the global path segments of the fifth optimized image and the boundary influence areas. Based on these positional relationships, it determines the central and edge roads of the road segmentation vectorized image. By cleaning up invalid line segments and constructing new path segments, it fills in the broken parts of the road network, improving the integrity and coherence of the road network. This optimizes the topology of the road network, reduces topological errors and anomalies, and makes the road network clearer and more hierarchical. The road network processed by this method has accurate geometric shape and a clear topological structure, better meeting the needs of subsequent applications such as navigation, route planning, and traffic flow analysis.

[0089] In one feasible implementation, the step of weighting the global path segments of the second optimized image and merging the main path segments in the second optimized image to obtain the third optimized image includes: Step S61: Weight the global path segments of the second optimized image to determine the segment weights corresponding to the global path segments of the second optimized image; It should be noted that the global path segments in step S61 refer to the set of all road segments in the second optimized image. Weighting refers to assigning different weight values ​​to path segments based on certain attributes (such as length, connectivity, etc.) to prioritize important path segments in subsequent processing. Segment weight is a quantitative indicator used to represent the importance of a path segment in the road network.

[0090] Specifically, in one possible implementation, the length of the path segment can be used as the weight, that is, the weight of each edge (i.e., the road segment) is its actual geometric length, and longer segments usually represent main roads, so they are given higher weights.

[0091] For example, for each path segment in the second optimized image, its length is calculated, and the length value is used as a weight. Assuming that the length of line segment A is 100 meters and the length of line segment B is 50 meters, then the weight of line segment A is 100 and the weight of line segment B is 50.

[0092] Step S62: For any weighted path segment in the global path segments, determine the longest path corresponding to the weighted path segment; It should be noted that the longest path refers to the longest continuous path from one starting point to one ending point in a weighted path segment, in order to identify the main paths in the road network, which usually represent the main roads.

[0093] Specifically, for each weighted path segment in the global path segment, a graph theory algorithm (such as depth-first search or breadth-first search) is used to determine its corresponding longest path. In one possible implementation, an iterative depth-first traversal algorithm starts from any unprocessed segment and explores paths in both directions along that segment. When the exploration reaches an intersection (i.e., an endpoint with a degree greater than 2), it continues to extend forward, traversing all possible path combinations. Furthermore, among all possible paths explored from a starting point, the algorithm calculates the total length of each path (i.e., the sum of the lengths of all segments on the path) and finally determines the path with the longest total length, which is considered a main road in the current network.

[0094] Step S63: Based on the longest paths, perform path merging to determine the main path segments in the second optimized image, and obtain the third optimized image.

[0095] It should be noted that path merging refers to combining path segments with similar directions and connections into a single, longer path segment. Main path segments are those that play a primary connecting role in the road network, typically possessing greater length and higher connectivity. The third optimized image is the image after path merging and main path segment determination, containing the optimized road network.

[0096] Specifically, to simplify the road network structure and highlight the main roads, in one embodiment, once a longest path is determined, the geometric information of all the original line segments constituting that path is extracted and seamlessly merged into a new, single, longer linear geometry through a geometric operation (linemerge). Simultaneously, all original line segments involved in the merging are marked as "processed." Furthermore, the algorithm continuously selects the next unprocessed line segment in the network as a new starting point, repeating the longest path finding and merging process until all line segments in the network are incorporated into a merged long road (i.e., the main road line segment), resulting in the third optimized image, thus simplifying the road network structure and highlighting the main roads.

[0097] Understandably, after the road network topology was repaired, the innovative use of the longest path algorithm based on graph theory depth-first search merged the physically connected short road segments separated by intersections into a single, logical, and geographically consistent long road, greatly improving the semantic value and usability of the road network data.

[0098] This embodiment determines the segment weights of the global path segments in the second optimized image by weighting them. Then, for any weighted path segment in the global path segments, the longest path corresponding to the weighted path segment is determined. Based on the longest paths, path merging is performed to determine the trunk path segments in the second optimized image, resulting in the third optimized image. Through weighting and path merging, broken parts in the road network can be filled, making the road network more complete and coherent. By determining the longest path and merging paths, the topology of the road network is optimized, redundant segments are reduced, and major roads are highlighted, thereby more accurately identifying trunk paths and providing reliable basic data for subsequent road network analysis and applications.

[0099] In one feasible implementation, the step of performing cross-map sheet stitching and road optimization on the edge roads to obtain the boundary road network includes: Step S71: Based on the position coordinates of each edge road, stitch the edge roads across the map frame to obtain a global edge road stitched image; It should be noted that edge roads refer to roads located at the edge of a map sheet that may need to be connected across map sheets. Location coordinates refer to the geographic coordinates of the start and end points of each edge road; these coordinates are used to determine the road's spatial location. Cross-map sheet stitching refers to connecting edge roads in different map sheets to form a continuous road network. The global edge road stitched image is the image obtained after cross-map sheet stitching, containing all edge roads.

[0100] Specifically, to form a seamless boundary road network, one possible implementation involves using spatial analysis tools (such as spatial query and data merging functions in GIS software) to identify and connect edge roads in different map sheets. Then, by comparing the endpoint coordinates of these edge roads, connectable edge roads in adjacent map sheets can be found and merged across map sheets, combining them into a unified GeoDataFrame—a global edge road mosaic image—within the same geographic coordinate system. At this point, the road network is broken at the seams of the global edge road mosaic image, as can be seen from [reference needed]. Figure 11 .

[0101] In one specific implementation, assume the following edge roads and their location coordinates: Edge road 1 in map A: starting point (10, 20), ending point (30, 40); Edge road 2 in map sheet B: starting point (30, 40), ending point (50, 60); Edge road 3 in map sheet C: starting point (50, 60), ending point (70, 80); By comparing the endpoint coordinates of these edge roads, it can be found that the endpoint (30, 40) of edge road 1 is the same as the starting point (30, 40) of edge road 2, and the endpoint (50, 60) of edge road 2 is the same as the starting point (50, 60) of edge road 3. Therefore, these edge roads can be stitched together across map sheets to obtain a global edge road stitched image.

[0102] Step S72: Connect endpoints and merge paths in the global edge stitching image to obtain the boundary road network.

[0103] It should be noted that endpoint connection refers to connecting disconnected points in the global edge stitching image to form a continuous road path. Path merging refers to merging path segments with similar directions and connections into a longer path segment. The boundary road network refers to the road network covering the entire boundary of the study area, obtained after endpoint connection and path merging.

[0104] Specifically, to further optimize the topology of the boundary road network and ensure its coherence and integrity, in one possible implementation, the same iterative connection logic (i.e., strategies for constructing a topology graph, interconnecting endpoints, merging paths, and connecting endpoints to line segments) from the above steps is reapplied to process the global edge stitching image. However, in this case, the algorithm's connection target is no longer breakpoints within the same map sheet, but rather geographically adjacent suspended endpoints that originally belonged to different map sheets. For example, it will connect an endpoint on the edge of map sheet A to an endpoint or line segment on the edge of map sheet B, as shown in the reference. Figure 12 This enables true, seamless cross-map splicing.

[0105] This embodiment stitches together edge roads across map sheets based on their location coordinates to obtain a global edge road stitched image. Then, it performs endpoint connection and path merging on this global edge road stitched image to obtain the boundary road network. This cross-map sheet stitching connects edge roads in different map sheets, forming a seamless boundary road network. This solves the problem of road breaks and misalignments at map sheet seams, ensuring the continuity and integrity of the boundary road network. This makes the road network of the entire study area more complete. Furthermore, through precise coordinate matching and path optimization, it improves the data quality of the boundary road network, reducing errors and anomalies. Thus, the seamless boundary road network provides more accurate basic data for applications such as navigation, route planning, and traffic flow analysis, enhancing the performance and reliability of these applications.

[0106] For example, to help understand the implementation process of the vector road network generation method, please refer to... Figure 13 , Figure 13 A simplified flowchart illustrating the vector road network generation method of this application.

[0107] Specifically, the process shown in the figure is a road network extraction and optimization process designed to automatically generate and optimize road networks from remote sensing imagery.

[0108] The process begins with the road polygon vector segmentation results from n remote sensing images, then proceeds through four main steps: S1, S2, S3, and center-boundary road separation, finally integrating the global road network and ending the process. Among these steps: In step S1, the road polygon is first smoothed by Gaussian filtering to eliminate noise and jagged edges. Then, the center line of the road polygon is extracted to lay the foundation for subsequent topology and road recognition.

[0109] Step S2 involves constructing a node-edge topology graph. This step involves analyzing the connectivity of each road centerline to build a graph model representing the road network connectivity. Subsequently, short line segments with a degree of 0 or 1 in the topology graph are cleaned up. These are usually isolated or suspended road sections that have little impact on the connectivity of the road network, so they are removed.

[0110] Step S3 involves constructing an "endpoint-segment" topology map by identifying dead ends (road endpoints) and attempting to connect these points to fill gaps in the road network. Afterward, trunk paths are identified and merged to form a more complete and coherent road network. Short segments with a degree of 0 or 1 are cleaned up again, and dead ends are extended to the nearest road segment to further optimize the road network structure.

[0111] In the center-boundary road separation step, the processed road network is divided into center roads and boundary roads, which can be referred to as follows: Figure 14 as well as Figure 15 Central roads are located within the map sheet and do not require cross-map sheet connections, while boundary roads are located at the map sheet edges and require cross-map sheet connections. By merging all boundary roads and repeating the processing flow of step S3, the coherence and integrity of the boundary road network are ensured.

[0112] Finally, all the collected central roads and the merged boundary roads are integrated into a global road network to form a complete road network. Through the separation of central and boundary roads and repeated optimization steps, a seamless and highly connected road network is generated.

[0113] It should be noted that the examples in the figure are only for understanding this application and do not constitute a limitation on the vector road network generation method of this application. Any simple transformations based on this technical concept are within the protection scope of this application.

[0114] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0115] This application also provides a vector road network generation device; please refer to [reference needed]. Figure 16 The vector road network generation device includes: The acquisition module 161 is used to acquire multiple road segmentation vectorized images of the target area; The recognition module 162 is used to identify the center road and edge roads of any of the road segmentation vectorized images; Optimization module 163 is used to perform cross-map splicing and road optimization on the edge roads to obtain the boundary road network; The integration module 164 is used to integrate and optimize each of the central roads and the boundary road network to obtain the target vector road network image corresponding to the target range.

[0116] The vector road network generation device is also used for: For any of the road segmentation vectorized images, geometric smoothing and centerline extraction are performed on the road segmentation vectorized images to obtain a first optimized image; Based on the first optimized image, new path segments are constructed and integrated for optimization to obtain a second optimized image; By weighting the global path segments of the second optimized image and merging the main path segments in the second optimized image, a third optimized image is obtained. Based on the third optimized image, the central road and edge roads of the road segmentation vectorized image are identified.

[0117] The vector road network generation device is also used for: Construct a first road topology map corresponding to the first optimized image, and clean up invalid line segments in the first road topology map to obtain a fourth optimized image; A second road topology map corresponding to the fourth optimized image is constructed, and new path segments are added to the second road topology map through spatial indexing to obtain the second optimized image.

[0118] The vector road network generation device is also used for: Obtain the degree value corresponding to each line segment in the first road topology map; Line segments whose degree values ​​conform to the preset isolated line segment degree values ​​are considered invalid line segments. Obtain the length of the line segment whose degree value meets the preset hanging line segment degree value, and take the line segment whose length meets the preset pruning length as the invalid line segment.

[0119] The vector road network generation device is also used for: After cleaning up the invalid line segments in the third optimized image, the newly added path line segments in the third optimized image are constructed and integrated through spatial indexing and path extension processing to obtain the fifth optimized image. Obtain the boundary lines of the fifth optimized image, and determine the boundary influence region in the fifth optimized image based on the preset distance and each boundary line; Determine the positional relationship between the global path segment of the fifth optimized image and the boundary influence region; Based on the aforementioned positional relationships, the central road and edge roads of the road segmentation vectorized image are determined.

[0120] The vector road network generation device is also used for: The global path segments of the second optimized image are weighted to determine the segment weights corresponding to the global path segments of the second optimized image; For any weighted path segment in the global path segments, determine the longest path corresponding to the weighted path segment; Based on merging the longest paths, the main path segments in the second optimized image are determined to obtain the third optimized image.

[0121] The vector road network generation device is also used for: Based on the location coordinates of each edge road, the edge roads are stitched together across map sheets to obtain a global edge road stitched image; The global edge stitching image is subjected to endpoint connection and path merging to obtain the boundary road network.

[0122] The vector road network generation apparatus provided in this application, employing the vector road network generation method in the above embodiments, can solve the technical problems mentioned in the background art. Compared with the prior art, the beneficial effects of the vector road network generation apparatus provided in this application are the same as those of the vector road network generation method provided in the above embodiments, and other technical features in the vector road network generation apparatus are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0123] This application provides a vector road network generation device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, which are executed by the at least one processor to enable the at least one processor to perform the vector road network generation method in the first embodiment described above.

[0124] The following is for reference. Figure 17The diagram illustrates a structural schematic of a vector road network generation device suitable for implementing embodiments of this application. The vector road network generation device in these embodiments may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 17 The vector road network generation device shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0125] like Figure 17 As shown, the vector road network generation device may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in the read-only memory 1002 or a program loaded from the storage device 1003 into the random access memory 1004. The random access memory 1004 also stores various programs and data required for the operation of the vector road network generation device. The processing unit 1001, the read-only memory 1002, and the random access memory 1004 are interconnected via a bus 1005. An input / output interface 1006 is also connected to the bus. Typically, the following systems can be connected to the input / output interface 1006: input devices 1007 including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices 1008 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 1003 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1009. The communication device 1009 allows the vector road network generating device to communicate wirelessly or wiredly with other devices to exchange data. Although the figure shows vector road network generating devices with various systems, it should be understood that it is not required to implement or possess all of the systems shown. More or fewer systems may be implemented alternatively.

[0126] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from read-only memory 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0127] The vector road network generation device provided in this application, employing the vector road network generation method in the above embodiments, can solve the technical problems mentioned in the background art. Compared with the prior art, the beneficial effects of the vector road network generation device provided in this application are the same as those of the vector road network generation method provided in the above embodiments, and other technical features in this vector road network generation device are the same as those disclosed in the method of the previous embodiment, and will not be repeated here.

[0128] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0129] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0130] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, the computer-readable program instructions being used to execute the vector road network generation method in the above embodiments.

[0131] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems or devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0132] The aforementioned computer-readable storage medium may be included in the vector road network generation device; or it may exist independently and not assembled into the vector road network generation device.

[0133] The aforementioned computer-readable storage medium carries one or more programs, which, when executed by the vector road network generation device, cause the vector road network generation device to: Obtain multiple road segmentation vectorized images within the target area; Identify the central road and edge roads in any of the road segmentation vectorized images; The edge roads are spliced ​​across map sheets and optimized to obtain the boundary road network. The central roads and the boundary road network are integrated and optimized to obtain the target vector road network image corresponding to the target range.

[0134] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a Local Area Network (LAN) or a Wide Area Network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0135] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0136] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0137] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described vector road network generation method, and is capable of solving the technical problems described in the background art. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as the beneficial effects of the vector road network generation method provided in the above embodiments, and will not be repeated here.

[0138] This application provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the vector road network generation method described above.

[0139] The computer program product provided in this application can solve the technical problems described in the background section. Compared with the prior art, the beneficial effects of the computer program product provided in the embodiments of this application are the same as the beneficial effects of the vector road network generation method provided in the above embodiments, and will not be repeated here.

[0140] The above description is only a part of the embodiments of this application and does not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A method for vector road network generation, characterized in that, The method comprises the following steps: acquiring a plurality of road segmentation vector images of a target range; identifying a center road and an edge road of any of the road segmentation vector images; splicing and optimizing each of the edge roads to obtain a boundary road network; integrating and optimizing each of the center roads and the boundary road network to obtain a target vector road network image corresponding to the target range.

2. The vector road network generation method of claim 1, wherein, The identification of the center road and the edge road of any of the road segmentation vector images comprises the following steps: performing geometric smoothing and center line extraction on any of the road segmentation vector images to obtain a first optimized image; constructing and optimizing new path segments based on the first optimized image to obtain a second optimized image; weighting the global path segments of the second optimized image and merging the main path segments in the second optimized image to obtain a third optimized image; identifying the center road and the edge road of the road segmentation vector image based on the third optimized image.

3. The vector road network generation method of claim 2, wherein, The construction and optimization of new path segments based on the first optimized image to obtain a second optimized image comprises the following steps: constructing a first road topology corresponding to the first optimized image and cleaning invalid segments in the first road topology to obtain a fourth optimized image; constructing a second road topology corresponding to the fourth optimized image and constructing new path segments in the second road topology through spatial indexing to obtain the second optimized image.

4. The vector road network generation method of claim 3, wherein, The determination process of the invalid segments comprises the following steps: acquiring a degree value corresponding to each segment in the first road topology; regarding a segment with a degree value meeting a preset isolated segment degree value as an invalid segment; acquiring a segment length corresponding to a segment with a degree value meeting a preset hanging segment degree value and regarding a segment with a segment length meeting a preset pruning length as an invalid segment.

5. The vector road network generation method of claim 2, wherein, The identification of the center road and the edge road of the road segmentation vector image based on the third optimized image comprises the following steps: after cleaning invalid segments in the third optimized image, constructing and integrating new path segments in the third optimized image through spatial indexing and path extension to obtain a fifth optimized image; acquiring boundary lines of the fifth optimized image and determining a boundary influence area in the fifth optimized image based on a preset distance and each of the boundary lines; determining a positional relationship between global path segments of the fifth optimized image and the boundary influence area; based on each of the positional relationships, determining the center road and the edge road of the road segmentation vector image.

6. The vector road network generation method of claim 2, wherein, The weighting of the global path segments of the second optimized image and the merging of the main path segments in the second optimized image to obtain a third optimized image comprises the following steps: weighting the global path segments of the second optimized image to determine a segment weight corresponding to the global path segments of the second optimized image; for any weighted path segment in the global path segments, determining a longest path corresponding to the weighted path segment; The main path segments in the second optimization image are determined based on path merging of the longest paths, to obtain the third optimization image.

7. The vector road network generation method of claim 1, wherein, The edge road splicing and road optimization are performed on each of the edge roads to obtain the boundary road network. The edge road splicing is performed on each of the edge roads according to the position coordinates of the edge roads to obtain a global edge road splicing image. The edge road splicing and road optimization are performed on each of the edge roads to obtain the boundary road network.

8. A vector road network generation apparatus characterized by comprising: The edge road splicing and road optimization are performed on each of the edge roads to obtain the boundary road network. The edge road splicing and road optimization are performed on each of the edge roads to obtain the boundary road network. The integration module is configured to integrate and optimize each of the center roads and the boundary road network to obtain a target vector road network image corresponding to the target range. The vector road network generation device comprises a memory, a processor, and a computer program stored on the memory and executable on the processor, and the computer program is configured to implement the steps of the vector road network generation method according to any one of claims 1 to 7. The storage medium is a computer readable storage medium, and the storage medium stores a computer program, and the computer program is executed by the processor to implement the steps of the vector road network generation method according to any one of claims 1 to 7.

9. A vector road network generation device, characterized by, ​ 10. A storage medium, characterized by ​

Images