A method for fusing electronic maps

Abstract

This invention discloses a method for fusing electronic maps, relating to the field of geographic information and digital map data processing technology. By continuously analyzing the line segment structure at the sampling node level, introducing structural relationships between adjacent nodes at each sampling node, and further converting and calculating local curvature, the bending changes of line segments in space can be expressed in a continuous sequence. This method can more meticulously reflect the bending changes of road crossing structures or waterway segments in local areas, enabling a more accurate description of the geometrical variations of the same geographic entity from different data sources. Through the joint construction of curvature and direction change sequences, the geometric structural features of road and waterway segments in local areas can be fully described, providing a stable structural basis for subsequent cross-layer entity recognition processes and helping to reduce the identification bias of entities with the same name due to differences in geometric representation.

Description

Technical Field

[0001] This invention relates to the field of geographic information and digital map data processing technology, specifically to a method for integrating electronic maps. Background Technology

[0002] With the increasing demand for the combined application of highway traffic data and inland waterway data in electronic maps, the spatial relationship representation between highway vector data and waterway vector data has gradually become an important aspect of electronic map construction. In practical applications, geographic objects such as bridges, waterway intersections, and road crossing structures often exist simultaneously in both highway and waterway networks. These cross-network structures are often expressed in different geometric forms in different data sources. For example, bridges appear as continuous road structures in highway data, while they appear as waterway restrictions or navigable structures in waterway data. Therefore, it is necessary to identify and associate entities with the same name in highway and waterway vector data using electronic map fusion methods.

[0003] In existing technologies, the identification of entities with the same name in highway and waterway data typically relies on name field matching or simple spatial proximity judgment methods, such as associating entities based on bridge names, node names, or spatial distance thresholds. However, in practical data applications, differences in naming rules, collection times, and data representation standards among different data sources can lead to situations where the same bridge or spanning structure may have missing names, name differences, or inconsistent naming rules in different datasets.

[0004] When entities with the same name in highway and waterway data cannot be accurately identified, issues such as entity duplication, structural mismatch, or missing spatial relationships can easily arise during electronic map fusion. For example, in areas where bridges cross waterways, if bridge nodes in the highway network and navigation restriction structures in the waterway network are not correctly associated, it may result in situations where roads cross waterways but the waterway does not record bridge information, or inconsistencies between the location of waterway restriction structures and roads. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method for fusing electronic maps, which solves the problems mentioned in the background section.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for fusing electronic maps, comprising the following steps: S1. Read the highway vector dataset and the waterway vector dataset, extract the spatial candidate units that form entities with the same name from the two types of data, and construct a sampling area with a radius of Rs at the center of each spatial candidate unit. Perform discrete sampling on the line segments in the sampling area according to a fixed sampling interval Ls to obtain the sampling node P and the node spacing sequence Ds={Di}. S2. Based on the sampling node P, analyze the three adjacent sampling nodes, calculate the discrete curvature K, and calculate the line segment direction angle xphe and direction change Δphe according to the two adjacent sampling nodes to form a curvature sequence and a direction change sequence. S3. Based on the curvature sequence, calculate the curvature change Vc. According to the direction change sequence and the node spacing sequence Ds={Di}, calculate the direction change structure Sd. At the same time, combine it with the node density ρ of the sampled nodes to form a set of geometric structure feature quantities of spatial candidate units. S4. Calculate the geometric fingerprint value Fg of the spatial candidate unit based on the curvature change Vc, the direction change structure Sd, the node density ρ, and the total length of the line segment Lt in the sampling area. S5. Establish candidate spatial candidate unit pairs between the highway vector dataset and the waterway vector dataset. Based on the corresponding geometric fingerprint value Fg, calculate the entity difference Df and compare the entity difference Df with the preset judgment value Td to determine the entity status.

[0007] Preferably, in step S11, the highway vector dataset and the waterway vector dataset are read, and the bridge nodes, road crossing nodes and waterway centerline intersection nodes in the highway vector dataset are extracted respectively, and the proximity association retrieval is performed based on the spatial position relationship of each node. The nodes that meet the preset proximity constraints are merged to generate a set of spatial candidate units; then, a sampling region with a radius of Rs is constructed with the center position (X, Y) of each spatial candidate unit as the center. The proximity constraint restricts the merging range of candidate nodes. It calculates the spatial proximity between any two candidate nodes to obtain the spatial interval Lab, using the following formula: ; Lab specifically represents the spatial interval between candidate node a and candidate node b, where (xa, ya) and (xb, yb) represent the coordinates of candidate node a and candidate node b, respectively. When the spatial interval Lab between candidate node a and candidate node b is less than the distance threshold, they are merged into the same spatial candidate unit. The center position (X, Y) is obtained by acquiring the coordinate values ​​xj and yj of each data point and calculating the average values ​​of the horizontal and vertical coordinates respectively. Specifically, the average value of the horizontal coordinate X is the sum of all horizontal coordinate values ​​xj divided by the number of data points Nq. Similarly, the average value of the vertical coordinate Y is the sum of all vertical coordinate values ​​yj divided by the number of data points Nq. The obtained (X, Y) represents the center position coordinates.

[0008] Preferably, in step S12, the road segments and waterway segments falling within the sampling area are extracted, and each segment is sampled at equal distances according to a fixed sampling interval Ls to form a set of sampling nodes P. Based on the coordinate order of the sampling nodes, the local curvature C, direction angle phe and node density ρ corresponding to each sampling node are calculated in sequence, and the node spacing D between adjacent sampling nodes is calculated to generate a node spacing sequence Ds={Di}. The local curvature C is obtained as follows: First, select three adjacent sampling nodes, namely Pi-1, Pi, and Pi+1, with coordinates (xi-1, yi-1), (xi, yi), and (xi+1, yi+1). First, calculate the product of two vectors: one is the product of the difference between the x and y coordinates from point Pi-1 to point Pi and the difference between the x and y coordinates from point Pi to point Pi+1; then subtract another similar product; and place the result in the numerator. The denominator calculates the lengths of two vectors, which are the sum of the squares of the lengths from point Pi-1 to point Pi and from point Pi to point Pi+1. The entire result is then cubed and a minimum value of 0.01 is added to obtain the local curvature C. The direction angle phe is obtained by calculating the ratio of the difference between the ordinates and the difference between the abscissas of two adjacent points to obtain the slope of the line segment; then, substituting this ratio into the arctangent function, the direction angle phe of the line segment is obtained. The node density ρ is obtained by dividing the total number of sampling nodes in the sampling area by the product of the square of the sampling area radius and the value of π. The formula for obtaining the node spacing D is: ; In the formula, Di represents the node spacing of sampling node Pi, and (xi, yi) and (xi+1, yi+1) represent the coordinates of sampling nodes Pi and Pi+1, respectively.

[0009] Preferably, in step S21, for each sampling node, select the preceding and following sampling nodes adjacent to the sampling node, first obtain the local curvature C of the sampling node, and then calculate the discrete curvature K based on the local curvature C: The formula for obtaining the discrete curvature K is as follows: ; In the formula, Ci represents the local curvature of sampling node Pi, Di+1 represents the node spacing of sampling node Pi+1, Di-1 represents the node spacing of sampling node Pi-1, Li represents the node spacing from sampling node Pi to its adjacent sampling points, and Ki represents the discrete curvature of sampling node Pi. Integrate all discrete curvatures K to obtain a curvature sequence.

[0010] Node spacing and path length both affect the calculation of discrete curvature. Changes in node spacing (i.e., variations in the interval between sampling points) are used to measure the degree of change in local curvature. Compared to traditional curvature calculation methods based on fixed spacing, considering both node spacing variations and path length allows for a more detailed capture of the path's bending and changes in different regions. Discrete curvature can more accurately reflect the variation characteristics of the path at each sampling point. Especially in sampling nodes with different precision, by considering the combined effects of node spacing and local curvature, a more reasonable curvature value can be obtained, and it can adapt to path data with different shapes (such as complex paths with alternating curved and straight segments).

[0011] Preferably, in step S22, the line segment direction angle xphe is calculated based on the direction angle phe of two adjacent sampling nodes; The line segment direction angle xphe is obtained as follows: read the original direction angle phe corresponding to the previous node, the current node, and the next node of the current sampling node, and calculate the absolute value of the difference between each direction angle phe; then combine each original direction angle with the corresponding absolute value of the difference to obtain the line segment direction angle xphe corresponding to the current sampling node. Based on the line segment direction angle xphe, the direction change Δphe is calculated and obtained, and the direction change Δphe of all sampling nodes is integrated to obtain the direction change sequence; The formula for obtaining the change in direction Δphe is: ; In the formula, Δphei represents the change in direction of sampling node Pi, xphei represents the line segment direction angle of sampling node Pi, and xphe(i-1) represents the line segment direction angle of sampling node Pi-1.

[0012] Preferably, in step S31, the adjacent difference expansion process is performed on the curvature sequence to obtain the change of the bending state of the line segment inside the spatial candidate unit along the sampling direction, and the curvature change amount Vc is obtained; then the direction change sequence and the node spacing sequence are combined and calculated to obtain the direction change structure amount Sd under the combined action of direction turning and spatial discretization relationship. The curvature change Vc is obtained as follows: First, read the discrete curvature values ​​corresponding to the current sampling node and its adjacent sampling nodes in the curvature sequence in sequence; then calculate the absolute difference between the current discrete curvature and the previous discrete curvature, and the absolute difference between the next discrete curvature and the current discrete curvature respectively. Add the absolute values ​​of the two differences to obtain the local curvature fluctuation at the current sampling node; read the distance between the two nodes adjacent to the sampling node and sum the distances of the two nodes as the local spatial expansion scale; finally, divide the local curvature fluctuation by the local spatial expansion scale and sum it point by point for all intermediate sampling nodes to obtain the curvature change Vc. The results indicate the degree of continuous change in curvature within the candidate solid element along the sampling direction: If the curvature change Vc value is large, it indicates that the bending state of the line segment within the candidate solid element fluctuates continuously in multiple local intervals; if the curvature change Vc value is small, it indicates that the bending state of the line segment within the candidate solid element is relatively stable. The structural quantity Sd representing the change in direction is obtained using the following formula: ; In the formula, Ns represents the number of sampling nodes.

[0013] When the directional change in a certain region is large, and there is a significant difference in the distance between the nodes before and after, the value of the directional change structure Sd will increase accordingly. When the direction change is continuous but the node distribution is relatively uniform, the structural quantity Sd of the direction change mainly reflects the continuous direction structure. When the direction change is small and the node spacing changes smoothly, the value of the structural quantity Sd of the direction change is relatively small. Preferably, in step S32, based on the curvature change Vc and the direction change structure Sd, and combined with the obtained node density ρ, a set of geometric structural feature quantities of spatial candidate units is constructed. Wherein, node density ρ represents the number of sampling nodes distributed within a unit sampling area.

[0014] If the node density ρ value is large, it indicates that the sampling nodes are densely distributed within a limited area; if the node density ρ value is small, it indicates that the sampling nodes are relatively sparsely distributed. Preferably, in step S41, the road segments and waterway segments falling into the sampling area are read, and the node spacing between each adjacent sampling node is extracted according to the sampling node order. Then, the total length Lt of the line segments in the sampling area corresponding to the current spatial candidate unit is obtained by performing cumulative calculation on all node spacings. If the total length Lt of the line segment is large, it means that the line segments contained in the current sampling area have a long unfolded length; if the total length Lt of the line segment is small, it means that the current sampling area contains only short local line segments. S42. After obtaining the total length Lt of the line segment, combine it with the curvature change Vc, the direction change structure Sd and the node density ρ, and construct the geometric fingerprint value Fg of the current spatial candidate unit through geometric coupling calculation. The geometric fingerprint value Fg is obtained using the following formula: .

[0015] The geometric fingerprint value Fg is not a single curvature quantity, a single direction quantity, or a single length quantity, but a structural characterization result that unifies the local bending change state, direction change state, node distribution state, and line segment unfolding state. in, Used to characterize the comprehensive change state of the current spatial candidate unit in local geometry. Used to characterize the relationship between the node distribution and line segment expansion of the current spatial candidate unit within the sampling area; When the Fg values ​​of two spatial candidate units are close, it indicates that they have similar structural expressions in terms of local geometric organization. When the Fg values ​​of two spatial candidate units differ significantly, it indicates that there is at least one significant difference between them in terms of bending change, direction change, node distribution, or line segment unfolding state. Preferably, in step S51, the set of spatial candidate units in the highway vector dataset and the set of spatial candidate units in the waterway vector dataset are read, and the geometric fingerprint value Fg corresponding to each spatial candidate unit obtained in step S4 is read. Based on the center location of the spatial candidate unit, the coverage area of ​​the sampling area, and the unit number to which the geometric fingerprint value belongs, a one-to-one pairing search is performed between the spatial candidate units on the highway side and the spatial candidate units on the waterway side to establish spatial candidate unit pairs. For each pair of spatial candidate units, the corresponding road-side geometric fingerprint value Fg(road) and waterway-side geometric fingerprint Fg(water) are read, and the entity difference Df is calculated. The entity difference Df is obtained by taking the absolute value of the difference between the road-side geometric fingerprint value Fg (road) and the waterway-side geometric fingerprint Fg (water). When the entity difference Df value is small, it indicates that the local geometric organization of the two spatial candidate units is similar. When the entity difference Df value is large, it indicates that there is a significant difference in the geometric structure between the two spatial candidate units. When obtaining the entity difference Df, a center interval LJ is introduced to constrain the space. The center interval LJ only performs geometric fingerprint difference calculation on spatial candidate units whose center interval is not greater than the preset pairing range value, thus avoiding the entry of distant irrelevant units into the judgment process.

[0016] The method for obtaining the center gap LJ is as follows: First, read the center coordinates of the candidate spatial units on the highway side and the candidate spatial units on the waterway side; then calculate the difference between the two center coordinates in the horizontal direction, that is, the difference between the horizontal coordinates of the highway side center and the horizontal coordinates of the waterway side center, and square the difference. Next, calculate the difference between the two center coordinates in the vertical direction, that is, the difference between the vertical coordinate of the center on the highway side and the vertical coordinate of the center on the waterway side, and square the difference; then sum the squared result of the difference in the horizontal coordinate and the squared result of the difference in the vertical coordinate; finally, perform a square root operation on the summed result to obtain the center interval LJ between the candidate spatial units on the highway side and the candidate spatial units on the waterway side. Preferably, in step S52, the acquired entity difference quantity Df is compared with a preset judgment value Td to determine the entity state; the method is as follows: If the entity difference Df is less than or equal to the preset judgment value Td, then the spatial candidate unit pair is determined to be entities with the same name, and spatial fusion processing is performed. When the entity difference Df is greater than the preset judgment value Td, the candidate unit pair in the space is judged to be different entities, and the original data structure is maintained. The formula for obtaining the judgment value Td is: Td = μDf + kb × σDf; In the formula, μDf represents the average value of entity differences, indicating the overall average level of entity differences in the current set of candidate spatial unit pairs. This value is obtained by summing the entity differences Df corresponding to all spatial candidate unit pairs and then dividing by the number of spatial candidate unit pairs. It is used to reflect the basic level of overall geometric fingerprint differences. σDf represents the standard deviation of entity differences, used to characterize the dispersion of entity differences among spatial candidate unit pairs. A larger standard deviation indicates greater fluctuation in geometric differences among different spatial candidate unit pairs; a smaller standard deviation indicates that the distribution of entity differences is relatively concentrated. kb represents the difference adjustment coefficient, with a value ranging from 0.5 to 2.5.

[0017] This invention provides a method for fusing electronic maps, which has the following beneficial effects: (1) By constructing spatial candidate units between highway vector data and waterway vector data, and performing discrete sampling on local line segment structures at a unified spatial scale, structural analysis can be performed on data from different sources within the same spatial range. In this way, even when name information is missing or naming rules differ, areas that may correspond to the same geographic entity can still be identified based on spatial structural relationships, thereby reducing the identification bias caused by relying solely on name matching or simple spatial proximity judgment.

[0018] (2) By performing adjacent difference expansion processing on the curvature sequence, the bending changes of line segments in the sampling direction can be expressed in a continuous variation form, thereby more accurately reflecting the bending change state of road crossing structures or waterway segments in local areas. Compared with methods that rely solely on single-point curvature or a single geometric index, this method, through cumulative analysis of adjacent curvature changes, enables the overall spatial trend of line segment bending changes to be fully expressed, thereby improving the stability of identifying the geometric features of the same geographic entity in different data sources.

[0019] (3) By constructing geometric fingerprint values, the curvature change features, direction change features, node density, and line segment unfolding length are uniformly expressed, so that the local geometric organization of spatial candidate units can be transformed into comparable structural feature results. By performing difference analysis on the geometric fingerprints of spatial candidate units on the highway side and the waterway side, spatial units with similar structures can be identified, and the center interval is introduced to constrain the spatial range during the identification process, thereby reducing the entry of distant and irrelevant spatial units into the judgment process, making the cross-layer entity identification process more targeted.

[0020] (4) Spatial candidate unit pairs are established between the highway vector dataset and the waterway vector dataset. By reading the geometric fingerprint values ​​corresponding to each spatial candidate unit and comparing them one by one, local structures that may correspond to the same geographic entity in different data sources can be matched under a unified rule. Compared with traditional methods that rely on name fields or simple spatial proximity judgment, this method performs one-by-one pairing and retrieval of spatial candidate units, enabling objects such as bridge crossing structures and waterway intersection structures in highway data and waterway data to be associated at the structural level, thereby reducing the difficulty of entity identification caused by inconsistent names or missing attribute information. Attached Figure Description

[0021] Figure 1 This is a schematic diagram illustrating the steps of an electronic map fusion method according to the present invention; Figure 2 This is a schematic diagram illustrating the process of determining different entities in this invention; Figure 3 This is a schematic diagram of the process for obtaining the curvature change amount according to the present invention. Detailed Implementation

[0022] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0023] Example 1 This invention provides a method for merging electronic maps; please refer to [link / reference]. Figures 1 to 3 This includes the following steps: S1. Read the highway vector dataset and the waterway vector dataset, extract the spatial candidate units that form entities with the same name from the two types of data, and construct a sampling area with a radius of Rs at the center of each spatial candidate unit. Perform discrete sampling on the line segments in the sampling area according to a fixed sampling interval Ls to obtain the sampling node P and the node spacing sequence Ds={Di}. S2. Based on the sampling node P, analyze the three adjacent sampling nodes, calculate the discrete curvature K, and calculate the line segment direction angle xphe and direction change Δphe according to the two adjacent sampling nodes to form a curvature sequence and a direction change sequence. S3. Based on the curvature sequence, calculate the curvature change Vc. According to the direction change sequence and the node spacing sequence Ds={Di}, calculate the direction change structure Sd. At the same time, combine it with the node density ρ of the sampled nodes to form a set of geometric structure feature quantities of spatial candidate units. S4. Calculate the geometric fingerprint value Fg of the spatial candidate unit based on the curvature change Vc, the direction change structure Sd, the node density ρ, and the total length of the line segment Lt in the sampling area. S5. Establish candidate spatial candidate unit pairs between the highway vector dataset and the waterway vector dataset. Based on the corresponding geometric fingerprint value Fg, calculate the entity difference Df and compare the entity difference Df with the preset judgment value Td to determine the entity status.

[0024] In this embodiment, by constructing spatial candidate units and performing unified sampling processing on highway vector data and waterway vector data within a local range, the line segment structures in different data sources can be analyzed at the same spatial scale, thereby avoiding the identification errors caused by relying solely on name fields or simple spatial proximity relationships, and enabling cross-layer entity recognition to be based on actual geometric structures.

[0025] This invention constructs a geometric fingerprint value that can characterize local spatial structure features by jointly analyzing the curvature changes, direction changes, and node distribution status of line segments within the sampling area. This enables the structural features of the same geographic entity in different data sources to be uniformly expressed. Even in the case of missing names or inconsistent naming rules, entities with the same name can be identified through spatial structure features, thereby improving the unstable identification problem caused by relying on name matching in traditional methods.

[0026] This invention calculates the difference between the geometric fingerprint values ​​of highway data and waterway data, and combines this with a judgment mechanism to determine the entity status. This gives the matching process between spatial candidate units a clear structural basis, thereby reducing the occurrence of duplicate records or spatial mismatches of geographical entities such as bridge crossing structures and waterway intersection structures during the fusion process. This makes the spatial relationship expression between highway networks and waterway networks in electronic maps more consistent and improves the overall structural coordination of fused data in electronic maps.

[0027] Example 2 Please refer to Figure 1 Specifically: S11, Read the highway vector dataset and the waterway vector dataset, extract the bridge nodes and road crossing nodes in the highway vector dataset and the waterway centerline intersection nodes in the waterway vector dataset, and perform a proximity association retrieval based on the spatial position relationship of each node; The nodes that meet the preset proximity constraints are merged to generate a set of spatial candidate units; then, a sampling region with a radius of Rs is constructed with the center position (X, Y) of each spatial candidate unit as the center. The center position (X, Y) is obtained by acquiring the coordinate values ​​xj and yj of each data point and calculating the average values ​​of the horizontal and vertical coordinates respectively. Specifically, the average value of the horizontal coordinate X is the sum of all horizontal coordinate values ​​xj divided by the number of data points Nq. Similarly, the average value of the vertical coordinate Y is the sum of all vertical coordinate values ​​yj divided by the number of data points Nq. The obtained (X, Y) represents the center position coordinates.

[0028] S12. Extract the highway and waterway segments that fall within the sampling area. Perform equidistant sampling on each segment according to a fixed sampling interval Ls to form a set of sampling nodes P. Based on the coordinate order of the sampling nodes, calculate the local curvature C, direction angle phe, and node density ρ corresponding to each sampling node in sequence, and calculate the node spacing D between adjacent sampling nodes to generate the node spacing sequence Ds={Di}. The local curvature C is obtained as follows: First, select three adjacent sampling nodes, namely Pi-1, Pi, and Pi+1, with coordinates (xi-1, yi-1), (xi, yi), and (xi+1, yi+1). First, calculate the product of two vectors: one is the product of the difference between the x and y coordinates from point Pi-1 to point Pi and the difference between the x and y coordinates from point Pi to point Pi+1; then subtract another similar product; and place the result in the numerator. The denominator calculates the lengths of two vectors, which are the sum of the squares of the lengths from point Pi-1 to point Pi and from point Pi to point Pi+1. The entire result is then cubed and a minimum value of 0.01 is added to obtain the local curvature C. The direction angle phe is obtained by calculating the ratio of the difference between the ordinates and the difference between the abscissas of two adjacent points to obtain the slope of the line segment; then, substituting this ratio into the arctangent function, the direction angle phe of the line segment is obtained. The node density ρ is obtained by dividing the total number of sampling nodes in the sampling area by the product of the square of the sampling area radius and the value of π. The formula for obtaining the node spacing D is: ; In the formula, Di represents the node spacing of sampling node Pi, and (xi, yi) and (xi+1, yi+1) represent the coordinates of sampling nodes Pi and Pi+1, respectively.

[0029] In this embodiment, bridge nodes, road crossing nodes, and waterway centerline intersection nodes are extracted from highway vector data and waterway vector data. Proximity correlation analysis is then performed based on the spatial relationships between these nodes, thereby constructing unified spatial candidate units across different data sources. This allows local areas potentially corresponding to the same geographic entity to be identified and processed in advance. Compared to entity matching methods that rely solely on name fields or simple spatial distance, this approach can still determine the analysis scope of potential entities with the same name through spatial structural relationships, even in cases of inconsistent data naming or missing attribute information, providing a stable spatial foundation for subsequent fusion analysis.

[0030] This embodiment constructs a fixed-scale sampling region within the spatial candidate unit area and performs discrete sampling with uniform spacing on highway and waterway segments within the region. This allows vector segments from different sources and with different precisions to be analyzed structurally at a unified spatial scale. By extracting geometric features such as the local curvature, segment orientation angle, and node distribution density of the sampling nodes, the bending state, orientation changes, and spatial distribution of nodes can be transformed into comparable structural parameters. This reduces the impact of differences in acquisition scale and geometric representation between different data sources on the entity recognition process.

[0031] This embodiment continuously calculates the spatial spacing between sampling nodes and forms a node spacing sequence, thus fully expressing the spatial unfolding state of line segments within the sampling area. This allows for a more comprehensive reflection of the local geometric features of road crossing structures or waterway intersection structures. Through this processing, the local structural information in highway and waterway data can be uniformly organized and expressed, providing a reliable structural foundation for subsequent geometric feature-based entity recognition and data fusion. This helps reduce entity mismatches or duplicate records in bridge crossing areas or waterway intersection areas during electronic map fusion.

[0032] Example 3 Please refer to Figure 1 Specifically: S2 includes S21 and S22; S21. For each sampling node, select the preceding and following sampling nodes adjacent to the sampling node, first obtain the local curvature C of the sampling node, and then calculate the discrete curvature K based on the local curvature C: The formula for obtaining the discrete curvature K is as follows: ; In the formula, Ci represents the local curvature of sampling node Pi, Di+1 represents the node spacing of sampling node Pi+1, Di-1 represents the node spacing of sampling node Pi-1, Li represents the node spacing from sampling node Pi to its adjacent sampling points, and Ki represents the discrete curvature of sampling node Pi. Integrate all discrete curvatures K to obtain a curvature sequence.

[0033] S22. Calculate and obtain the line segment direction angle xphe based on the direction angle phe of two adjacent sampling nodes; The line segment direction angle xphe is obtained as follows: read the original direction angle phe corresponding to the previous node, the current node, and the next node of the current sampling node, and calculate the absolute value of the difference between each direction angle phe; then combine each original direction angle with the corresponding absolute value of the difference to obtain the line segment direction angle xphe corresponding to the current sampling node. Based on the line segment direction angle xphe, the direction change Δphe is calculated and obtained, and the direction change Δphe of all sampling nodes is integrated to obtain the direction change sequence; The formula for obtaining the change in direction Δphe is: ; In the formula, Δphei represents the change in direction of sampling node Pi, xphei represents the line segment direction angle of sampling node Pi, and xphe(i-1) represents the line segment direction angle of sampling node Pi-1.

[0034] In this embodiment, by continuously analyzing the line segment structure at the sampling node level, the structural relationship between adjacent nodes is introduced at each sampling node, and the local curvature is further transformed and calculated to obtain discrete curvature. This allows the bending changes of the line segment in space to be expressed in the form of a continuous sequence. Compared with the traditional method that only uses single-point curvature or overall curvature indices, this method can more meticulously reflect the bending change state of road crossing structures or waterway segments in local areas, and can more accurately describe the geometric morphological changes of the same geographic entity in different data sources.

[0035] In this embodiment, the original direction angle is not used directly during the direction feature calculation. Instead, the original direction angle is reconstructed by combining the direction relationship between adjacent nodes to obtain the line segment direction angle. In this way, the impact of abrupt changes in the direction of a single node or local sampling errors on the overall direction judgment can be reduced, so that the line segment direction feature can more stably reflect the true direction of the line segment in the local area.

[0036] This embodiment, after obtaining the direction angle of a line segment, calculates the change relationship between the direction angles of adjacent line segments to construct a direction change sequence, enabling the direction turning of the line segment in space to be expressed in a continuous change form. Through the joint construction of the curvature sequence and the direction change sequence, the geometric structural features of road and waterway line segments in local areas can be fully described, thus providing a stable structural basis for subsequent cross-layer entity recognition processes and helping to reduce the identification deviation of entities with the same name due to differences in geometric representation.

[0037] Example 4 Please refer to Figure 3 Specifically: S3 includes S31; S31. Perform adjacent difference expansion processing on the curvature sequence to obtain the change of the bending state of the line segment inside the spatial candidate unit along the sampling direction, and obtain the curvature change amount Vc; then combine the direction change sequence and the node spacing sequence to calculate the direction change structure amount Sd under the combined action of direction turning and spatial discretization relationship. The curvature change Vc is obtained as follows: First, read the discrete curvature values ​​corresponding to the current sampling node and its adjacent sampling nodes in the curvature sequence in sequence; then calculate the absolute difference between the current discrete curvature and the previous discrete curvature, and the absolute difference between the next discrete curvature and the current discrete curvature respectively. Add the absolute values ​​of the two differences to obtain the local curvature fluctuation at the current sampling node; read the distance between the two nodes adjacent to the sampling node and sum the distances of the two nodes as the local spatial expansion scale; finally, divide the local curvature fluctuation by the local spatial expansion scale and sum it point by point for all intermediate sampling nodes to obtain the curvature change Vc. The structural quantity Sd representing the change in direction is obtained using the following formula: ; In the formula, Ns represents the number of sampling nodes.

[0038] S32. Based on the curvature change Vc and the direction change structure Sd, and combined with the obtained node density ρ, a set of geometric structural feature quantities of spatial candidate units is constructed. Wherein, node density ρ represents the number of sampling nodes distributed within a unit sampling area.

[0039] S41. Read the highway and waterway segments that fall into the sampling area, extract the node spacing between each adjacent sampling node in the order of sampling nodes, and then perform cumulative calculation on all node spacings to obtain the total length Lt of the line segments in the sampling area corresponding to the current spatial candidate unit. S42. After obtaining the total length Lt of the line segment, combine it with the curvature change Vc, the direction change structure Sd and the node density ρ, and construct the geometric fingerprint value Fg of the current spatial candidate unit through geometric coupling calculation. The geometric fingerprint value Fg is obtained using the following formula: .

[0040] In this embodiment, by performing adjacent difference expansion processing on the curvature sequence, the bending changes of line segments in the sampling direction can be expressed in a continuous variation form, thereby more accurately reflecting the bending change state of road crossing structures or waterway segments in local areas. Compared with methods that rely solely on single-point curvature or a single geometric index, this method, through cumulative analysis of adjacent curvature changes, allows the overall spatial trend of line segment bending changes to be fully expressed, thereby improving the stability of identifying the geometric features of the same geographic entity from different data sources.

[0041] This embodiment combines the direction change sequence with the node spacing sequence, enabling the line segment direction turning pattern to reflect not only the direction change itself but also the spatial distribution of nodes. This allows the direction change structural quantity to simultaneously reflect both the turning state and the spatial discrete state of the line segment. In this way, it is possible to maintain an effective description of the line segment structural features under data conditions with different sampling densities or geometric representations, reducing structural identification bias caused by differences in sampling density.

[0042] This embodiment, based on the acquisition of curvature change and orientation change structural quantities, introduces node density and the total length of line segments within the sampling area, and constructs geometric fingerprint values ​​through geometric coupling. This allows the curvature features, orientation change features, and node distribution status of spatial candidate units to be uniformly expressed as a comprehensive structural feature. In this way, the local geometric structure of the same geographic entity in highway and waterway data can be transformed into comparable structural representations, thus providing a stable basis for subsequent identification of entities with the same name and reducing entity identification errors caused by differences in geometric representation methods from different data sources.

[0043] Example 5 Please refer to Figure 2 Specifically: S51, Read the set of spatial candidate units in the highway vector dataset and the set of spatial candidate units in the waterway vector dataset, and read the geometric fingerprint value Fg corresponding to each spatial candidate unit obtained in step S4; Based on the center location of the spatial candidate unit, the coverage area of ​​the sampling area, and the unit number to which the geometric fingerprint value belongs, a one-to-one pairing search is performed between the spatial candidate units on the highway side and the spatial candidate units on the waterway side to establish spatial candidate unit pairs. For each pair of spatial candidate units, the corresponding road-side geometric fingerprint value Fg(road) and waterway-side geometric fingerprint Fg(water) are read, and the entity difference Df is calculated. The entity difference Df is obtained by taking the absolute value of the difference between the road-side geometric fingerprint value Fg (road) and the waterway-side geometric fingerprint Fg (water). When obtaining the entity difference Df, a center interval LJ is introduced to constrain the space. The center interval LJ only performs geometric fingerprint difference calculation on spatial candidate units whose center interval is not greater than the preset pairing range value, thus avoiding the entry of distant irrelevant units into the judgment process.

[0044] S52. Compare the acquired entity difference quantity Df with the preset judgment value Td to determine the entity status; the method is as follows: If the entity difference Df is less than or equal to the preset judgment value Td, then the spatial candidate unit pair is determined to be entities with the same name, and spatial fusion processing is performed. When the entity difference Df is greater than the preset judgment value Td, the candidate unit pair in the space is judged to be different entities, and the original data structure is maintained. The formula for obtaining the judgment value Td is: Td = μDf + kb × σDf; In the formula, μDf represents the average value of entity variance, σDf represents the standard deviation of entity variance, and kb represents the variance adjustment coefficient.

[0045] In this embodiment, spatial candidate unit pairs are established between the highway vector dataset and the waterway vector dataset. Each spatial candidate unit's corresponding geometric fingerprint value is read and compared one by one, enabling local structures that may correspond to the same geographic entity in different data sources to be matched under unified rules. Compared to traditional methods that rely on name fields or simple spatial proximity judgments, this approach, by performing a one-to-one pairing retrieval of spatial candidate units, allows objects such as bridge crossing structures and waterway intersection structures in highway and waterway data to be associated at the structural level, thereby reducing the difficulty of entity identification caused by inconsistent names or missing attribute information.

[0046] This embodiment introduces a center interval to constrain the spatial range during the calculation of entity differences, ensuring that only spatial candidate units within a reasonable spatial distance participate in the geometric fingerprint difference calculation, thereby preventing distant spatial units from entering the matching process. In this way, while ensuring the reasonableness of the matching range, interference from irrelevant data on the recognition process can be reduced, making cross-layer entity recognition more targeted and stable.

[0047] This embodiment determines the entity status by comparing the entity difference quantity with the judgment value, and dynamically determines the judgment range based on the distribution of the difference quantity in the overall data, enabling the entity recognition process to adapt to the data distribution characteristics of different regions. In this way, spatial candidate units with similar geometric structures in highway and waterway data can be fused, while spatial candidate units with significant geometric differences retain their original structure. This reduces duplicate records or spatial mismatches in bridge crossing areas or waterway intersection areas during electronic map fusion, resulting in a more harmonious expression of the spatial relationships between highway and waterway networks in the electronic map.

[0048] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended technical solutions and their equivalents.

Claims

1. A method for fusing electronic maps, characterized in that: Includes the following steps: S1. Read the highway vector dataset and the waterway vector dataset, extract the spatial candidate units that form entities with the same name from the two types of data, and construct a sampling area with a radius of Rs at the center of each spatial candidate unit. Perform discrete sampling on the line segments in the sampling area according to a fixed sampling interval Ls to obtain the sampling node P and the node spacing sequence Ds={Di}. S2. Based on the sampling node P, analyze the three adjacent sampling nodes, calculate the discrete curvature K, and calculate the line segment direction angle xphe and direction change Δphe according to the two adjacent sampling nodes to form a curvature sequence and a direction change sequence. S3. Based on the curvature sequence, calculate the curvature change Vc. According to the direction change sequence and the node spacing sequence Ds={Di}, calculate the direction change structure Sd. At the same time, combine it with the node density ρ of the sampled nodes to form a set of geometric structure feature quantities of spatial candidate units. S4. Calculate the geometric fingerprint value Fg of the spatial candidate unit based on the curvature change Vc, the direction change structure Sd, the node density ρ, and the total length of the line segment Lt in the sampling area. S5. Establish candidate spatial candidate unit pairs between the highway vector dataset and the waterway vector dataset. Based on the corresponding geometric fingerprint value Fg, calculate the entity difference Df and compare the entity difference Df with the preset judgment value Td to determine the entity status.

2. The electronic map fusion method according to claim 1, characterized in that: S1 includes S11; S11. Read the highway vector dataset and the waterway vector dataset, extract the bridge nodes and road crossing nodes in the highway vector dataset and the waterway centerline intersection nodes in the waterway vector dataset, and perform a proximity association retrieval based on the spatial position relationship of each node. The nodes that meet the preset proximity constraints are merged to generate a set of spatial candidate units; then, a sampling region with a radius of Rs is constructed with the center position (X, Y) of each spatial candidate unit as the center. The center position (X, Y) is obtained by acquiring the coordinate values ​​xj and yj of each data point and calculating the average values ​​of the horizontal and vertical coordinates respectively. Specifically, the average value of the horizontal coordinate X is the sum of all horizontal coordinate values ​​xj divided by the number of data points Nq. Similarly, the average value of the vertical coordinate Y is the sum of all vertical coordinate values ​​yj divided by the number of data points Nq. The obtained (X, Y) represents the center position coordinates.

3. The electronic map fusion method according to claim 1, characterized in that: S1 also includes S12; S12. Extract the highway and waterway segments that fall within the sampling area. Perform equidistant sampling on each segment according to a fixed sampling interval Ls to form a set of sampling nodes P. Based on the coordinate order of the sampling nodes, calculate the local curvature C, direction angle phe, and node density ρ corresponding to each sampling node in sequence, and calculate the node spacing D between adjacent sampling nodes to generate the node spacing sequence Ds={Di}. The local curvature C is obtained as follows: First, select three adjacent sampling nodes, namely Pi-1, Pi, and Pi+1, with coordinates (xi-1, yi-1), (xi, yi), and (xi+1, yi+1). First, calculate the product of two vectors: one is the product of the difference between the x and y coordinates from point Pi-1 to point Pi and the difference between the x and y coordinates from point Pi to point Pi+1; then subtract another similar product; and place the result in the numerator. The denominator calculates the lengths of two vectors, which are the sum of the squares of the lengths from point Pi-1 to point Pi and from point Pi to point Pi+1. The entire result is then cubed and a minimum value of 0.01 is added to obtain the local curvature C. The direction angle phe is obtained by calculating the ratio of the difference between the ordinates and the difference between the abscissas of two adjacent points to obtain the slope of the line segment; then, substituting this ratio into the arctangent function, the direction angle phe of the line segment is obtained. The node density ρ is obtained by dividing the total number of sampling nodes in the sampling area by the product of the square of the sampling area radius and the value of π. The formula for obtaining the node spacing D is: ; In the formula, Di represents the node spacing of sampling node Pi, and (xi, yi) and (xi+1, yi+1) represent the coordinates of sampling nodes Pi and Pi+1, respectively.

4. The electronic map fusion method according to claim 3, characterized in that: S2 includes S21 and S22; S21. For each sampling node, select the preceding and following sampling nodes adjacent to the sampling node, first obtain the local curvature C of the sampling node, and then calculate the discrete curvature K based on the local curvature C: The formula for obtaining the discrete curvature K is as follows: ； In the formula, Ci represents the local curvature of sampling node Pi, Di+1 represents the node spacing of sampling node Pi+1, Di-1 represents the node spacing of sampling node Pi-1, Li represents the node spacing from sampling node Pi to its adjacent sampling points, and Ki represents the discrete curvature of sampling node Pi. Integrate all discrete curvatures K to obtain a curvature sequence.

5. The electronic map fusion method according to claim 4, characterized in that: S2 also includes S22; S22. Calculate and obtain the line segment direction angle xphe based on the direction angle phe of two adjacent sampling nodes; The line segment direction angle xphe is obtained as follows: read the original direction angle phe corresponding to the previous node, the current node, and the next node of the current sampling node, and calculate the absolute value of the difference between each direction angle phe; then combine each original direction angle with the corresponding absolute value of the difference to obtain the line segment direction angle xphe corresponding to the current sampling node. Based on the line segment direction angle xphe, the direction change Δphe is calculated and obtained, and the direction change Δphe of all sampling nodes is integrated to obtain the direction change sequence; The formula for obtaining the change in direction Δphe is: ; In the formula, Δphei represents the change in direction of sampling node Pi, xphei represents the line segment direction angle of sampling node Pi, and xphe(i-1) represents the line segment direction angle of sampling node Pi-1.

6. The electronic map fusion method according to claim 5, characterized in that: S3 includes S31; S31. Perform adjacent difference expansion processing on the curvature sequence to obtain the change of the bending state of the line segment inside the spatial candidate unit along the sampling direction, and obtain the curvature change amount Vc; then combine the direction change sequence and the node spacing sequence to calculate the direction change structure amount Sd under the combined action of direction turning and spatial discretization relationship. The curvature change Vc is obtained as follows: First, read the discrete curvature values ​​corresponding to the current sampling node and its adjacent sampling nodes in the curvature sequence in sequence; then calculate the absolute difference between the current discrete curvature and the previous discrete curvature, and the absolute difference between the next discrete curvature and the current discrete curvature respectively. Add the absolute values ​​of the two differences to obtain the local curvature fluctuation at the current sampling node; read the distance between the two nodes adjacent to the sampling node and sum the distances of the two nodes as the local spatial expansion scale; finally, divide the local curvature fluctuation by the local spatial expansion scale and sum it point by point for all intermediate sampling nodes to obtain the curvature change Vc. The structural quantity Sd representing the change in direction is obtained using the following formula: ； In the formula, Ns represents the number of sampling nodes.

7. The electronic map fusion method according to claim 6, characterized in that: S3 also includes S32; S32. Based on the curvature change Vc and the direction change structure Sd, and combined with the obtained node density ρ, a set of geometric structural feature quantities of spatial candidate units is constructed. Wherein, node density ρ represents the number of sampling nodes distributed within a unit sampling area.

8. The electronic map fusion method according to claim 7, characterized in that: S4 includes S41 and S42; S41. Read the highway and waterway segments that fall into the sampling area, extract the node spacing between each adjacent sampling node in the order of sampling nodes, and then perform cumulative calculation on all node spacings to obtain the total length Lt of the line segments in the sampling area corresponding to the current spatial candidate unit. S42. After obtaining the total length Lt of the line segment, combine it with the curvature change Vc, the direction change structure Sd and the node density ρ, and construct the geometric fingerprint value Fg of the current spatial candidate unit through geometric coupling calculation. The geometric fingerprint value Fg is obtained using the following formula: 。 9. The electronic map fusion method according to claim 8, characterized in that: S5 includes S51; S51. Read the set of spatial candidate units in the highway vector dataset and the set of spatial candidate units in the waterway vector dataset, and read the geometric fingerprint value Fg corresponding to each spatial candidate unit obtained in step S4. Based on the center location of the spatial candidate unit, the coverage area of ​​the sampling area, and the unit number to which the geometric fingerprint value belongs, a one-to-one pairing search is performed between the spatial candidate units on the highway side and the spatial candidate units on the waterway side to establish spatial candidate unit pairs. For each pair of spatial candidate units, the corresponding road-side geometric fingerprint value Fg(road) and waterway-side geometric fingerprint Fg(water) are read, and the entity difference Df is calculated. The entity difference Df is obtained by taking the absolute value of the difference between the road-side geometric fingerprint value Fg (road) and the waterway-side geometric fingerprint Fg (water). When obtaining the entity difference Df, a center interval LJ is introduced to constrain the space. The center interval LJ only performs geometric fingerprint difference calculation on spatial candidate units whose center interval is not greater than the preset pairing range value, thus avoiding the entry of distant irrelevant units into the judgment process.

10. The electronic map fusion method according to claim 9, characterized in that: S5 also includes S52; S52. Compare the acquired entity difference quantity Df with the preset judgment value Td to determine the entity status; the method is as follows: If the entity difference Df is less than or equal to the preset judgment value Td, then the spatial candidate unit pair is determined to be entities with the same name, and spatial fusion processing is performed. When the entity difference Df is greater than the preset judgment value Td, the candidate unit pair in the space is judged to be different entities, and the original data structure is maintained. The formula for obtaining the judgment value Td is: Td = μDf + kb × σDf; In the formula, μDf represents the average value of entity variance, σDf represents the standard deviation of entity variance, and kb represents the variance adjustment coefficient.

Images