A road network generalization method under the constraint of populated areas

Abstract

The present invention provides a road network generalization method under the constraint of populated areas, belonging to the technical field of road network cartography. First, the populated areas are clustered to generate settlements, and then the roads are classified according to the topological relationship between the roads and the settlements; then neurons are constructed according to the road types included in the settlements, and effective paths between adjacent neurons are searched; finally, according to the principle of minimizing the path travel time cost, the paths of adjacent neuron pairs and adjacent neuron groups forming a triangular structure are simplified in turn, so that the connectivity between settlements and the functionality of the road network can be better maintained. Compared with the prior art that mainly relies on the semantic, geometric and topological information of the road network for road simplification, the present invention uses populated area elements as the constraints for road network generalization, fully considering the geographical association between roads and populated areas, making the simplified road network more in line with the actual travel needs.

Description

Technical Field

[0001] The present invention relates to a comprehensive method for road network under the constraint of populated areas, belonging to the technical field of road network cartography. Background Art

[0002] Roads are important geographical elements that constitute the framework of a map and are the key research objects of cartographic generalization. The goal of road network generalization is to retain important roads during the process of scale reduction, discard redundant roads that do not conform to the target scale, and maintain the overall shape and connectivity of the road network.

[0003] Currently, the main methods for road network generalization include: methods based on semantic levels, methods based on graph theory, methods based on Stroke, methods based on mesh density, and other hybrid methods. The methods based on semantic levels mainly rely on the semantic information of roads, do not establish an effective model for the road network, and ignore the topological and geometric information of the road network. The method for evaluating the importance of roads based on graph theory mainly focuses on the analysis of the connectivity and topological structure between road networks, plays an important role in maintaining the connectivity of the road network, but ignores the semantic and geometric information of roads. The method based on Stroke can effectively simulate the influence of the visual length of roads on the evaluation of road importance during the manual selection process, and can comprehensively consider the extensibility, topological consistency, and context information between road targets while maintaining the integrity of roads during the road comprehensive selection process. The method based on mesh density can better maintain the relative density of the road network before and after generalization and the topological semantic consistency of the road network. However, most of the above methods do not consider the association between the road network and other geographical elements, which will affect the coordination between geographical elements and even lead to situations that violate the laws of reality. Roads mainly serve people's lives. If the roads after generalization cannot guarantee the connectivity between populated areas, they will deviate from the laws of reality, violate people's actual needs, and thus lead to a decrease in the connectivity between populated areas. Summary of the Invention

[0004] The purpose of the present invention is to provide a comprehensive method for road network under the constraint of populated areas to solve the problem of reduced connectivity between populated areas caused by the failure to fully consider the geographical relevance between populated areas and roads during the existing road network generalization process.

[0005] The present invention proposes a comprehensive method for road network under the constraint of populated areas, which includes the following steps:

[0006] 1) Obtain the populated area data and road network data within the target area;

[0007] 2) Cluster the obtained residential area data to obtain different settlements; classify the roads according to the topological relationship between the roads and the settlements; the road classification includes E-type roads and P-type roads. An E-type road means that one end of the road is inside the residential area settlement and the other end is outside the residential area settlement. A P-type road means that the road runs through the residential area settlement and both ends of the road are outside the residential area settlement;

[0008] 3) Construct neurons for the settlements according to the road types included in the settlements; the neurons are divided into type I neurons and type II neurons. A type I neuron means that the neuron includes an E-type road and a type I road or only includes an E-type road. A type II neuron means that the neuron only includes a P-type road;

[0009] 4) Search for all valid paths between any two adjacent neurons, and find the optimal path among them. The optimal path refers to the path with the lowest travel time cost among all valid paths between adjacent neurons. Simplify all valid paths between the two adjacent neurons according to the optimal path to obtain the simplified path between the adjacent neurons;

[0010] 5) After completing the path simplification between all adjacent neurons, treat all valid paths between adjacent neurons as a whole for processing. Search for the triangular structure formed by three mutually adjacent settlements, find the optimal path for each side in the triangular structure, and determine whether the side with the largest time cost in the formed triangular structure can be replaced by the combination of the other two sides. If the time cost corresponding to the side with the largest time cost is greater than the sum of the time costs of the other two sides, delete all valid paths corresponding to the side with the largest time cost.

[0011] The present invention uses residential area elements as the constraints for road integration, clusters the residential areas to obtain different residential area settlements, classifies the roads considering the topological relationship between the roads and the settlements, constructs neurons by combining the settlements and the roads topologically connected to the settlements, searches for valid paths between adjacent neurons, and simplifies the paths of adjacent neuron pairs and adjacent neuron groups forming a triangular structure in turn according to the principle of the minimum travel time cost of the paths, so that the connectivity between settlements and the functionality of the road network can be better maintained. Compared with the prior art that mainly relies on the semantic, geometric and topological information of the road network for road simplification, the present invention uses residential area elements as the constraints for road network integration, fully considers the geographical relationship between the roads and the residential areas, and makes the simplified road network more in line with the actual travel needs.

[0012] Further, the roads also include: type I roads and O-type roads. A type I road means that each end point of the road is inside the residential area settlement. An O-type road means that the road is completely outside the residential area settlement.

[0013] Classifying roads according to the topological relationship between roads and settlements in the above way facilitates the subsequent construction of neurons and the search for effective paths.

[0014] Furthermore, the effective path between adjacent neurons refers to all paths formed by the E-type roads from the starting neuron to the E-type roads of the ending neuron. Among them, when the neuron is a type-II neuron, the P-type road within the type-II neuron is used as the E-type road of this neuron.

[0015] Since the E-type roads connect inside and outside the settlement and the P-type roads run through the settlement, the passage paths connecting two adjacent neurons can be determined through these two types of roads, while the type-I roads are the internal paths of neurons.

[0016] Furthermore, the effective path refers to the path with the lowest travel time cost, and the calculation formula for this travel time cost is:

[0017]

[0018] In the formula, L is the road length, V is the design speed of different grades of highways, and ST is the road curvature.

[0019] Since there may be multiple E-type roads for a neuron in different orientations of the settlement, there may be multiple paths between adjacent neurons. Calculate the travel time costs between all combinations of E-type roads, select the effective paths that meet the requirements, further reduce data redundancy, and make it more convenient for subsequent path simplification.

[0020] Furthermore, the basis for simplifying the effective path between adjacent neurons is:

[0021] α × (κ f + κ s + κ e ) < κ t

[0022] In the formula, α is the tolerable cost ratio; κ f is the travel time cost of the optimal path; κ s is the internal travel time cost of the optimal path and the current path in the starting neuron; κ e is the internal travel time cost of the optimal path and the current path in the ending neuron; κ t is the travel time cost of the current path; if the travel time cost of the current path is greater than the sum of the travel time cost of the optimal path and the internal travel time cost under the tolerable cost ratio, then remove this path.

[0023] Furthermore, simplify the paths in the adjacent neuron group that can form a triangular structure through the effective path. The basis for simplifying the paths in the adjacent neuron group is:

[0024] α × (κ e1 + κ e2 ) < κ em

[0025] where α is the tolerable cost ratio; κ em is the maximum travel time cost among neighboring neurons; κ e1 and κ e2 are the other two travel time costs.

[0026] By the above method, the path simplification between neighboring neurons and neighboring neuron groups is realized. The core of the two simplification methods is to select the path with the minimum travel time cost, so as to eliminate redundant paths.

[0027] Furthermore, the tolerable cost ratio refers to the degree to which the service capacity of the optimal path in the target area can be accepted to be weaker than that of the current path, representing the simplification strength of the road network. Its value range is [0.6, 1]. When the simplification strength of the road network is stronger, the tolerable cost is smaller.

[0028] When performing path simplification, the tolerable cost ratio can be used to adjust the simplification strength of the road network to achieve path simplification when the travel time costs are not much different.

[0029] Furthermore, before constructing neurons for settlements, the settlements are screened according to the importance of the settlements, and neurons are constructed for the selected settlements.

[0030] The settlements are screened according to their importance to eliminate a few weak settlements without affecting the overall road simplification, so as to enhance the constraint effect.

[0031] Furthermore, the importance of the settlement is determined according to the area of the settlement, the residential density, the Voronoi area, and the road index, where the Voronoi area refers to the spatial influence range of the settlement, and the road index refers to the road connectivity of the settlement.

[0032] Four parameters are used to determine the importance of the settlement. Among them, the larger the area and the residential density, the more important the settlement can be considered; the Voronoi area represents the spatial influence range, and the larger the Voronoi area, the more representative the settlement; the road index is also considered to better ensure the connectivity of the roads between settlements after subsequent simplification; through these four parameters, the importance of the settlement can be more accurately characterized.

[0033] Furthermore, the calculation formula for the importance of the settlement is:

[0034] SP(v i ) = μ × [w 1 × S(v i) + w 2 ×RD(v i ) + w 3 ×VS(v i )] + w 4 ×RI(v i )

[0035] In the formula, v i (i = 1, 2,..., n) is the i-th residential settlement, n is the total number of generated settlements, S is the settlement area, RD is the residential density, VS is the Voronoi area, RI is the road index, w 1 , w 2 , w 3 and w 4 are the weights of S, RD, VS, and RI respectively, and w 1 + w 2 + w 3 + w 4 = 1, μ is the adjustment coefficient. When the settlement belongs to type I neurons, μ = 1; when the settlement belongs to type II neurons, μ < 1.

[0036] The importance of different settlements is determined by the above formula, and by setting the adjustment coefficient under different types of neurons, the adaptive adjustment of the importance of settlements under different types of neurons is achieved. Brief Description of the Drawings

[0037] Figure 1 is the specific implementation flowchart of the road network integration method under the constraint of residential areas of the present invention;

[0038] Figure 2(a) is an example diagram of type E road in the road type;

[0039] Figure 2(b) is an example diagram of type P road in the road type;

[0040] Figure 2(c) is an example diagram of type I road in the road type;

[0041] Figure 2(d) is an example diagram of type O road in the road type;

[0042] Figure 3(a) is an example diagram of a typical type I neuron;

[0043] Figure 3(b) is an example diagram of a type I neuron including a type P road that cannot directly serve settlements;

[0044] Figure 3(c) is an example diagram of a typical type II neuron;

[0045] Figure 3(d) is an example diagram of a type II neuron including a type P road that cannot directly serve settlements;

[0046] Figure 4 is a schematic diagram of the path simplification between adjacent neurons;

[0047] Figure 5 It is a schematic diagram of path simplification between adjacent neuron groups. Specific implementation manner

[0048] The following further describes the specific implementation manner of the present invention in conjunction with the accompanying drawings.

[0049] The present invention proposes a method for road network generalization under the constraint of residential areas. The specific process of this method is as Figure 1 shown. First, cluster the residential area data to generate settlements, and then classify the roads according to the topological relationship between the roads and the settlements; then, construct neurons according to the road types included in the settlements; after that, search for all effective paths between adjacent neurons and find the optimal path; finally, according to the principle of minimizing the path passing time cost, simplify the paths between adjacent neurons and the paths between the neuron groups forming a triangular structure in turn.

[0050] Step 1. Obtain data

[0051] The present invention first obtains the residential area data and road network data in the target area. Both of these two types of data can be obtained from the already established geographical database to obtain the vector data of residential areas and road networks in the area. According to the actual requirements for the accuracy of the road network, select the data with an appropriate scale in the existing database. For example, if it is necessary to generate road network data of 1:5000, it can be obtained by simplifying the road network data of 1:2000. The present invention aims to use the obtained residential area data as a constraint and perform subsequent path simplification according to the geographical correlation between the roads and the residential areas.

[0052] Step 2. Generate settlements and classify roads

[0053] According to the obtained residential area data and road network data, first cluster the residential area data to generate settlements, and then classify the roads according to the topological relationship between the roads and the settlements.

[0054] First, use a clustering algorithm to cluster the obtained residential area data to obtain different settlements, and extract the settlement contours. In the clustering of residential areas, generally use the density-based clustering method. In this embodiment, use the DBSCAN (Density-Based Spatial Clustering of Applications with Noise) algorithm to cluster the residential areas, and use the ball pivoting method to extract the contours of the settlements. As other implementation manners, the CFSFDP algorithm or K-means algorithm can also be used for the clustering of residential areas. As other implementation manners, the edge detection method can also be used for the extraction of settlement contours.

[0055] Then, classify the roads according to the topological relationship between the roads and the settlements. In the present invention, the roads are classified into E-type roads, P-type roads, I-type roads, and O-type roads. Among them, as shown in Fig. 2(a), the E-type road means that one end of the road is inside the settlement and the other end is outside the settlement, and it can directly connect the inside and outside of the settlement; as shown in Fig. 2(b), the P-type road means that the road runs through the settlement and both ends of the road are outside the settlement. Similarly, this type of road can connect the inside and outside of the settlement and is common in small settlements distributed along the road and occasionally appears in larger settlements; as shown in Fig. 2(c), the I-type road means that all endpoints of the road are inside the residential settlement, and its function is to connect different areas inside the settlement and the E-type road. The larger the settlement scale, the more I-type roads and the more complex the structure; as shown in Fig. 2(d), the O-type road is separated from the target settlement and is completely outside the settlement and does not participate in the construction of subsequent neurons. In addition, the classification of roads is settlement-oriented, so the type of the same road may be different in different clusters.

[0056] Step 3. Settlement screening

[0057] Screen the settlements according to the importance of the settlements. Since the present invention conducts road network generalization under the constraint of residential areas, some important settlements are selected as constraint conditions, and some weak settlements, that is, settlements with low importance, are removed to enhance the constraint effect. The importance of the settlements is determined according to the area of the settlements, residential density, Voronoi area, and road index. The specific calculation formula is as follows:

[0058] SP(v i ) = μ × [w 1 ×S(v i ) + w 2 ×RD(v i ) + w 3 ×VS(v i )] + w 4 ×RI(v i ) (1)

[0059] In the formula, v i (i = 1, 2,..., n) is the i-th residential settlement, n is the total number of generated settlements, S is the settlement area, RD is the residential density, VS is the Voronoi area, RI is the road index, w 1 , w 2 , w 3 and w 4 are the weights of S, RD, VS, and RI respectively, and w 1 + w 2 + w 3 + w 4= 1, where μ is an adjustment coefficient. Since settlements located at road intersections have more significant geographical significance than those distributed along roads when parameters such as area and density are the same, μ = 1 when the settlement belongs to type I neurons; μ < 1 when the settlement belongs to type II neurons. Among them, the weight values of different parameters can be adjusted according to actual application requirements.

[0060] Among them, the settlement area is the most intuitive manifestation of the importance of the settlement. Generally, settlements with a large area are more important than those with a small area; the residential density reflects the density of residential areas within the settlement. When the settlement areas are the same, settlements with a large residential density are more important. The calculation of residential density is shown in formula (2); the Voronoi area refers to the spatial influence range of the settlement. The larger the Voronoi area, the more representative the settlement is and the greater the possibility of being selected; the road index refers to the road connectivity of the settlement. The more roads and the higher the grades involved in the settlement, the higher the importance of the settlement. The road index is determined according to the grades of the roads between settlements. The settlement is mainly connected to the outside and inside by type E roads and type P roads. Therefore, the road index is mainly calculated based on the quantity and grades of type E roads and type P roads, and its calculation formula is shown in formula (3):

[0061]

[0062] In the formula, m is the number of residential areas in settlement v i ; S(p j ) is the area of residential area p i in settlement v j ; S(v j ) is the area of settlement v i .

[0063]

[0064] In the formula, k is the total number of effective connected roads of settlement v i ; Rank(r j ) is the numerical grade of road r j . The numerical grade of the road refers to the road weight value divided according to the grade of the road, as shown in Table 1 specifically:

[0065] Table 1:

[0066]

[0067] Determine the importance of each settlement according to formulas (1)-(3), and eliminate some settlements according to the set proportional threshold, which can be set according to the actual situation of the obtained settlements. All subsequent processes are based on the retained settlements. As other implementation manners, the importance of the settlements can also be determined according to actual requirements. For example, without considering road factors, it can be determined only by using the area of the settlements, the residential density, and the Voronoi area, or other factors affecting the settlements can also be considered, such as population density, terrain, etc. As other implementation manners, if the residential data is at the target scale, settlement screening may not be required.

[0068] Step 4. Neuron construction

[0069] Construct neurons according to the road types included in the settlements, that is, combine the settlements with the roads topologically connected to the settlements. The neurons are divided into two types according to the road types in the neurons, type I neurons and type II neurons. Among them, type I neurons refer to those neurons that include type E roads and type I roads or only include type E roads, as shown in Fig. 3(a); at the same time, the present invention follows the principle of "only the connection with the settlement core is the real connection", that is, the target road can directly serve the settlement only when it intersects with other roads inside the settlement or the influence range of the target road covers the core area of the settlement. As shown in Fig. 3(b), in Fig. 3(b), there are type E roads and type P roads in the settlement, but the type P road is located at the edge of the settlement and cannot directly serve the settlement. Therefore, this type P road can be regarded as an O-type road and does not participate in the construction of this neuron, and this neuron is a type I neuron.

[0070] Among them, type II neurons refer to those neurons that only include type P roads. Most type II neurons only include one type P road, and the settlement completely depends on this road, as shown in Fig. 3(c); a small number of type II neurons include multiple type P roads, but the principle of "only the connection with the settlement core is the real connection" also needs to be followed. As shown in Fig. 3(d), the settlement is penetrated by two type P roads, but the right type P road is located at the edge of the settlement and cannot directly serve the settlement. Therefore, this type P road does not participate in the construction of this neuron.

[0071] Step 5. Search for effective paths

[0072] Search for all valid paths between any two adjacent neurons. A valid path between neurons refers to the path generated from the E-type road of the starting neuron to the E-type road of the ending neuron. Other E-type roads and I-type roads in the starting neuron and the ending neuron do not participate in path construction. The P-type roads within the effectively connected settlements in type-II neurons can be regarded as the E-type roads of that neuron. Since there may be more than one E-shaped road in a neuron, the valid path between adjacent neurons refers to the path generated by directly connecting the E-type road of the starting neuron to the E-type road of the ending neuron, and the roads connecting to external roads of neurons or the E-type roads of other neurons do not participate in the path construction of adjacent neurons this time. A valid path refers to the path with the minimum travel time cost. Because searching for all paths between two network nodes in a large network is very time-consuming, and people usually choose the shortest and fastest path when traveling. When planning a path, people are more concerned about the travel time than the travel distance and tend to choose a higher-level road with a faster speed. Therefore, the travel time is used as the cost. The travel time is mainly determined by the road length, road grade, and road tortuosity. The calculation formula is as follows:

[0073]

[0074] In the formula, L is the road length, V is the design speed of different grades of highways, and ST is the road curvature.

[0075] Since there may be multiple E-type roads in different orientations of a neuron in a settlement, there may be multiple paths between a pair of neurons. When searching for the valid path between a pair of adjacent neurons, the E-type roads of the starting neuron and the E-type roads of the ending neuron are arranged and combined, and the Dijkstra algorithm is used to search for the path with the minimum travel time cost between all combinations of E-type roads to determine the valid path.

[0076] Step 6. Path simplification

[0077] The path simplification of the present invention mainly includes two stages. One is the path simplification between adjacent neurons, and the other is the path simplification between groups of adjacent neurons.

[0078] Path simplification between adjacent neurons:

[0079] Find the optimal path among all the generated valid paths. The optimal path refers to the path with the lowest travel time cost among all valid paths. Simplify all the valid paths of two adjacent neurons according to the optimal path. The specific process is as follows:

[0080] Arbitrarily select a valid path and compare it with the optimal path. If the cost of this path is greater than the sum of the optimal path cost and the internal travel cost, then it is considered that this path can be replaced. It can be formally expressed as:

[0081] α × (κ f + κ s + κ e ) < κ t (5)

[0082] In the formula, α is the tolerable cost ratio; κ f is the travel time cost of the optimal path; κ s is the internal travel time cost between the optimal path and the current path in the starting neuron; κ e is the internal travel time cost between the optimal path and the current path in the ending neuron; κ t is the travel time cost of the current path; among them, the tolerable cost ratio represents the degree to which the service capacity of the optimal path in the target area is weaker than that of the current path can be accepted, that is, the simplification intensity of the road network. Its value range is [0.6, 1]. When the simplification intensity of the road network is higher, the value of α is smaller, and it can be selected according to the actual simplification requirements of the road network.

[0083] For example Figure 4 As shown, there are 3 paths between settlement A and settlement B. Path I starts from p1 and reaches p2 along road ea. Path II starts from p1 and reaches p3 along roads ec and eb. Path III starts from p1 and reaches p4 along roads ec and ed. Among them, Path I is the optimal path between settlement A and settlement B. The starting and ending points of Path II are p1 and p3 respectively, and its travel cost is 4.1. However, the total cost of starting from p1, reaching p2 along Path I, and then reaching p3 through the internal road ib is 2.8, which is less than the cost of Path II. Therefore, Path II can be replaced by Path I. Reducing the value of α can increase the simplification intensity. For example, the cost of Path III is 4.7, and the total cost of starting from p1, reaching p2 along Path I, and then reaching p4 through roads ib and ia is 5.0. When α = 1.0, formula (5) does not hold, and Path I cannot replace Path III. But when α < 0.94, formula (5) holds, and Path I can replace Path III.

[0084] Path simplification of adjacent neuron groups:

[0085] All effective paths between adjacent neurons are treated as a whole to search for a triangular structure composed of three mutually adjacent settlements. For adjacent neuron groups that can form a triangular structure through effective paths, the paths in the adjacent neuron groups are simplified according to the path with the largest travel time cost among the simplified paths in the adjacent neuron groups. It is judged whether the side with the largest time cost in the formed triangular structure can be replaced by the combination of the other two sides. If the time cost corresponding to the side with the largest time cost is greater than the sum of the time costs of the other two sides, the simplified path corresponding to the largest time cost is deleted. Among them, the paths representing the sides of the triangle are all the paths with the lowest travel time cost between two neurons on the corresponding side, that is, the optimal paths between two neurons. The formal expression of the path simplification of this adjacent neuron group is as follows:

[0086] α×(κ e1 +κ e2 )<κ em (6)

[0087] In the formula, α is the tolerable cost ratio; κ em is the largest travel time cost among adjacent neurons; κ e1 and κ e2 are the other two travel time costs. Among them, the value range of α is the same as that in the path simplification between the above adjacent neurons.

[0088] For example Figure 5 as shown, Figure 5 in the neuron group composed of settlements A, B, and C. Among them, the travel time cost between settlement A and settlement B is 4, the travel time cost between settlement B and settlement C is 2, and the travel time cost between settlement A and settlement C is 8. It can be seen from formula (6) that the path between settlement A and settlement C can be replaced by the combination of the paths between settlement A and settlement B and between settlement B and settlement C. Therefore, all paths between settlement A and settlement C are deleted to achieve path simplification.

[0089] Through the above process, using the residential area to constrain the road network integration, fully considering the geographical relevance between roads and residential areas, the overall selection of the complex road network is decomposed into relatively simple path simplification between neurons, redundant paths are removed, so that the obtained comprehensive result is highly consistent with the spatial distribution of residential area elements, and the connectivity between settlements and the functionality of the road network can be ensured.

Claims

1. A road network integration method under the constraint of residential areas, characterized in that, the method comprises the following steps: 1) Obtain the residential area data and road network data in the target area; 2) Cluster the obtained residential area data to obtain different settlements; and classify the roads according to the topological relationship between the roads and the settlements; the road classification includes E-type roads and P-type roads. An E-type road means that one end of the road is inside the residential area settlement and the other end is outside the residential area settlement. A P-type road means that the road runs through the residential area settlement and both ends of the road are outside the residential area settlement; 3) Construct neurons for the settlements according to the road types included in the settlements; the neurons are divided into type I neurons and type II neurons. A type I neuron means that the neuron includes an E-type road and a type I road or only includes an E-type road. A type II neuron means that the neuron only includes a P-type road; 4) Search for all effective paths between any two adjacent neurons, and find the optimal path among them. The optimal path refers to the path with the lowest travel time cost among all effective paths between adjacent neurons. Simplify all effective paths between the two adjacent neurons according to the optimal path to obtain the simplified path between the adjacent neurons; 5) After completing the path simplification between all adjacent neurons, treat all effective paths between adjacent neurons as a whole for processing. Search for the triangular structure formed by three mutually adjacent settlements, find the optimal path for each side in the triangular structure, and determine whether the side with the largest time cost in the formed triangular structure can be replaced by the combination of the other two sides. If the time cost corresponding to the side with the largest time cost is greater than the sum of the time costs of the other two sides, delete all effective paths corresponding to the side with the largest time cost; 2. The road network integration method under the constraint of residential areas according to claim 1, characterized in that, the road classification further includes: type I roads and O-type roads; a type I road means that all endpoints of the road are inside the residential area settlement, and an O-type road means that the road is completely outside the residential area settlement; 3. The road network integration method under the constraint of residential areas according to claim 2, characterized in that, the effective path between adjacent neurons refers to all paths formed by the E-type road of the starting neuron to the E-type road of the ending neuron. Among them, when the neuron is a type II neuron, the P-type road in the type II neuron is used as the E-type road of the neuron; 4. The road network integration method under the constraint of residential areas according to claim 3, characterized in that, the effective path refers to the path with the lowest travel time cost, and the calculation formula for the travel time cost is: In the formula, L is the road length, V is the design speed of different grades of highways, and ST is the road curvature.

5. The road network integration method under the constraint of residential areas according to claim 1, characterized in that, the basis for simplifying the effective path between adjacent neurons is: α×(κ f +κ s +κ e )<κ t Where α is the tolerable cost ratio; κ f is the travel time cost of the optimal path; κ s is the internal travel time cost between the optimal path and the current path in the starting neuron; κ e is the internal travel time cost between the optimal path and the current path in the ending neuron; κ t is the travel time cost of the current path; if the travel time cost of the current path is greater than the sum of the travel time cost of the optimal path and the internal travel time cost under the tolerable cost ratio, then remove this path.

6. The road network integration method under the constraint of residential areas according to claim 1, characterized in that, Simplify the paths in adjacent neuron groups that can form a triangular structure through valid paths. The basis for path simplification in the adjacent neuron groups is as follows: α×(κ e1 +κ e2 )<κ em where α is the tolerable cost ratio; κ em is the maximum travel time cost among neighboring neurons; κ e1 and κ e2 are the other two travel time costs.

7. The road network generalization method under residential area constraints according to claim 5 or 6, characterized in that the tolerable cost ratio refers to the degree to which the optimal path can be accepted to have weaker service capabilities in the target area than the current path, representing the simplification intensity of the road network. Its value range is [0.6, 1]. When the simplification intensity of the road network is stronger, the tolerable cost ratio is smaller.

8. The road network generalization method under residential area constraints according to any one of claims 1-6, characterized in that before constructing neurons for settlements, first screen the settlements according to the importance of the settlements, and construct neurons for the selected settlements.

9. The road network generalization method under residential area constraints according to claim 8, characterized in that the importance of the settlement is determined according to the area of the settlement, residential area density, Voronoi area, and road index, where the Voronoi area refers to the spatial influence range of the settlement, and the road index refers to the road connectivity of the settlement.

10. The road network generalization method under residential area constraints according to claim 9, characterized in that the calculation formula for the importance of the settlement is: SP(v i ) = μ × [w 1 × S(v i ) + w 2 × RD(v i ) + w 3 × VS(v i )] + w 4 × RI(v i ) where v i (i = 1, 2, …, n) is the i-th residential settlement, n is the total number of generated settlements, S is the settlement area, RD is the residential density, VS is the Voronoi area, RI is the road index, w 1 , w 2 , w 3 and w 4 are the weights of S, RD, VS, and RI respectively, and w 1 + w 2 + w 3 + w 4 = 1, μ is the adjustment coefficient. When the settlement belongs to type I neurons, μ = 1; when the settlement belongs to type II neurons, μ < 1.

Images