Method for matching road network in freehand sketch based on local shape

Abstract

The invention discloses a local shape-based road network matching method in a freehand sketch, which comprises the following steps of: 1, calculating centrality and intermediate characteristics of each road in the freehand sketch in a road network, and extracting main roads in the area from the road network; 2, calculating the shape similarity between the main road extracted from the freehand sketch and the road in the actual map by using the shape distance factor; and 3, quantifying morphological characteristics of the road. Within a preset threshold range, calculating the similarity between the road in the freehand sketch and the road in the actual map on key point features; 4, quantifying the direction change of the road, and calculating the similarity of the road in the freehand sketch and the road in the measurement map in the advancing direction; and 5, calculating the comprehensive similarity between the roads in the freehand sketch and the roads in the actual map by using a weighted average method. And generating a sorting list of the road matching sets in the actual map based on the calculated sum of the comprehensive similarities.

Description

Technical Field

[0001] The present invention relates to a method for matching hand-drawn sketch road networks based on local shapes, and particularly to a method for matching hand-drawn sketches of roads using factors such as main roads, shape distances of roads, key nodes of roads, and the forward directions of roads. Background Art

[0002] With the wide application of geospatial data in diversified fields such as intelligent transportation and smart cities, intelligent location services, as a key node in the integration of geospatial data and industry applications, are increasingly highlighting their importance. Intelligent Location Services is a comprehensive service that combines technologies such as Geographic Information System (GIS), Global Positioning System (GPS), Internet of Things (IoT), and Artificial Intelligence (AI). It can provide location-related information and services in real time and accurately, and make intelligent responses and decisions according to the needs of users and environmental changes.

[0003] In the research field of intelligent location services, spatial data query plays a crucial role. Spatial data query refers to the process of retrieving, analyzing, and operating on spatial data in a Geographic Information System (GIS) or spatial database, allowing users to search for and extract spatial information based on geographical location, spatial relationships, or attribute conditions. However, existing spatial data query methods for ordinary users are usually database queries based on metadata or geographical ranges, such as queries based on the name of geographical features - finding the "Zifeng Tower", or queries based on the spatial distance between geographical features - finding hotels around the "Zifeng Tower".

[0004] Spatial data query based on hand-drawn geographical sketches is to convert the spatial information in the hand-drawn sketches into query conditions and submit them to the database to retrieve geographical spatial scenes similar to the hand-drawn sketches. This type of query method has been applied to fields such as navigation and crime scene identification. However, due to problems such as incorrect drawing and incomplete drawing, the styles and spatial relationships (including topological relationships, sequential relationships, and positional relationships) of objects between hand-drawn sketches and real maps are not completely consistent. This poses challenges to geographical spatial data query based on hand-drawn sketches.

[0005] As the main geographical elements in geographical spatial scenes, roads show a high degree of similarity between hand-drawn sketches and actual maps. In sketches, people tend to draw roads more frequently and accurately than other roads. The reason for this is consistent with people's visual judgment. Specifically, two places are considered similar not only because they have similar spatial relationships, but also because they have similar structural features, that is, the structural features of the road network.

[0006] However, due to the randomness and subjectivity of hand-drawn sketches, there are the following problems in the matching of hand-drawn sketches based on roads: (1) Inaccuracy of hand-drawn objects: In the sketch, roads are often distorted and only partial roads are drawn; (2) Omission of the number of roads: There are often differences between the number of roads in the sketch and the actual map because people tend to draw only the objects they are familiar with; (3) Inaccurate description of spatial relationships (such as distance and orientation): For example, the distance between two features in a hand-drawn sketch is only a general representation of the approximate distance relationship and cannot accurately and quantitatively express the actual distance between the two features. Summary of the Invention

[0007] The object of the present invention is to propose a method for matching road networks in hand-drawn sketches based on local shapes in view of the limitations of the prior art in the spatial data query service based on hand-drawn maps. This method takes the query of hand-drawn sketches based on the local shapes of roads as the core and systematically designs a matching process for the road network based on hand-drawn maps. The present invention fully considers the inherent diversity and subjectivity characteristics of hand-drawn maps and innovatively introduces a matching mode based on the local shapes of roads. This mode significantly improves the accuracy of spatial data query and matching based on hand-drawn maps and effectively overcomes the difficulties of traditional methods in processing hand-drawn inputs. Through this innovative method, the present invention not only improves the accuracy of spatial data query, but also greatly enhances the quality of spatial data services for ordinary users. This technological breakthrough makes the spatial data query service more in line with the usage habits and cognitive methods of ordinary users, thus laying a solid foundation for improving the user experience and expanding the scope of service applications. The present invention first calculates the centrality and betweenness characteristics of each road in the hand-drawn sketch in the road network and extracts the main roads of the road network in this geographical area. On the basis of the extraction of the main roads, multiple key features are introduced for similarity measurement, including the shape distance of the road, the key node features and the forward direction, and a comprehensive similarity measurement is completed between the hand-drawn sketch and the actual map. The matching results are sorted based on the similarity values, and the best matching results are displayed ( Figure 1 as shown).

[0008] The technical solution adopted by the present invention to solve its technical problems is: A method for matching road networks in hand-drawn sketches based on local shapes, the method comprising the following steps ( Figure 2 as shown):

[0009] Step 1: Extraction of main roads of the road network in the hand-drawn sketch

[0010] The purpose of this step is to identify and extract the main roads in the hand-drawn sketch. As the core elements of regional traffic, the main roads show a high degree of similarity between the hand-drawn sketch and the actual map. The present invention designs a specific algorithm to extract the main roads of the scene from the hand-drawn sketch for subsequent matching with the roads in the actual map.

[0011] Step 2: Comparison of Shape Differences between Roads

[0012] In this step, the shape distance method is used to calculate the shape difference between two roads. For each matching result in the road set retrieved in Step 1, the present invention applies the shape distance method to calculate the shape similarity between the main road in the hand-drawn sketch and the matching road in the actual map retrieval set.

[0013] Step 3: Comparison of the Number of Key Points of Roads

[0014] This step is mainly used to calculate the morphological features of roads and compare the number of key points. Based on the shape difference matching results in Step 2, the present invention calculates the number of key points of the main road in the hand-drawn sketch and the matching road in the actual map retrieval set respectively to quantify the morphological features of the roads. Within the preset threshold range, the similarity of the number of key points between the road in the hand-drawn sketch and the road in the metric map is calculated.

[0015] Step 4: Comparison of the Forward Direction Features of Roads

[0016] This step aims to measure the curvature features of roads. Based on the matching results of the number of key points in Step 3, the present invention calculates the forward direction values of the main road in the hand-drawn sketch and the matching road in the actual map retrieval set respectively to quantify the direction change of the roads. Within the preset threshold range, the similarity of the forward direction between the road in the hand-drawn sketch and the road in the actual map is calculated.

[0017] Step 5: Comprehensive Comparison of Similarity between Roads

[0018] In this step, the foregoing features are comprehensively considered for the final similarity measurement. The present invention assigns weights to matching factors such as shape distance, number of key points, and forward direction, and uses the weighted average method to calculate the comprehensive similarity between the road in the hand-drawn sketch and the road in the actual map. Based on the sum of the calculated comprehensive similarities, a sorted list of the road set in the matching actual map is generated in descending order of the similarity values.

[0019] Technical Features and Beneficial Effects of the Present Invention:

[0020] (1) The matching technology proposed by the present invention has two main features, namely, expanding the service scope of spatial data query and enhancing the user experience. This technology adopts a hand-drawn sketch main road extraction algorithm, which can accurately identify and extract the main road network in the geographical scene. In the display of matching results, this technology introduces a user preference setting mechanism to achieve dynamic adjustment of matching results. The system can flexibly adjust the matching strategy according to the personalized needs and matching tendencies of users, so as to generate query results that better meet the user's expectations. This dual optimization method combining main road extraction and user preferences greatly improves the accuracy and adaptability of spatial data matching, provides more accurate and personalized spatial data services for different users, and significantly enhances the robustness and service quality of this technology in practical applications.

[0021] (2) The road network matching method in hand-drawn sketches based on local shape proposed by the present invention has brought significant improvements to spatial data query services. Taking hand-drawn sketches as the input of query data is closer to the spatial cognition of ordinary users, greatly reducing the technical usage threshold, being able to meet the query needs of ordinary users for specific geographical areas, and improving the accessibility and user-friendliness of the query service; different from the traditional query methods that only rely on geographical attributes (such as names, types, etc.), the present invention realizes a non-semantic spatial data query mode, which can significantly enhance the reliability and flexibility of the query service. The present invention is not only applicable to traditional fields such as tourism planning and urban management, but also can be extended to multiple industries such as intelligent transportation and emergency response, and has broad industrial application prospects.

[0022] (3) The road network matching method based on local shape proposed by the present invention has certain universality and portability. The proposed technical solution is not only applicable to spatial data query services based on hand-drawn sketches, but also can be applied to diverse scenarios such as multi-scale geographical data integration and multi-resolution geographical data linkage update. Through appropriate improvement, the present invention can be organically combined with the building group matching technology, comprehensively considering the spatial correlation between the road network and the building group, so as to achieve accurate matching and query of multiple types of geographical scenes. This diversified application ability can significantly improve the accuracy and applicable scope of spatial data query services, meet the diverse needs under different application scenarios, and further expand its application scope.

[0023] By providing more intuitive and accurate spatial data query services for ordinary users, the present invention is expected to play an important role in improving the efficiency of public services and promoting the construction of smart cities. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 is a schematic diagram of the outline of the road network matching method in hand-drawn sketches based on local shape proposed by the present invention;

[0025] Figure 2It is the specific step flow chart of the hand-drawn sketch road network matching method based on local shape proposed by the present invention;

[0026] Figure 3 It is the calculation diagram of the simple road network and main road features in the hand-drawn sketch proposed by the present invention;

[0027] Figure 4 It is the schematic diagram of the road key point features in the hand-drawn sketch proposed by the present invention;

[0028] Figure 5 It is the schematic diagram of the road forward direction features in the hand-drawn sketch proposed by the present invention;

[0029] Figure 6 It is the hand-drawn sketch data set of the experimental area collected by the present invention;

[0030] Figure 7 It is the automatic extraction result diagram of the main roads in the hand-drawn sketch implemented by the present invention;

[0031] Figure 8 It is the road network matching result diagram in the hand-drawn sketch based on local shape implemented by the present invention. Detailed implementation manner

[0032] Next, in combination with the drawings in the embodiments of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts shall fall within the protection scope of the present invention.

[0033] In order to further understand the content, features and effects of the present invention, the following embodiments are listed and described in detail in conjunction with the drawings as follows.

[0034] Next, in combination with Figure 3-8 A detailed description will be given to a hand-drawn sketch road matching method based on local shape of the present invention. The main variables and parameters in the process are shown in Table 1.

[0035] Table 1 Functions of main variables and parameters

[0036]

[0037] The first step, calculation of the main road features of the road network in the hand-drawn sketch ( Figure 3 ), which includes the calculation of two factors, and the implementation steps are as follows:

[0038] (1.a) Calculation of the degree centrality (DC) of a single road.

[0039] In road network analysis, the centrality of a single road is a key indicator quantifying the degree of connection between the road and other roads in the network. It measures the importance and centrality of the road in the overall network structure by calculating the number of other roads directly connected to a specific road. Specifically, the centrality of a certain road is positively correlated with the number of its connections in the road network. In other words, if a road is connected to more other roads in the network, its centrality is higher. The calculation method of centrality involves in-depth analysis of the topological structure of the road network. Specifically, by counting the number of other road nodes directly connected to the nodes of the target road (usually the intersection points or endpoints of the road), the centrality value of the road can be obtained.

[0040] The specific calculation method is as follows:

[0041]

[0042] In the formula: DC i represents the connection quantity value between the i-th road and other roads, represents that the topological relationship between the i-th road and the j-th road is connected, that is, the two roads are connected, then the Num value is incremented by 1.

[0043] (1.b) Calculation of the betweenness centrality (BC) of a single road. In road network analysis, the betweenness centrality of a single road is an important topological indicator used to quantify the degree to which a specific road serves as a mediator or bridge connecting other roads in the entire network. It measures its importance in the overall structure by evaluating the frequency of its appearance in all the shortest paths in the network. Specifically, the betweenness centrality of a road is positively correlated with the frequency of its appearance in the shortest paths in the network. If a road appears more frequently in the shortest paths in the network, its betweenness centrality is greater. The calculation of betweenness centrality involves a complex network analysis process. The present invention adopts the following method: First, calculate the shortest paths from one endpoint of the target road to all other road endpoints in the network. Then, count the frequency of the appearance of this road in these shortest paths, and use this frequency as the betweenness centrality indicator of the road in a specific scenario.

[0044] The specific calculation method is as follows:

[0045] BC i = Freq(ShortDis (i，0..n) ) (2)

[0046] In the formula: BC i represents the number of times the i-th road appears in all the shortest paths in the scenario, and ShortDis (i，0..n) represents the shortest path between the i-th road and other road nodes (from node 0 to node n).

[0047] The present invention extracts main roads from a hand-drawn sketch road network using two factors, betweenness centrality and closeness centrality, which means that roads with higher betweenness centrality and closeness centrality can be regarded as main roads in the road network.

[0048] In the second step, the shape differences between roads are compared. The shape distance (SD) is used to calculate the shape differences between the main roads extracted from the hand-drawn sketch and the roads in the metric map.

[0049] (2.a) Resampling of the feature point set. The present invention first resamples the point clouds of two corresponding roads to n points, then calculates the Euclidean distance between each pair of matching points, and takes the sum of the distances as the dissimilarity between the two point clouds. During the calculation, unordered points are used to represent the shape, ignoring the node numbers, order, and direction.

[0050] The specific calculation method is as follows:

[0051]

[0052] In the formula: C and T are two point sets; dis (C,T) refers to the distance between the two point sets calculated according to this formula; Cx i and Tx i are the x - coordinate values of the i - th matching point in the two point sets; Cy i and Ty i are the y - coordinate values of the i - th matching point in the two point sets.

[0053] (2.b) Calculation of shape differences. Based on the above calculation results, the shape distance is regarded as the shape difference SD between the two roads.

[0054] The specific calculation method is as follows:

[0055]

[0056] In the formula: SD (S,O) represents the shape distance between the road in the hand-drawn sketch and the road in OSM; R s represents the road from the scene sketch; R o represents the road from OSM; represents the shape distance between the road in the hand-drawn sketch and the OSM road calculated according to formula (3).

[0057] In the third step, the comparison of the key point features of the roads ( Figure 4 ). The key points refer to the points on a single road where the direction angle exceeds a certain threshold among all nodes. The number of critical points (NCP) can reflect the morphological characteristics and direction changes of the road.

[0058] (3.a) The method for calculating the direction angle is as follows:

[0059]

[0060] In the formula: X 1 and Y 1 respectively represent the x and y coordinate values of point 1, X 2 and Y 2 respectively represent the x and y coordinate values of point 2.

[0061] The threshold depends on the characteristics of the road itself. The method for obtaining its value is to first calculate the direction angles of all adjacent nodes on the road, then extract the direction angle values with relatively large deviations among the adjacent direction angles, and finally take the smallest value among all the extracted direction angles as the threshold.

[0062] (3.b) The comparison of the characteristics of key points (NCP) refers to calculating the similarity of two roads by comparing the number of key points on the roads. Only when the NCP of a road in OSM is equal to or exceeds the NCP of the comparable road in the sketch can these two roads be similar.

[0063] The specific calculation method is as follows:

[0064]

[0065] Among them, SimNCP (S,O) represents the NCP similarity between two roads; NCP o represents the NCP of a certain road in the actual map OSM; NCP s represents the NCP of a certain road in the sketch. If NCP o ≥NCP s , it is considered that the similarity of the two roads in terms of key point characteristics is true, otherwise it is false.

[0066] Fourth step, calculation of the road forward direction characteristics ( Figure 5 ). The forward direction (CirculationDirection, CD) of a single road is a key geometric feature that reflects the spatial orientation of the road and helps to identify and quantify the direction changes between adjacent road segments within the road.

[0067] (4.a) Calculation of the change value of adjacent road segments of a single road.

[0068] The specific calculation method is:

[0069] v = (X p0 - X p1 ) * (Y p1 - Y p2)-(X p1 -X p2 )*(Y p0 -Y p1 ) (7)

[0070] where p 0 is the starting point of the first road segment; p 1 is the ending point of the first road segment and also the starting point of the second road segment; p 2 is the ending point of the second road segment.

[0071] (4.b) Direction change of adjacent segments of a single road. Based on the change value of adjacent segments, the direction change of two adjacent road segments is obtained, where "l" represents the left direction and "r" represents the right direction.

[0072] The specific calculation method is as follows:

[0073]

[0074] where D represents the direction change of two adjacent road segments. The value of 0 indicates no direction change of adjacent road segments, r indicates that the road segment turns to the right, and l indicates that the road segment turns to the left. If the direction change value v of adjacent road segments is equal to 0, the D value of these two road segments is 0; if the direction change value v of adjacent road segments is less than 0, the D value of these two road segments is r; otherwise, the D value of these two road segments is l.

[0075] (4.c) Set of direction changes of a single road segment. The CD of a single road is the set of direction change values of all adjacent line segments on the single road.

[0076] The specific calculation method is as follows:

[0077] CD = (D 0 , D 1 , …, D n-1 ) (9)

[0078] where CD represents the set of direction changes of all adjacent line segments on a single road. D 0 represents the direction change value between the first road segment and the second road segment, D 1 represents the direction change value between the second road segment and the third road segment, D n-1 represents the direction change value between the (n - 1)th road segment and the nth road segment.

[0079] (4.d) Calculation of forward direction feature similarity. If the CD of one road is the same as the CD of another road, these two roads are considered similar. If the CD of one road is a part of the CD of another road, these two roads are regarded as partially similar.

[0080] The specific calculation method is as follows:

[0081]

[0082] Among them, SimCD (s,o) represents the similarity of two roads in the sketch and OSM in terms of CD features; CD s represents the CD of the road in the sketch; CD o represents the CD of the road in the actual map OSM.

[0083] Step 5, comprehensive comparison of the similarity between roads. The similarity between two roads is the sum of the similarities considering features such as shape distance, the number of key points, and forward direction.

[0084] The specific calculation method is as follows:

[0085] Sim R = w 1 *(NUM - SD)+ w 2 *Sim NCP + w 3 *Sim CD (11)

[0086] Among them, Sim R represents the total similarity between two roads; NUM is the number of sampling points involved in the shape distance feature comparison; SD represents the shape distance between two roads; w 1 represents the weight of the similarity in SD; Sim NCP represents the similarity of NCP between two roads; w 2 represents the weight given to Sim NCP ; Sim CD represents the similarity of CD between two roads; w 3 represents the weight given to Sim CD . Assume that w 1 + w 2 + w 3 = 1.0. In addition, the larger the value of Sim R , the more similar the two roads are. In the invention, to make all features in formula () have the same importance, all weights have the same value: 1 / 3. If in other studies, more emphasis needs to be placed on a certain feature, these weights can be adjusted. For example, if the road shape feature in the scene is more special, the weight of the factor key point number feature can be increased.

[0087] Next, a partial area of the north campus of Nanjing Normal University Xianlin Campus is selected as a sample area to further illustrate this embodiment.

[0088] I. Summary of the test area

[0089] The reasons for selecting some areas in the northern part of Xianlin Campus of Nanjing Normal University as the experimental area are as follows: (1) Integrity of geographical elements: This scene includes various geographical objects, such as teaching buildings, playgrounds, dormitories, roads, bridges, etc. These objects can be drawn in various styles such as polygons, ellipses, rectangles, line segments, etc. in the sketch. (2) Complexity of spatial relationships: There are various spatial relationships in this scene, such as the intersection relationship between roads, the adjacent relationship between roads and buildings, the separation relationship between buildings and buildings, and the intersection relationship between roads and bridges, etc.

[0090] II. Experimental data collection

[0091] When drawing the scene sketch, it is based on the volunteers' own perception of the scene, but there are some basic requirements: Volunteers can use any shape to draw and express geographical objects, including shapes such as polygons, ellipses, rectangles, line segments, etc., and can add annotations on the scene sketch, such as the type or name of the geographical object; The special requirement for drawing the sketch is that all volunteers must draw the scene sketch completely from memory, including the areas and roads they are familiar with, without the help of any electronic devices such as Baidu Maps or other data sources. Figure 6 Shows the scene sketches drawn by 11 volunteers collected in this experiment, numbered S1, S2, S3, S4, S5, S6, S7, S8, S9, S10, and S11 respectively. Among them, sketches S10 and S11 were drawn by volunteers without a geographical science background.

[0092] III. Experimental results

[0093] After completing the drawing and collection of the scene sketch, the present invention digitizes the scene sketch for further analysis and comparison of geographical object features.

[0094] The first step: Extraction of the main roads in the road network of the hand-drawn sketch ( Figure 7 ), which involves the calculation of two factors, and the implementation steps are as follows:

[0095] (1.a) Calculation of the centrality of a single road. The calculation method of centrality is specifically to obtain the centrality value of the road by counting the number of other road nodes directly connected to the nodes of the target road (usually the intersection points or endpoints of the road).

[0096] (1.b) Calculation of the betweenness of a single road. First, calculate the shortest paths from one endpoint of the target road to all other road endpoints in the network. Then, count the frequency of the appearance of this road in these shortest paths, and use this frequency as the betweenness index of this road in a specific scene.

[0097] The present invention extracts main roads from a hand-drawn sketch road network using two factors, centrality and betweenness, which means that roads with higher centrality and betweenness can be regarded as main roads in the road network.

[0098] In the second step, the shape differences between roads are compared. The shape distance is used to calculate the shape differences between the main roads extracted from the hand-drawn sketch and the roads in the metric map.

[0099] (2.a) Resampling of the feature point set. The present invention first resamples the point clouds of two corresponding roads to n points, then calculates the Euclidean distance between each pair of matching points, and takes the sum of the distances as the dissimilarity between the two point clouds. During the calculation, unordered points are used to represent the shape, ignoring the node numbers, order, and direction.

[0100] The specific calculation method is as follows:

[0101]

[0102] In the formula: C and T are two point sets; dis (C,T) refers to the distance between the two point sets calculated according to this formula; Cx i and Tx i are the x-coordinate values of the i-th matching point in the two point sets; Cy i and Ty i are the y-coordinate values of the i-th matching point in the two point sets.

[0103] (2.b) Calculation of shape differences. Based on the above calculation results, the shape distance is regarded as the shape difference SD between the two roads.

[0104] In the third step, the feature comparison of road key points is carried out. Key points refer to the points on a single road whose direction angles exceed a certain threshold among all nodes. The threshold depends on the characteristics of the road itself. The method of obtaining the value is to first calculate the direction angles of all adjacent nodes on the road, then extract the direction angle values with relatively large deviations among the adjacent direction angles, and finally take the smallest value among all the extracted direction angles as the threshold.

[0105] (3.b) Feature comparison of key points (NCP) refers to comparing the similarity of two roads by calculating the number of key points on the roads. Only when the NCP of a road in OSM is equal to or exceeds the NCP of the compared road in the sketch can these two roads be similar.

[0106] In the fourth step, the feature comparison of the road forward direction is carried out. The forward direction (Circulation Direction, CD) of a single road is a key geometric feature, which reflects the spatial orientation of the road and helps to identify and quantify the direction changes between adjacent road segments inside the road.

[0107] (4.a) Calculation of the change value of adjacent road segments of a single road.

[0108] (4.b) Change direction of adjacent road segments of a single road. Based on the change value of adjacent road segments, the direction change of two adjacent road segments is obtained, where "l" represents the left direction and "r" represents the right direction.

[0109] (4.c) Set of direction changes of road segments of a single road. The CD of a single road is the set of direction change values of all adjacent line segments on the single road.

[0110] (4.d) Calculation of the similarity of the forward direction feature. If the CD of one road is the same as the CD of another road, then these two roads are considered similar. If the CD of one road is a part of the CD of another road, then these two roads are regarded as partially similar.

[0111] Step 5, Comprehensive comparison of the similarity between roads. The similarity between two roads is the sum of the similarities considering features such as shape distance, number of key points, and forward direction.

[0112] The specific calculation method is as follows:

[0113] Sim R = w 1 *(NUM - SD)+ w 2 *Sim NCP + w 3 *Sim CD (11)

[0114] where Sim R represents the total similarity between two roads; NUM is the number of sampling points involved in the shape distance feature comparison; SD represents the shape distance between two roads; w 1 represents the weight of the similarity in SD; Sim NCP represents the similarity of NCP between two roads; w 2 represents the weight given to Sim NCP ; Sim CD represents the similarity of CD between two roads; w 3 represents the weight given to Sim CD . Assume that w 1 + w 2 + w 3 = 1.0. In addition, the larger the value of Sim R , the more similar the two roads are. In the invention, to make all features in formula () have the same importance, all weights have the same value: 1 / 3.

[0115] Figure 8Shows the road network matching results in this example, where:

[0116] (1) Each row shows the main roads extracted from each sketch, the matching parameters, and the matching results of OSM;

[0117] (2) The first column shows the main roads extracted from each sketch and lists the matching parameters, including: MaxMN represents the maximum number of main roads with the first matching priority participating in the composite road matching in OSM; ThresholdCon is the threshold used to calculate the CD feature;

[0118] (3) The remaining columns show the matching results with higher similarity retrieved from OSM and the similarity values. The greater the similarity, the more similar the road network between the scene sketch and OSM.

[0119] (4) The similarity of the road network is calculated between the scene sketch and OSM using formula (11). The weights in each similarity calculation formula are averaged: the values of w1, w2, and w3 in the formula are all set to 0.33.

Claims

1. A road network matching method in hand-drawn sketches based on local shapes, characterized in that The following steps are involved: Step 1: Extraction of main roads from hand-drawn sketch road network; Step 2: Compare the shape differences between roads; Step 3: Comparison of the number and characteristics of key road points; Step 4: Compare the characteristics of the road direction; Step 5: Comprehensive comparison of similarities between roads.

2. A road network matching method in a hand-drawn sketch based on local shapes according to claim 1, characterized in that: The specific process of step 1 is as follows: Step 11, calculate the degree centrality (DC) of a single road by calculating the number of other roads connected to a certain road. The more other roads connected to a certain road, the higher the degree centrality of the road. Where: DC i Indicates the number of connections between the i-th road and other roads, Indicates that the topological relationship between the i-th road and the j-th road is connected, that is, the two roads are connected, then the Num value is ++, Step 12, Betweenness Centrality (BC) calculation of a single road, by evaluating the frequency of the road appearing in all shortest paths in the network to calculate its betweenness in the entire geographical scenario, BC i = Freq(ShortDis (i，0..n) ) (2) Where: BC i Indicates the number of times the i-th road appears in all the shortest paths in the scene, ShortDis (i，0..n) represents the shortest path between the ith road and other road nodes (from 0 to n nodes), The present invention uses two factors, centrality and betweenness, to extract main roads from a hand-drawn sketch road network, which means that roads with higher centrality and betweenness can be regarded as main roads in the road network.

3. The road network matching method in a hand-drawn sketch based on local shapes according to claim 1 is characterized in that: The specific process of step 2 is as follows: Step 21, resampling of feature point set. The present invention resamples the point clouds of two corresponding roads to n points, calculates the Euclidean distance between each matching point, and uses the sum of the distances as the dissimilarity between the two point clouds. In the calculation process, unordered points are used to represent the shape, and the number, order and direction of the nodes are ignored. The specific calculation method is: Where: C and T are two point sets; dis (C,T) Refers to the distance between two point sets calculated according to this formula; Cx i and Tx i is the x-coordinate value of the i-th matching point in the two point sets; Cy i and Ty i is the y coordinate value of the i-th matching point in the two point sets, Step 22, shape difference calculation, based on the above calculation results, the shape distance is regarded as the shape difference SD between the two roads, The specific calculation method is: In the formula: SD (S,O) Represents the shape distance between the hand-drawn sketch and the road in OSM; R s represents the road from the scene sketch; R o Represents roads from OSM; It represents the shape distance between the road in the hand-drawn sketch and the OSM road calculated according to formula (3).

4. The road network matching method in a hand-drawn sketch based on local shapes according to claim 1 is characterized in that: The specific process of step 3 is as follows: Step 31, comparing the key point features of the road. The key point refers to the point whose direction angle exceeds a certain threshold among all nodes on a single road. The number of critical points (NCP) can reflect the morphological characteristics and direction changes of the road. The direction angle calculation method is as follows: Where: X1 and Y1 represent the x and y coordinates of point 1, respectively; X2 and Y2 represent the x and y coordinates of point 2, respectively. The threshold depends on the characteristics of the road itself. The method for obtaining the value is to first calculate the direction angles of all adjacent nodes on the road, then extract the direction angle values ​​with large deviations among the adjacent direction angles, and finally take the smallest value of all the extracted direction angles as the threshold. Step 32, comparing key points (NCP) features, is to compare the similarity of two roads by calculating the number of key points on the roads. Only when the NCP of a road in OSM is equal to or exceeds the NCP of the compared road in the sketch, the two roads have possible similarity. The specific calculation method is as follows: Among them, SimNCP (S,O) Indicates the NCP similarity between two roads; NCP o Represents the NCP of a road in the actual map OSM; NCP s Represents the NCP of a road in the sketch. If NCP o ≥NCP s , then the similarity between the two roads in terms of key point features is considered to be true, otherwise it is considered to be false.

5. The road network matching method in a hand-drawn sketch based on local shapes according to claim 1 is characterized in that: The specific process of step 4 is as follows: Step 41, calculation of the change value of adjacent sections of a single road, the specific calculation method is: v=(X p0 -X p1 )*(AND p1 -AND p2 )-(X p1 -X p2 )*(AND p0 -AND p1 ) (7) Among them, p0 is the starting point of the first section; p1 is the end point of the first section and the starting point of the second section; p2 is the end point of the second section. Step 42, the change direction of adjacent sections of a single road, based on the change value of the adjacent section, obtain the direction change of two adjacent sections, with "l" representing the left direction and "r" representing the right direction, The specific calculation method is: Where D represents the direction of change of two adjacent road sections. A value of 0 indicates that there is no direction change of the adjacent road sections. r indicates that the road section turns right, and l indicates that the road section turns left. If the direction change value v of the adjacent road sections is equal to 0, the D values ​​of the two road sections are 0; if the direction change value v of the adjacent road sections is less than 0, the D values ​​of the two road sections are r; otherwise, the D values ​​of the two road sections are l. Step 43, a set of direction changes of a single road segment. The CD of a single road is a set of direction change values ​​of all adjacent line segments on the single road. The specific calculation method is: CD=(D0,D1,…,D n-1 ) (9) Among them, CD represents the set of direction changes of all adjacent line segments on a single road, D0 represents the direction change value between the first segment and the second segment, D1 represents the direction change value between the second segment and the third segment, and D n-1 Indicates the direction change value of the n-1th road segment and the nth road segment. Step 44, forward direction feature similarity calculation, if the CD of one road is the same as the CD of another road, then the two roads are considered similar, if the CD of one road is a part of the CD of another road, then the two roads are considered partially similar, The specific calculation method is: Among them, SimCD (s,o) Indicates the similarity of CD features between two roads in the sketch and OSM; CD s CD representing the road in the sketch; CD o CD representing the road in the actual map OSM.

6. The method for matching road networks in hand-drawn sketches based on local shapes according to claim 1, characterized in that: The specific process of step 5 is as follows: The comprehensive comparison of similarity between roads refers to the similarity between two roads, which is the sum of similarities that comprehensively considers features such as shape distance, number of key points, and direction of travel. The specific calculation method is: Sim R =w1*(NUM-SD)+w2*Sim NCP +w3*Sim CD (11) Among them, Sim R represents the total similarity between two roads; NUM is the number of sampling points involved in the shape distance feature comparison; SD represents the shape distance between two roads; w1 represents the weight of the similarity in SD; Sim NCP represents the similarity of NCP between two roads; w2 represents the value assigned to Sim NCP Weight of Sim CD Indicates the similarity of CD between two roads; w3 indicates the value given to Sim CD weights, assuming w1+w2+w3=1.0, in addition, Sim R The larger the value, the more similar the two roads are. If you need to focus more on a certain feature in other studies, you can adjust these weights. For example, if the road morphology in the scene is more special, you can increase the weight of the factor key point quantity feature.

Images