A road intersection positioning method based on triangle matching strategy

Abstract

The present invention discloses a method for locating road intersections based on a triangle matching strategy, which belongs to the field of image processing technology. The method comprises: inputting image data to be matched; extracting all intersections from the image data and counting intersection point sets; counting the triangular topological information description between each point in the intersection point set; counting the two-dimensional distribution of all triangle internal angles of the triangle topological information of the intersection of the data to be matched in the basic intersection topology information library; combining matching constraints to obtain matching triangles for the data to be matched; and obtaining intersection positioning results for the data to be matched by matching the intersection matching results corresponding to each vertex of the triangle. The present invention makes full use of the strong uniqueness and high robustness of the topological relationship between road intersections, and utilizes the matching principle of stable triangle structure to ultimately achieve accurate and reliable road intersection positioning, which can solve the problem of difficult road extraction and inability to quickly and accurately match.

Description

Technical Field

[0001] The present invention belongs to the field of geographic information processing, and in particular relates to a road intersection positioning method based on a triangle matching strategy. Background Art

[0002] Currently, scholars at home and abroad have conducted in-depth research on matching algorithms for road networks, primarily focusing on geometry-based matching, topological feature-based matching, and hybrid feature-based matching. Among these, geometric feature-based matching algorithms primarily use the distance between nodes and polylines to represent the similarity of matching objects. Topological feature-based matching primarily classifies roads based on topological relationships and divides them into matching and non-matching layers according to road type, achieving hierarchical matching of road networks. Hybrid feature-based matching algorithms effectively combine geometric, semantic, topological, node, and arc matching by selecting similarity metrics such as node, direction, and distance to establish a comprehensive matching strategy. Furthermore, some researchers have proposed matching algorithms based on both geometric and topological features, using distance as an evaluation metric and the results of a road network arc breakpoint neighbor analysis as a matching threshold to determine whether matching objects are identical entities. However, these matching algorithms require human intervention and control and are computationally intensive.

[0003] As a key element of the road network, intersections form a topological structure with roads that is highly recognizable in a map network, providing an important basis for matching multi-source data. Therefore, the main research direction of this patent is to study how to match the topological structure of intersections to complete road positioning. Summary of the Invention

[0004] In view of this, the present invention provides a road intersection positioning method based on a triangle matching strategy, which fully utilizes the strong uniqueness and high robustness of the topological relationship between road intersections, and uses the matching principle of stable triangle structures to ultimately achieve accurate and reliable road intersection positioning.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is:

[0006] A road intersection positioning method based on a triangle matching strategy comprises the following steps:

[0007] Step 1: Use the road vectors in OSM data as the basic road network data, extract all intersections, and form the intersection point set R i ,Each point in the intersection point set is used to represent the coordinates of the center point of an intersection;

[0008] Step 2: Input the image data to be matched, extract all the intersections in the image data to be matched, and form the intersection point set R j ;

[0009] Step 3: Set the constraint range Tdis according to the size of the image to be matched. i For each point in R i The points whose distance to the point does not exceed Tdis form a triangle with the point, and all different triangles form R i Triangle description library Will Each triangle in is represented as a two-dimensional coordinate, and we get The two-dimensional distribution Ki (x,y) ;

[0010] Step 4: Count the intersection point set R j All triangles in R j Triangle description library Will Each triangle in is represented as a two-dimensional coordinate, and we get The two-dimensional distribution Kj (x,y) ;

[0011] Step 5, according to Kj (x,y) The coordinates of each coordinate point in Ki (x,y) Find the closest coordinate point as the approximate matching point, and define the candidate area with each matching point as the center and the preset length as the radius;

[0012] Step 6: Find Kj in each candidate area based on vertex constraints and edge length constraints (x,y) The midpoint is Ki (x,y) The exact matching point in Kj is obtained (x,y) with Ki (x,y) Exact matching relationship;

[0013] Step 7, according to Kj (x,y) with Ki (x,y) The exact matching relationship of R j With R i The matching relationship is then obtained to obtain the positioning results of each intersection in the image data to be matched.

[0014] Furthermore, in step 3, a constraint range Tdis is set according to the size of the image to be matched. Specifically, Tdis is set to the farthest distance between two intersections in the image to be matched.

[0015] Furthermore, in step 3 and step 4, the triangle is represented as a two-dimensional coordinate in the following manner: the maximum inner angle value and the minimum inner angle value of the triangle are used as the two-dimensional coordinates.

[0016] Furthermore, the specific method of step 6 is:

[0017] Step 601: For a candidate area, count all the vertices of all triangles in it. If a vertex is in If the frequency p of the vertex does not satisfy p ≥ n-1, where n represents the number of intersections in the image to be matched, the triangle containing the vertex will be deleted from the candidate area;

[0018] Step 602: For a candidate area, count the side lengths of all triangles in it and select the one that matches the candidate area. The two-dimensional coordinates of the triangle whose side length matches the triangle to be matched are the exact matching points.

[0019] The present invention has the following beneficial effects:

[0020] (1) The present invention proposes a road intersection positioning method based on a triangle matching strategy, which does not rely on the degree of refinement of road extraction and completes the topological structure description between intersections through the relationship between intersections.

[0021] (2) The present invention utilizes the principle of triangle similarity and completes matching according to the topological relationship distribution between intersections, which can reduce the accumulation of extraction accuracy errors of intersections.

[0022] (3) The method of the present invention can be used to describe the topological structure of road intersections, providing effective technical support for UAV positioning. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 Schematic diagram of the process of a road intersection positioning method based on a triangle matching strategy in an embodiment of the present invention.

[0024] Figure 2 Schematic diagram of the construction of a triangular topological structure at an intersection in an embodiment of the present invention.

[0025] Figure 3 Schematic diagram of two-dimensional distribution statistics of triangle topology information within different constraint ranges in an embodiment of the present invention.

[0026] Figure 4 Schematic diagram of the distribution of interior angles of triangles to be matched in a matching triangle library according to an embodiment of the present invention. DETAILED DESCRIPTION

[0027] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0028] A method for locating a road intersection based on a triangle matching strategy comprises: inputting image data to be matched; extracting all intersections from the image data, and counting intersection point sets, wherein the point sets include: the position coordinates of the intersections and corresponding points in the image; counting the triangle topology information descriptions between the points in the intersection point sets; counting the two-dimensional distribution of all triangle interior angles of the triangle topology information of the intersection of the data to be matched in a basic intersection topology information database; combining matching constraints to obtain matching triangles for the data to be matched; and obtaining intersection positioning results for the data to be matched by matching the intersection matching results corresponding to the vertices of the triangles.

[0029] like Figure 1 As shown, the method specifically includes the following steps:

[0030] Step 1: Input the image data to be matched;

[0031] Step 2: Extract all intersections from the image data and count the intersection point set R j (j=1,2...n), the point set content includes: intersection id', the position coordinates x', y' of the corresponding point in the image;

[0032] Step 3: Count the intersection point set R j Description of the triangle topology information between the points

[0033] Step 4: Count the triangle topology information of the intersection of the data to be matched The two-dimensional distribution of all triangle interior angles Kj (x,y) In the basic intersection topology information database Ki (x,y) distribution in;

[0034] Step 5: Combine the matching constraints to obtain the matching triangles of the data to be matched;

[0035] Step 6: Match the corresponding vertices of the triangles in R i The corresponding intersection matching result in is the intersection positioning result of the data to be matched.

[0036] The construction process of the basic intersection topology information database is as follows:

[0037] First, the road vectors in OSM data are used as the basic road network data to extract the road intersection point set R i , build a triangle description library between intersection points within the constraint range Tdis For all triangle topology information Perform two-dimensional statistics of the interior angles, and record the statistical results as Ki (x,y) Among them, the intersection point set R iThe content includes: intersection id, location coordinates x, location coordinates y, and the triangle description library construction range Tdis (Tdis ≥ L) is set according to the maximum size L of the image to be matched. The triangle description library between any intersection points is constructed within this range. The content includes: triangle id, vertex 1 id and position coordinates: p1_id, p1_x, p1_y, vertex 2 id and position coordinates: p2_id, p2_x, p2_y, vertex 3 id and position coordinates: p3_id, p3_x, p3_y, the maximum and minimum internal angles of the triangle: anglemax, anglemin, and the three side lengths: edge1, edge2, edge3. The principle is as follows Figure 2 As shown. For the triangle topology information within the constraint range d The maximum and minimum interior angles are statistically analyzed in two dimensions, as shown in the following diagram: Figure 3 The statistical results are recorded as:

[0038] Ki (x,y) =Ki (x＝anglemax,y＝anglemin) (i corresponds to triangle id)

[0039] That is Ki (x,y) Can represent the corresponding id triangle.

[0040] The specific implementation of step 3 is:

[0041] intersection point set R j Description of the triangle topology information between the points Description method is the same as

[0042] The specific implementation of step 4 is as follows:

[0043] statistics Describe the two-dimensional distribution Kj of the interior angles of all triangles in the library (x,y) in Ki (x,y) The distribution in Figure 4 As shown. Set in Kj (x,y) The point in is the center and the radius threshold is r k The points falling within the buffer zone are considered similar to the center point of the buffer zone, i.e., candidate approximate matching triangles.

[0044] The specific implementation of the matching restriction conditions in step 5 is as follows:

[0045] If the frequency p of the vertex of the candidate approximate triangle satisfies p≥n-1 (n represents the number of intersections to be matched in the image), then the vertex belongs to the approximate matching point, and the triangles that do not include the vertex are removed to further screen out the approximate matching triangles; further, by The edge range in further constrains the selection of matching triangles.

[0046] This method makes full use of the strong uniqueness and high robustness of the topological relationship between road intersections, and uses the stable matching principle of triangular structure to ultimately achieve accurate and reliable road intersection positioning, which can solve the problems of difficult road extraction and inability to quickly and accurately match.

[0047] Here's a more specific example:

[0048] A road intersection positioning method based on a triangle matching strategy comprises the following steps:

[0049] Step 1: Input the image data to be matched; use a drone to take the image, the size of which is 3056 pixels * 2243 pixels (1400m * 1300m).

[0050] Step 2: Extract all intersections from the image data and count the intersection point set R j (j=1,2...n), the point set content includes: intersection id', the position coordinates x', y' of the corresponding point in the image;

[0051] The intersection extraction model uses the YOLOv5 model, and the training data uses the intersection class in NWPU-RESISC45 and 349 images with road intersections selected from LoveDAU;

[0052] Step 3: Count the intersection point set R j Description of the triangle topology information between the points

[0053] Step 4: Count the triangle topology information of the intersection of the data to be matched The two-dimensional distribution of all triangle interior angles Kj (x,y) In the basic intersection topology information database Ki (x,y) distribution in;

[0054] Specifically, a buffer is created, and the triangles that fall within the buffer are candidate similar triangles. The buffer size is set to the 10 points closest to the triangle to be matched.

[0055] Step 5: Combine the matching constraints to obtain the matching triangles of the data to be matched;

[0056] The matching restriction condition is to quickly filter out the occurrence frequency of candidate approximate triangle vertices falling within the buffer zone (at least L-1 times, where L is the number of intersection points).

[0057] Step 6: Match the corresponding vertices of the triangles in R iThe corresponding intersection matching result in is the intersection positioning result of the data to be matched.

[0058] In summary, the proposed method utilizes intersection information, a key element of the road network. This significantly reduces matching difficulty by not relying on road extraction results. Furthermore, it leverages the topological relationships between intersections to reduce the impact of accumulated errors in intersection extraction accuracy. Furthermore, the matching is robust, utilizing the triangle similarity principle. This method can provide an important basis for matching multi-source data.

Claims

1. A road intersection positioning method based on triangle matching strategy, characterized in that: The following steps are involved: Step 1: Use the road vectors in OSM data as the basic road network data, extract all intersections, and form the intersection point set R i ,Each point in the intersection point set is used to represent the coordinates of the center point of an intersection; Step 2: Input the image data to be matched, extract all the intersections in the image data to be matched, and form the intersection point set R j ; Step 3: Set the constraint range Tdis according to the size of the image to be matched. i For each point in R i The points whose distance to the point does not exceed Tdis form a triangle with the point, and all different triangles form R i Triangle description library Will Each triangle in is represented as a two-dimensional coordinate, and we get The two-dimensional distribution Ki (x,y) ; Step 4: Count the intersection point set R j All triangles in R j Triangle description library Will Each triangle in is represented as a two-dimensional coordinate, and we get The two-dimensional distribution Kj (x,y) ; Step 5, according to Kj (x,y) The coordinates of each coordinate point in Ki (x,y) Find the closest coordinate point as the approximate matching point, and define the candidate area with each matching point as the center and the preset length as the radius; Step 6: Find Kj in each candidate area based on vertex constraints and edge length constraints (x,y) The midpoint is Ki (x,y) The exact matching point in Kj is obtained (x,y) with Ki (x,y) Exact matching relationship; Step 7, according to Kj (x,y) with Ki (x,y) The exact matching relationship of R j With R i The matching relationship is then obtained to obtain the positioning results of each intersection in the image data to be matched.

2. A road intersection positioning method based on triangle matching strategy according to claim 1, characterized in that: In step 3, the constraint range Tdis is set according to the size of the image to be matched. Specifically, Tdis is set to the farthest distance between two intersections in the image to be matched.

3. The method for locating a road intersection based on a triangle matching strategy according to claim 1, characterized in that: In step 3 and step 4, the triangle is represented as a two-dimensional coordinate in the following manner: the maximum interior angle value and the minimum interior angle value of the triangle are used as the two-dimensional coordinates.

4. The method for locating a road intersection based on a triangle matching strategy according to claim 1, characterized in that: The specific method of step 6 is: Step 601: For a candidate area, count all the vertices of all triangles in it. If a vertex is in If the frequency p of the vertex does not satisfy p≥n-1, where n represents the number of intersections in the image to be matched, the triangle containing the vertex will be deleted from the candidate area; Step 602: For a candidate area, count the side lengths of all triangles in it and select the one that matches the candidate area. The two-dimensional coordinates of the triangle whose side length matches the triangle to be matched are the exact matching points.

Images