Method and system for identifying clusters with road lane changing characteristics

Abstract

The invention relates to the technical field of intelligent traffic, and discloses a method and system for identifying clusters with road lane change features, and the method comprises the following specific steps: S1, carrying out the clustering of lane lines of a target road to be subjected to lane change; s2, extracting geometric information of the cluster 1; s3, judging whether the obtained geometric information is in a list form or not; S4, obtaining long edges and short edges; s5, obtaining a long side point sequence list; s6, carrying out space coverage judgment to obtain a lane changing pre-selection list; s7, re-organizing lane line information; s8, judging whether the lane line information is greater than a threshold value or not; s9, obtaining a point sequence of geometric information of the lane line, and obtaining a long side and a short side of the point sequence; further obtaining a point sequence list; s10, further screening a point sequence list; and S11, obtaining all families with road lane changing characteristics. The method solves the problems that in the prior art, a prior sample is relied on, and a lane line with lane changing characteristics is difficult to accurately recognize, and has the advantages of being capable of improving clustering precision, flexible and easy to use.

Description

Technical Field

[0001] The present invention relates to the field of intelligent transportation technology, and more specifically, to a method and system for identifying a cluster with a road lane change feature. Background Art

[0002] At present, the method used for lane clustering in crowdsourced maps is DBSCAN density clustering, but this clustering method can easily lead to spreading, causing lanes that should not be clustered to cluster together, which is mainly reflected in places where buses change lanes or emergency lanes. In such places, lanes are stuck at intersections, the distance between non-sticky objects is close, and the attributes corresponding to lanes are basically the same, so it is difficult to separate lanes with current solutions.

[0003] To address this problem, there is a lane geometry information extraction method based on road segment trajectories. Trajectory clustering analysis is performed on trajectory subsets passing through the same road to obtain trajectory clusters belonging to the same lane. The trajectory geometry center of the trajectory subset within the same cluster is extracted in the form of a non-parametric curve, and the candidate lane center lines are obtained by fitting. A lane change lane line recognition method based on maximum clique search is established, an intersection relationship graph is established, and lane change lane lines are identified and eliminated through maximum clique analysis to obtain a set of non-intersecting lane center lines as basic lane center lines. The number of lanes at both ends of the road segment is reviewed through a Gaussian mixture model, and candidate lane center lines are extracted as new lane center lines based on the newly added Gaussian components and parameters.

[0004] However, the existing technology still relies on prior samples and is difficult to accurately identify lane lines with lane change characteristics. Therefore, how to invent a method for identifying clusters with road lane change characteristics that does not rely on prior samples to facilitate subsequent lane changes is an urgent problem to be solved in this technical field. Summary of the invention

[0005] In order to solve the problem that the existing technology relies on prior samples and is difficult to accurately identify lane lines with lane change characteristics, the present invention provides a method and system for identifying clusters with road lane change characteristics, which has the characteristics of improving clustering accuracy and being flexible and easy to use.

[0006] In order to achieve the above-mentioned purpose of the present invention, the technical scheme adopted is as follows:

[0007] A method for identifying a cluster having a lane change feature comprises the following specific steps:

[0008] S1: Clustering the lane lines of the target road to be changed;

[0009] S2: Extract the geometric information of one cluster from the result list after clustering is completed;

[0010] S3: Determine whether the obtained geometric information is in the form of a list. If so, convert the geometric information list into a unified coordinate system and execute step S4; otherwise, return to step S2;

[0011] S4: Obtain the long and short sides of the geometric information of the transformed cluster;

[0012] S5: Get the list of long edge point sequences near the long edge; get the starting point and end point of the short edge;

[0013] S6: Perform spatial coverage judgment on the long edge point sequence list according to the starting point and the end point of the short edge; if the spatial coverage judgment is not passed, return to step S2; if the long edge point sequence passes the spatial coverage judgment, obtain a lane change pre-selection list according to the long edge point sequence list;

[0014] S7: extracting a lane line information list from the result list; constructing an index tree according to the lane line information list; retrieving lane line information intersecting with the geometric information of the lane change pre-selection list through the index tree; and reorganizing the lane line information according to the timestamp;

[0015] S8: Determine whether the reorganized lane line information is greater than a threshold value. If it is less than the threshold value, return to step S7; if it is greater than the threshold value, execute step S9;

[0016] S9: Obtain a point sequence of geometric information of the lane line in the lane line information, and obtain its long side and short side; further obtain the azimuth of the short side, and perform a point-adding operation on the long side to obtain a point sequence list;

[0017] S10: further filter the point sequence list. If the point sequence list meets the filter condition, determine that the corresponding cluster has the lane change feature, and execute step S11; if the point sequence list does not meet the filter condition, return to step S9; if the point sequence lists of all lane lines in the lane line information do not meet the filter condition, return to step S2;

[0018] S11: Output the clusters with the road lane change feature; return to step S2 until all the clusters in the result list after clustering are traversed, thereby obtaining all the clusters with the road lane change feature.

[0019] Preferably, in step S2, the minimum width and step length of the gap in the geometric information list are also set.

[0020] Furthermore, in the step S3, the geometric information list is converted into a unified coordinate system, specifically: determining whether the input data is in the form of a list, organizing the geometric information list into data G in gdf format, and converting the data G into the WGS84 coordinate system.

[0021] Furthermore, in the step S4, the long side and short side of the geometric information of the converted cluster are obtained, specifically: the circumscribed rectangular frame of the converted cluster and the point sequence list of the circumscribed rectangular frame are obtained, and the two sides of the circumscribed rectangular frame are obtained; the lengths of the two sides are compared and judged to divide them into long sides and short sides.

[0022] Furthermore, in the step S5, a list of long edge point sequences near the long edge is obtained, specifically:

[0023] S501: Obtain a point sequence that is a certain distance from the long side; obtain a point sequence that is a step length as a starting point and a length of the long side as an end point, and gradually obtains a point sequence that is a step length as an interval from the long side; obtain a point sequence obtained by subtracting a certain distance from the length of the long side;

[0024] S502: Put all three point sequences into a list to obtain a long edge point sequence list; if the length of the obtained point sequence is less than the integer obtained by dividing the minimum width of the gap by the step length, it is not put into the list.

[0025] Furthermore, in step S5, the starting point and the end point of the short side are obtained, specifically:

[0026] Get the first and last points of the short side as the starting point and end point of the short side;

[0027] If the distance between the end point of the short side and the long side is smaller than the distance between the start point of the short side and the long side, the start point and the end point of the short side are swapped, otherwise they do not need to be swapped; the coordinate difference x, y of the start point and the end point of the short side is obtained simultaneously.

[0028] Furthermore, in step S6, the spatial coverage of the long edge point sequence list is judged according to the starting point and the end point of the short edge, and the specific steps are as follows:

[0029] S601: Convert the long edge point sequence list into gdf format;

[0030] S602: performing coordinate conversion on the long side point sequence list in the gdf format according to the coordinate difference x, y to obtain a complete long side point sequence list in the gdf format;

[0031] S603: Subtract the complete gdf format long edge point sequence list from the data G to obtain the inspection result; if the column of the inspection result is line type data, and the length of the inspection result is not less than the integer obtained by dividing the minimum width of the gap by the step size, then the long edge point sequence list corresponding to the inspection result passes the spatial coverage judgment; otherwise, the long edge point sequence list fails the spatial coverage judgment.

[0032] Furthermore, in step S9, the point filling operation is specifically as follows: the length of the long side of the point to be filled is divided by the number of points to be filled and rounded up, and after obtaining the position of each point to be filled from the starting point, the points are filled in sequence.

[0033] Furthermore, in the step S10, the point sequence list is further screened, and the specific steps are as follows:

[0034] S101: Obtain a perpendicular line L with a point in the point sequence list as the long side; the length of the perpendicular line is the length of the short side, and define count to count whether the perpendicular line intersects with the lane line;

[0035] S102: If the vertical line L intersects the lane line, count is increased by 1;

[0036] S103: looping steps S101 to S102 until all points in the point sequence list are traversed; if count is greater than or equal to the set threshold, it is determined that the screening principle is met; if count is still less than the set threshold, it is determined that the screening principle is not met.

[0037] A system for identifying clusters with road lane change characteristics, comprising a clustering module, an extraction module, a first screening module, a long and short edge module, a second screening module, a lane line module, a third screening module, a lane line point sequence module, a fourth screening module, and an output module;

[0038] The clustering module is used to cluster the lane lines of the target road to be changed;

[0039] The extraction module is used to extract the geometric information of one cluster from the result list after clustering is completed;

[0040] The first screening module is used to determine whether the obtained geometric information is in the form of a list, and if so, convert the geometric information list into a unified coordinate system;

[0041] The long and short edge module is used to obtain the long and short edges of the geometric information of the converted cluster; obtain a list of long edge point sequences near the long edge; and obtain the starting point and end point of the short edge;

[0042] The second screening module is used to perform spatial coverage judgment on the long side point sequence list according to the starting point and the end point of the short side; if the long side point sequence passes the spatial coverage judgment, a lane change pre-selection list is obtained according to the long side point sequence list;

[0043] The lane line module is used to extract the lane line information list in the result list; construct an index tree according to the lane line information list; retrieve the lane line information that intersects with the geometric information of the lane change pre-selection list through the index tree; and reorganize the lane line information according to the timestamp;

[0044] The third screening module is used to determine whether the reorganized lane line information is greater than a threshold;

[0045] The lane line point sequence module is used to obtain the point sequence of the geometric information of the lane line in the lane line information, and obtain its long side and short side; further obtain the azimuth of the short side, and perform a point addition operation on the long side to obtain a point sequence list;

[0046] The fourth screening module is used to further screen the point sequence list. If the point sequence list meets the screening condition, it is determined that the corresponding cluster has a lane change feature.

[0047] The output module is used to traverse all clusters in the result list after clustering is completed, so as to obtain and output all clusters with road change characteristics.

[0048] The beneficial effects of the present invention are as follows:

[0049] The present invention fully considers the collective characteristics of the lane lines of the target road to be changed, converts the geometric information list into a unified coordinate system, obtains the long and short sides of the geometric information of the converted cluster, and performs spatial coverage judgment on the long side point sequence list according to the starting point and end point of the short side to obtain a lane change pre-selection list; further filters the lane line information according to the lane change pre-selection list, and finally obtains all families with road lane change characteristics. After clustering is completed, the present invention identifies all families with road lane change characteristics and re-post-processes them to avoid affecting the subsequent module effects. Therefore, the present invention solves the problem of relying on prior samples and being difficult to accurately identify lane lines with lane change characteristics in the prior art, and has the characteristics of improving clustering accuracy and being flexible and easy to use. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 It is a schematic diagram of a process of identifying a cluster having a lane change feature according to the present invention;

[0051] Figure 2 It is a schematic diagram of a bus lane change for identifying a cluster having a road lane change characteristic according to the present invention;

[0052] Figure 3 It is a schematic diagram of an emergency lane change for identifying a cluster having a road lane change characteristic according to the present invention;

[0053] Figure 4 It is a schematic diagram of a complete method for identifying a cluster having a lane change feature in Embodiment 2 of the present invention.

[0054] Figure 5 Schematic diagram of a system for identifying clusters with lane change characteristics according to the present invention.

[0055] Figure 6It is a schematic diagram of a process of obtaining a list of long edge point sequences near a long edge for identifying a cluster with a road change feature according to the present invention.

[0056] Figure 7 It is a schematic diagram of a flow chart of spatial coverage judgment for identifying a cluster with a road change feature according to the present invention.

[0057] Figure 8 It is a schematic diagram of a process of further screening a point sequence list for identifying a cluster with a road change feature according to the present invention. DETAILED DESCRIPTION

[0058] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0059] Example 1

[0060] like Figure 1 As shown, a method for identifying a cluster with a lane change feature includes the following specific steps:

[0061] S1: Clustering the lane lines of the target road to be changed;

[0062] S2: Extract the geometric information of one cluster from the result list after clustering is completed;

[0063] S3: Determine whether the obtained geometric information is in the form of a list. If so, convert the geometric information list into a unified coordinate system and execute step S4; otherwise, return to step S2;

[0064] S4: Obtain the long and short sides of the geometric information of the transformed cluster;

[0065] S5: Get the list of long edge point sequences near the long edge; get the starting point and end point of the short edge;

[0066] S6: Perform spatial coverage judgment on the long edge point sequence list according to the starting point and the end point of the short edge; if the spatial coverage judgment is not passed, return to step S2; if the long edge point sequence passes the spatial coverage judgment, obtain a lane change pre-selection list according to the long edge point sequence list;

[0067] S7: extracting a lane line information list from the result list; constructing an index tree according to the lane line information list; retrieving lane line information intersecting with the geometric information of the lane change pre-selection list through the index tree; and reorganizing the lane line information according to the timestamp;

[0068] S8: Determine whether the reorganized lane line information is greater than a threshold value. If it is less than the threshold value, return to step S7; if it is greater than the threshold value, execute step S9;

[0069] S9: Obtain a point sequence of geometric information of the lane line in the lane line information, and obtain its long side and short side; further obtain the azimuth of the short side, and perform a point-adding operation on the long side to obtain a point sequence list;

[0070] S10: further filter the point sequence list. If the point sequence list meets the filter condition, determine that the corresponding cluster has the lane change feature, and execute step S11; if the point sequence list does not meet the filter condition, return to step S9; if the point sequence lists of all lane lines in the lane line information do not meet the filter condition, return to step S2;

[0071] S11: Output the clusters with the road lane change feature; return to step S2 until all the clusters in the result list after clustering are traversed, thereby obtaining all the clusters with the road lane change feature.

[0072] like Figure 2 As shown, in this embodiment, the vehicle to be changed lanes is a bus, and the lane line of the target road is a bus lane.

[0073] In a specific embodiment, in step S2, the minimum width of the gap, the maximum width of the gap, the judgment step length, and the minimum gap interval are also defined.

[0074] In a specific embodiment, the geometric information list is converted into a unified coordinate system, specifically: determining whether the input data is in the form of a list, organizing the geometric information list into data G in a gdf format, and converting the data G into the WGS84 coordinate system.

[0075] In a specific embodiment, in the step S4, the long side and short side of the geometric information of the converted cluster are obtained, specifically: the circumscribed rectangular frame of the converted cluster and the point sequence list of the circumscribed rectangular frame are obtained, and the two sides of the circumscribed rectangular frame are obtained; the lengths of the two sides are compared and judged to divide them into long sides and short sides.

[0076] In a specific embodiment, Figure 6 As shown, in the step S5, a list of long edge point sequences near the long edge is obtained, specifically:

[0077] S501: obtaining a point sequence having a certain distance from the long side; obtaining a point sequence having a step length as a starting point and a length of the long side as an end point, and gradually obtaining a point sequence having a step length as an interval from the long side; obtaining a point sequence obtained by subtracting a certain distance from the length of the long side;

[0078] S502: Put all three point sequences into a list to obtain a long edge point sequence list; if the length of the obtained point sequence is less than the integer obtained by dividing the minimum width of the gap by the step length, it is not put into the list.

[0079] In a specific embodiment, in step S5, the starting point and the end point of the short side are obtained as follows:

[0080] Get the first and last points of the short side as the starting point and end point of the short side;

[0081] If the distance between the end point of the short side and the long side is smaller than the distance between the start point of the short side and the long side, the start point and the end point of the short side are swapped, otherwise they do not need to be swapped; the coordinate difference x, y of the start point and the end point of the short side is obtained simultaneously.

[0082] In a specific embodiment, Figure 7 In step S6, the spatial coverage of the long edge point sequence list is judged according to the starting point and the end point of the short edge. The specific steps are as follows:

[0083] S601: Convert the long edge point sequence list into gdf format;

[0084] S602: performing coordinate conversion on the long side point sequence list in the gdf format according to the coordinate difference x, y to obtain a complete long side point sequence list in the gdf format;

[0085] S603: Subtract the complete gdf format long edge point sequence list from the data G to obtain the inspection result; if the column of the inspection result is line type data, and the length of the inspection result is not less than the integer obtained by dividing the minimum width of the gap by the step size, then the long edge point sequence list corresponding to the inspection result passes the spatial coverage judgment; otherwise, the long edge point sequence list fails the spatial coverage judgment.

[0086] In a specific embodiment, in step S9, the point filling operation is specifically as follows: the length of the long side of the point to be filled is divided by the number of points to be filled and rounded up, and after obtaining the position of each point to be filled from the starting point, the points are filled in sequence.

[0087] In a specific embodiment, Figure 8 As shown, in step S10, the point sequence list is further screened, and the specific steps are:

[0088] S101: Obtain a perpendicular line L with a point in the point sequence list as the long side; the length of the perpendicular line is the length of the short side, and define count to count whether the perpendicular line intersects with the lane line;

[0089] S102: If the vertical line L intersects the lane line, count is increased by 1;

[0090] S103: looping steps S101 to S102 until all points in the point sequence list are traversed; if count is greater than or equal to the set threshold, it is determined that the screening principle is met; if count is still less than the set threshold, it is determined that the screening principle is not met.

[0091] Example 2

[0092] More specifically, Figure 3 As shown, in this embodiment, the lane line of the target road is the bus lane and the emergency lane.

[0093] In a specific embodiment, Figure 4 As shown, the specific implementation process of the method for identifying a cluster with a lane change feature of the present invention is as follows:

[0094] Clustering the lane lines of the target road to be changed;

[0095] Get the result list list_ret_datas after clustering is completed; loop the list to get the geometric information list list_ret_data of one cluster;

[0096] Set the minimum gap width min_length = 1, the maximum gap width max_length = 5, the step length step = 0.5, and the minimum gap range min_range = 2;

[0097] Determine whether the input data is in the form of a list: if so, organize it into GeoDataFrame format to form data G, and convert its data into the WGS84 coordinate system. If not, return to obtain the geometric information list list_ret_data of one cluster;

[0098] Get the two sides of the circumscribed rectangle of the current cluster, exterior_line_a and exterior_line_b, by obtaining the circumscribed rectangle of the current cluster and the point sequence geom_rectangle_points list of the circumscribed rectangle. Compare and judge the lengths of the two sides to get the long side long_line and the short side short_line;

[0099] Returns a sequence of points that are 0.01 meters away from the long side long_line; takes the step length step as the starting point and the length of the long side as the end point, and gradually obtains a sequence of points that are separated by the step length step from the long side; obtains the sequence of points at the length of the long side minus 0.01 meters, and puts all the points obtained by these three judgment methods into the list long_point; if the length of the obtained long_point is less than the integer obtained by dividing the minimum width of the gap min_length by the step length step, returns null, indicating that no qualified points are found;

[0100] Get the start and end points of the short side. If the distance between the end point of the short side and the long side is smaller than the distance between the start point of the short side and the long side, the start and end points of the short side need to be swapped, otherwise they do not need to be swapped; synchronously obtain the coordinate difference x, y between the start and end points of the short side; this is used for subsequent conversion of coordinate points in GeoDataFrame format;

[0101] Convert the obtained long_point to GeoDataFrame format and reorganize it to get long_point_gdf. Among them, long_point_gdf contains a point sequence formed by the column geometry, the value long_point, and the column index, the value of the long_point length. Then pass x and y as parameters, convert the coordinates of the column geometry as the value of the column move_point; copy the value of the column geometry to get the value of the column org_point; correspond the value of the column geometry to the value of the column move_point one by one to form a line, overwriting the original value of the column geometry. In this way, the complete long_point_gdf data is organized. Subtract the data from G to get the result check_result, filter out the data in the column geometry of check_result that is a line type and the length of the line is less than a certain threshold, and further judge whether the length of the filtered check_result is less than a certain threshold. The threshold is the integer obtained by dividing the minimum width of the gap min_length by the step step. If it is less than, it does not meet the filter requirements, and returns empty, indicating that no qualified clusters are found. Return the list list_ret_data of geometric information of one of the clusters. Otherwise, further obtain the value of the column index of check_result as col, perform mathematical statistics on it, and obtain the value of the column group_id of check_result.

[0102] After organizing the values ​​of check_result, loop through the values ​​of group_id in the check_result column, convert the values ​​of move_point and org_point in the same group_id into a list, and finally obtain the inter_poly_list list of rectangular boxes that meet the requirements through mathematical judgment, that is, obtain the pre-selected list of lane change behaviors; it contains boxes and lines that meet the requirements.

[0103] Get the line_infos that intersect with the box. Input the line information list line_info_list and build the STRtree index tree; loop the inter_poly_list list, retrieve the line_infos that intersect with this box, and reorganize the data according to the timestamp; if the length of line_infos is less than a certain threshold, it means that the screening is unsuccessful, otherwise continue to judge;

[0104] Get information related to line. Loop line_infos to get the point sequence of the circumscribed rectangle of line line, get its long side and short side, further get the azimuth of the short side, and perform point-filling operation on the long side to get the point sequence list sup_points; the point-filling operation requires first connecting two points into a line, dividing the length of the line to be filled by the number of points to be filled and rounding, and after obtaining the position of each point to be filled from the starting point, fill in the points one by one.

[0105] Further filter the clusters that meet the requirements. Loop sup_points, get the perpendicular line L with one of the points as the long side. The length of the perpendicular line is the length of the short side. Define count to count whether the perpendicular line intersects with the line. Loop line_infos again. If the perpendicular line L intersects with the line, add 1 to count. If count is greater than or equal to 2, that is, the number of lines intersecting with the perpendicular line is greater than or equal to 2, then it is judged to meet the screening principle. On the contrary, if after the sup_points loop is completed, count is still less than the set threshold, then it is judged to not meet the screening principle. If the line in the line_info found and the corresponding perpendicular line do not meet the screening principle, no output is performed, and the collection information list list_ret_data of a certain family is returned to be re-obtained. If all the rectangular boxes found do not meet the screening principle, no output is performed. Return the geometric information list list_ret_data of one of the clusters again until all clusters are looped through.

[0106] Therefore, the present invention can identify lane changes through their geometric features without any prior conditions, accurately identify lane lines with lane change features, and effectively extract problematic clusters, which is convenient for subsequent post-processing and improves the accuracy of the clustering module. The present invention does not require a large amount of data sets or the extraction of a large number of features for training, has a small amount of calculation, and has a fast recognition speed, and is not dependent on changes in the data set.

[0107] Example 3

[0108] like Figure 5As shown, a system for identifying clusters with road lane change characteristics includes a clustering module, an extraction module, a first screening module, a long and short edge module, a second screening module, a lane line module, a third screening module, a lane line point sequence module, a fourth screening module, and an output module;

[0109] The clustering module is used to cluster the lane lines of the target road to be changed;

[0110] The extraction module is used to extract the geometric information of one cluster from the result list after clustering is completed;

[0111] The first screening module is used to determine whether the obtained geometric information is in the form of a list, and if so, convert the geometric information list into a unified coordinate system;

[0112] The long and short edge module is used to obtain the long and short edges of the geometric information of the converted cluster; obtain a list of long edge point sequences near the long edge; and obtain the starting point and end point of the short edge;

[0113] The second screening module is used to perform spatial coverage judgment on the long side point sequence list according to the starting point and the end point of the short side; if the long side point sequence passes the spatial coverage judgment, a lane change pre-selection list is obtained according to the long side point sequence list;

[0114] The lane line module is used to extract the lane line information list in the result list; construct an index tree according to the lane line information list; retrieve the lane line information that intersects with the geometric information of the lane change pre-selection list through the index tree; and reorganize the lane line information according to the timestamp;

[0115] The third screening module is used to determine whether the reorganized lane line information is greater than a threshold;

[0116] The lane line point sequence module is used to obtain the point sequence of the geometric information of the lane line in the lane line information, and obtain its long side and short side; further obtain the azimuth of the short side, and perform a point addition operation on the long side to obtain a point sequence list;

[0117] The fourth screening module is used to further screen the point sequence list. If the point sequence list meets the screening condition, it is determined that the corresponding cluster has a lane change feature.

[0118] The output module is used to traverse all clusters in the result list after clustering is completed, so as to obtain and output all clusters with road change characteristics.

[0119] Obviously, the above embodiments of the present invention are only examples for clearly explaining the present invention, and are not intended to limit the implementation methods of the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention shall be included in the protection scope of the claims of the present invention.

Claims

1. A method for identifying clusters with road change characteristics, characterized in that: The specific steps include: S1: Clustering the lane lines of the target road to be changed; S2: Extract the geometric information of one cluster from the result list after clustering is completed; S3: Determine whether the obtained geometric information is in the form of a list. If so, convert the geometric information list into a unified coordinate system and execute step S4; otherwise, return to step S2; S4: Obtain the long and short sides of the geometric information of the transformed cluster; S5: Get the list of long edge point sequences near the long edge; get the starting point and end point of the short edge; S6: Perform spatial coverage judgment on the long edge point sequence list according to the starting point and the end point of the short edge; if the spatial coverage judgment is not passed, return to step S2; if the long edge point sequence passes the spatial coverage judgment, obtain a lane change pre-selection list according to the long edge point sequence list; S7: extracting a lane line information list from the result list; and constructing an index tree according to the lane line information list; Retrieve lane line information that intersects with the geometric information in the lane change pre-selection list through the index tree; Reorganize the lane line information according to the timestamp; S8: Determine whether the reorganized lane line information is greater than a threshold value. If it is less than the threshold value, return to step S7; if it is greater than the threshold value, execute step S9; S9: Obtain a point sequence of geometric information of the lane line in the lane line information, and obtain its long side and short side; further obtain the azimuth of the short side, and perform a point-adding operation on the long side to obtain a point sequence list; S10: further filter the point sequence list. If the point sequence list meets the filter condition, determine that the corresponding cluster has the lane change feature, and execute step S11; if the point sequence list does not meet the filter condition, return to step S9; if the point sequence lists of all lane lines in the lane line information do not meet the filter condition, return to step S2; S11: Output the family with the characteristics of road change; return to Step S2, until all clusters in the result list after clustering are traversed, so as to obtain all clusters with road change characteristics.

2. The method for identifying a cluster having a lane change characteristic according to claim 1, characterized in that: In the step S2, the minimum width and step length of the gap in the geometric information list are also set.

3. The method for identifying a cluster having a lane change characteristic according to claim 2, characterized in that: In the step S3, the geometric information list is converted into a unified coordinate system, specifically: determining whether the input data is in the form of a list, organizing the geometric information list into data G in gdf format, and converting the data G into the WGS84 coordinate system.

4. The method for identifying a cluster having a lane change characteristic according to claim 3, characterized in that: In the step S4, the long side and the short side of the geometric information of the converted cluster are obtained, specifically: the circumscribed rectangular frame of the converted cluster and the point sequence list of the circumscribed rectangular frame are obtained, and the two sides of the circumscribed rectangular frame are obtained; the lengths of the two sides are compared and judged to divide them into the long side and the short side.

5. The method for identifying a cluster having a lane change characteristic according to claim 4, characterized in that: In the step S5, a list of long edge point sequences near the long edge is obtained, specifically: S501: obtaining a point sequence having a certain distance from the long side; obtaining a point sequence having a step length as a starting point and a length of the long side as an end point, and gradually obtaining a point sequence having a step length as an interval from the long side; obtaining a point sequence obtained by subtracting a certain distance from the length of the long side; S502: Put all three point sequences into a list to obtain a long edge point sequence list; if the length of the obtained point sequence is less than the integer obtained by dividing the minimum width of the gap by the step length, it is not put into the list.

6. The method for identifying a cluster having a lane change characteristic according to claim 4, characterized in that: In step S5, the starting point and the end point of the short side are obtained, specifically: Get the first and last points of the short side as the starting point and end point of the short side; If the distance between the end point of the short side and the long side is smaller than the distance between the start point of the short side and the long side, the start point and the end point of the short side are swapped, otherwise they do not need to be swapped; the coordinate difference x, y of the start point and the end point of the short side is obtained simultaneously.

7. The method for identifying a cluster having a lane change characteristic according to claim 6, characterized in that: In step S6, the spatial coverage of the long edge point sequence list is judged according to the starting point and the end point of the short edge, and the specific steps are as follows: S601: Convert the long edge point sequence list into gdf format; S602: performing coordinate conversion on the long side point sequence list in the gdf format according to the coordinate difference x, y to obtain a complete long side point sequence list in the gdf format; S603: Subtract the complete gdf format long edge point sequence list from the data G to obtain the inspection result; if the column of the inspection result is line type data, and the length of the inspection result is not less than the integer obtained by dividing the minimum width of the gap by the step size, then the long edge point sequence list corresponding to the inspection result passes the spatial coverage judgment; otherwise, the long edge point sequence list fails the spatial coverage judgment.

8. The method for identifying a cluster having a lane change characteristic according to claim 1, characterized in that: In step S9, the point filling operation is specifically as follows: the length of the long side of the point to be filled is divided by the number of points to be filled and rounded up, and after obtaining the position of each point to be filled from the starting point, the points are filled in sequence.

9. The method for identifying a cluster having a lane change characteristic according to claim 1, characterized in that: In the step S10, the point sequence list is further screened, and the specific steps are as follows: S101: Obtain a perpendicular line L with a point in the point sequence list as the long side; the length of the perpendicular line is the length of the short side, and define count to count whether the perpendicular line intersects with the lane line; S102: If the vertical line L intersects the lane line, count is increased by 1; S103: looping steps S101 to S102 until all points in the point sequence list are traversed; if count is greater than or equal to the set threshold, it is determined that the screening principle is met; if count is still less than the set threshold, it is determined that the screening principle is not met.

10. A system for identifying clusters with road change characteristics, characterized by: It includes a clustering module, an extraction module, a first screening module, a long and short edge module, a second screening module, a lane line module, a third screening module, a lane line point sequence module, a fourth screening module, and an output module; The clustering module is used to cluster the lane lines of the target road to be changed; The extraction module is used to extract the geometric information of one cluster from the result list after clustering is completed; The first screening module is used to determine whether the obtained geometric information is in the form of a list, and if so, convert the geometric information list into a unified coordinate system; The long and short edge module is used to obtain the long and short edges of the geometric information of the converted cluster; obtain a list of long edge point sequences near the long edge; and obtain the starting point and end point of the short edge; The second screening module is used to perform spatial coverage judgment on the long edge point sequence list according to the starting point and the end point of the short edge; If the long edge point sequence passes the spatial coverage judgment, then the lane change pre-selection list is obtained according to the long edge point sequence list; The lane line module is used to extract the lane line information list in the result list; and construct an index tree according to the lane line information list; Retrieve lane line information that intersects with the geometric information in the lane change pre-selection list through the index tree; Reorganize the lane line information according to the timestamp; The third screening module is used to determine whether the reorganized lane line information is greater than a threshold; The lane line point sequence module is used to obtain the point sequence of the geometric information of the lane line in the lane line information, and obtain its long side and short side; further obtain the azimuth of the short side, and perform a point addition operation on the long side to obtain a point sequence list; The fourth screening module is used to further screen the point sequence list. If the point sequence list meets the screening condition, it is determined that the corresponding cluster has a lane change feature. The output module is used to traverse all clusters in the result list after clustering is completed, so as to obtain and output all clusters with road change characteristics.

Images