Intelligent image recognition method for high-precision map

Abstract

The invention relates to the technical field of image processing, and provides an intelligent image recognition method for a high-precision map, which comprises the following steps of: acquiring various map elements required by the high-precision map and contour information of the map elements, and recording related data of a plurality of attributes of each map element; obtaining a plurality of categories of each attribute of each map element; obtaining a comprehensive basic value of each map element, and constructing a map element basic sequence from large to small; constructing a plurality of scales of the map element quantity from small to large; constructing a map distribution diagram of each scale, and obtaining a chain code vector of each edge in the map distribution diagram; screening to obtain a plurality of reserved scales; marking a plurality of nodes; obtaining a path of each node of each reserved scale; and screening a plurality of reserved paths of each node, and generating a high-precision map in combination with the map distribution diagram of each reserved scale. The invention aims to solve the problem that the matching of a road network and a lane network at an intersection is not accurate enough in a high-precision map making process.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of image processing, and in particular to an intelligent image recognition method for high-precision maps. BACKGROUND

[0002] With the popularization of intelligent driving technology, the requirement for map precision is also higher and higher; traditional maps mainly provide drivers with routes to destinations, but there is no more detailed spatial information on the entire route, and intelligent driving technology needs to be aware of the vehicle-mounted sensors (such as cameras, laser radars, etc.) to perform centimeter-level positioning, so it often needs the service of high-precision maps.

[0003] In the process of making high-precision maps, one of the main tasks is to build the topology of the road network. Topology describes the connection between map elements and the traffic rules. Intelligent transportation systems can implement lane-level control based on topology. When building the topology of the road network, the SD road network (traditional navigation map road) and the HD lane network (high-precision map lane element) need to be combined, that is, the key point matching of the SD road network and the HD lane network needs to be performed. Therefore, in order to ensure the accuracy of the subsequent key point matching results and thus the accuracy of the high-precision map, the collected data needs to be preprocessed in the early stage to accurately reflect the information points for matching, especially in complex places such as road intersections. It is often difficult to accurately match due to too complex lane elements and information, thereby affecting the accuracy of high-precision map making. SUMMARY

[0004] The application provides an intelligent image recognition method for high-precision maps to solve the problem of inaccurate matching of road networks and lane networks at intersections in the existing high-precision map making process. The technical solution adopted is as follows: The application provides an intelligent image recognition method for high-precision maps, which comprises the following steps: Collecting various map elements and their contour information required for high-precision maps and recording the relevant data of several attributes of each map element; Classifying the relevant data of each attribute of each map element to obtain several categories of each attribute of each map element; analyzing the distribution relationship of the relevant data in the categories of different attributes of different map elements in the same geographical location to obtain the basic degree of each attribute of each map element, and then obtaining the comprehensive basic value of each map element and constructing a map element basic sequence from large to small; constructing several scales with the number of map elements from small to large based on the map element basic sequence; Based on the relevant data of the attributes of the map elements at each scale, a map distribution map of each scale is constructed, and the chain code vector of each edge is obtained. The differences in the similarity relationship between the chain code vectors of the closest edges at adjacent scales are analyzed to screen out several retained scales. Based on the distribution relationship of the edges in the map distribution map of the retained scale with the least number of map elements among the retained scales, several nodes are marked. Based on the edge distribution and related data of each attribute of each map element in the map distribution map of each retention scale, the path of each node at each retention scale is obtained in combination with the Markov model; the difference between the paths of nodes at adjacent retention scales is analyzed, and several retention paths of each node are screened. The high-precision map is generated in combination with the map distribution map of each retention scale.

[0005] Optionally, the specific method of obtaining several categories of attributes of each map element includes: For any attribute of any map element, if the attribute is a numeric attribute, cluster the related data of the attribute to obtain several clusters, and take the related data in any cluster as a category of the attribute; If the attribute is a text attribute, the text information corresponding to the relevant data of the attribute is numbered, and the same text information has the same number. The relevant data of the attribute with the same number is regarded as a category of the attribute.

[0006] Optionally, the method of obtaining the basic degree of each attribute of each map element, and then obtaining the comprehensive basic value of each map element, and constructing a basic sequence of map elements from large to small includes the following specific methods: Build regions based on longitude and latitude. For a number of related data of any category, record the areas corresponding to the related data of the category as the distribution area of ​​the category, obtain the related data of any attribute of any other map element in the distribution area except the map element, and calculate the first The joint distribution probability of the attribute of this category and the attribute of other map elements is The mean of the joint distribution probability of each category of the attribute and each attribute of other map elements is used as the first The attribute and the basic factor of the other map elements are used to The average of the attributes and the basic factors of all other map elements except this map element is used as the first The basic level of each attribute; The average value of the basic degree of each attribute of the map element is used as the comprehensive basic value of the map element; the map elements are arranged in descending order of the comprehensive basic value, and the obtained sequence is recorded as the map element basic sequence.

[0007] Optionally, the method of constructing several scales of map elements from small to large in number based on the basic sequence of map elements includes the following specific methods: According to the order of map elements in the basic sequence of map elements, the first scale is constructed with the first map element, and the second scale is constructed with the first and second map elements, and so on, to obtain several scales.

[0008] Optionally, the construction of the map distribution diagrams of each scale and obtaining the chain code vector of each edge therein includes the following specific methods: For any scale, for each map element contained in the scale, the map distribution map of the scale is constructed based on the relevant data of each attribute of each map element, combined with its spatial contour information and location information; Obtain several edges in the map distribution map, and treat two completely parallel edges corresponding to the same map element as an edge group; for any edge, obtain several key points on the edge including the starting point, end point, and inflection point, and obtain the chain code of the edge based on the key points; Get the chain code of each edge in the map distribution map of this scale. Based on the chain code of the edge with the largest number of elements in the chain code, get the chain code vector of this edge. Get the chain code vectors for the chain codes of other edges. If the number of chain code elements is insufficient, add 0 at the end of the chain code.

[0009] Optionally, the obtaining of the plurality of retention scales includes the following specific methods: Analyze the differences in the similarity relationship between the chain code vectors of the nearest edge at adjacent scales to obtain the neighbor change factor of each edge group at adjacent scales; Get the At each scale, each edge group is A scale and The neighbor change factor of each scale is the average of all neighbor change factors as the first The scale relative to the Information change characteristics of each scale; The judgment starts from the information change characteristics of the second scale relative to the first scale. If the information change characteristics are less than the change threshold, the second scale is discarded, and the information change characteristics of the third scale relative to the first scale are calculated. If the third scale is retained, the subsequent calculation is based on the information change characteristics of the fourth scale relative to the third scale. And so on. Several scales are discarded based on the information change characteristics, and the remaining scales are retained.

[0010] Optionally, the obtaining of the neighbor change factor of each edge group at adjacent scales includes the following specific methods: For the obtaining an edge closest to the edge group based on the Euclidean distance average of the key points of the two edges in the edge group and other edges, taking the edge with the minimum Euclidean distance average as the edge closest to the edge group, obtaining the cosine similarity between the chain code vector of the edge group and the chain code vector of the edge closest to the edge group as the first scale neighbor similarity of the edge group; re-obtaining the edge closest to the edge group in the first scale, and re-obtaining the first scale neighbor similarity of the edge group; obtaining the absolute value of the difference between the neighbor similarities of the edge group in the first scale and the second scale as the neighbor change factor of the edge group in the first

[0011] Optionally, the marking of the nodes comprises the following specific method: taking the reserved scale with the least number of map elements as the maximum reserved scale; and obtaining a plurality of combinations of regional edge groups based on the distribution of the edges in each edge group in the distribution map of the maximum reserved scale; For any one combination of regional edge groups, the combination of regional edge groups is used for regional division, and two edge groups form a division region; and each edge in the two edge groups is extended until the edges of the two edge groups intersect, and the intersection part is marked as a node. For the original edges in the two edge groups, for any one edge group, a preset segmentation ratio is obtained, a plurality of straight lines perpendicular to the two edges are obtained, and from the starting point of each edge, a perpendicular straight line is extracted every segmentation ratio between the starting point and the next key point as a grid line in the edge group, and then a plurality of grid lines are obtained, and the intersection of the grid line and the two edges is marked as a node.

[0012] Optionally, the obtaining of the plurality of combinations of regional edge groups comprises the following specific method: For each edge group in the distribution map of the maximum reserved scale, an edge closest to the edge group is obtained, and based on the edge closest to the edge group, an edge group closest to the edge of each edge group is obtained, and a plurality of combinations of edge groups closest to the edge are obtained. For any one combination of edge groups closest to the edge, the direction difference between the two parts of the edge group closest to the edge is obtained, and if the direction difference is greater than or equal to a difference threshold, the combination of edge groups closest to the edge is taken as a combination of regional edge groups.

[0013] Optionally, the screening of the plurality of reserved paths of each node comprises the following specific method: For any node, two paths in the adjacent reserved scales are obtained, DTW distances of the two paths are obtained through DTW matching, a maximum value of the DTW distances of the node between all the paths in the reserved scales is obtained, a ratio of the DTW distance corresponding to the adjacent reserved scale to the maximum value of the DTW distances is taken as a path difference factor of the adjacent reserved scale, a preset path difference threshold is taken, if the path difference factor is greater than the path difference threshold, two paths of the node in the adjacent reserved scale are reserved, if the path difference factor is less than or equal to the path difference threshold, the path of the node in the latter one of the adjacent reserved scales is removed, and the paths of the node reserved in each reserved scale are taken as a plurality of reserved paths of the node.

[0014] The beneficial effects of the present application are: in the high-precision map generation process, the influence of intersection complex information on the SD road network and the HD lane network is considered, wherein through information collection and attribute data analysis of various map elements, according to the influence relationship between data changes under the attribute of the map element and data changes under other map elements, the basic map elements are screened and a plurality of scales from large to small are constructed, the number of map elements corresponding to the map elements is from small to large, and the comprehensive basic value of the map elements gradually decreases; further, according to the map elements and their attribute data, a map distribution graph is constructed for each scale, the difference change of the similarity relationship between the edges under the adjacent scales is analyzed, the scales corresponding to the repeated information are removed, the reserved scales are obtained, and the grid lines and the nodes intersected by the edges are constructed according to the map distribution graph, so as to preliminarily obtain the intersection information; through the historical vehicle trajectory combined with the prediction model, the paths of each node are predicted, and the reserved paths are further screened to reduce the interference of repeated information on the learning of the prediction model, and finally the matching between the road maps is performed through the nodes and their reserved paths, so as to improve the accuracy of high-precision map generation and recognition. BRIEF DESCRIPTION OF DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, brief descriptions will be given to the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0016] Figure 1 A high-precision map intelligent image recognition method flowchart provided by an embodiment of the present application; Figure 2 A high-precision road map schematic diagram. DETAILED DESCRIPTION

[0017] With reference to the accompanying drawings, the technical solutions in the embodiments of the present application will be clearly and completely described below, obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0018] Please refer to Figure 1 , which shows a high-precision map intelligent image recognition method flow chart provided by an embodiment of the present application, the method comprises the following steps: Step S001, collect various map elements and their contour information required for high-precision map, and record the relevant data of several attributes of each map element.

[0019] The purpose of the embodiment is to construct a high-precision map based on the topological relationship of the road network, and the road network contains various map elements, which need to be collected and the contour information of each map element recorded, and each map element contains several attributes, which also need to record the data of each attribute as the relevant data of each attribute.

[0020] Specifically, the map elements such as road lines, traffic signs, traffic signs, traffic lighting equipment and other traffic facilities are collected, and the collection means include but are not limited to vehicle-mounted mobile measurement system, laser radar, oblique photography and other methods; In addition to the collection of map elements, the contour information of each map element is recorded during the collection process, and the spatial contour information of the map elements is represented in the form of points, lines and surfaces.

[0021] Further, the data of multiple attributes of each map element are also recorded, for example, the attributes of road line map elements include line type, line width, line length, line color, turning information, vehicle type setting and speed limit information, and the corresponding data of each attribute of the map element is collected, and each attribute of the same map element contains several data (corresponding to different roads, traffic signs, etc. in the road network), which are stored as the relevant data of each attribute of each map element in SHP format.

[0022] It should be noted that, as Figure 2 shown, it shows a schematic diagram of a road high-precision map, which provides a city high-precision map collection work, and the collected elements include but are not limited to highway main road, ramp, interchange, bridge, tunnel, entrance and exit, monitoring equipment, signs and markings, various signs, information board, lighting facilities, isolation belt, reflective facility, guardrail, vehicle-road cooperation equipment, etc., with an absolute accuracy of 20-50 cm and a relative accuracy of 10-20 cm, which can be applied to high-speed or urban expressway, urban ordinary road, holographic intersection and other scenes, and provides support for intelligent management and control such as congestion control, safety prevention and control, and fine management and control.

[0023] It should be noted that in the process of high-precision map making, one of the main works is to construct the road network topology relationship, and in the construction of the road network topology relationship, key point matching of the SD road network and the HD lane network needs to be performed, so as to ensure the accuracy of the subsequent key point matching result, thereby ensuring the accuracy of the high-precision map; the road network is composed of roads and intersections, and the intersection is an important node connecting various roads, and it is often the most complex place in the entire road network, and if accurate matching is required, the information at the intersection needs to be more explicit; by analyzing the lane change information at the intersection, the topology relationship at the intersection is determined, thereby providing conditions for the subsequent matching process of the SD road and the HD lane network.

[0024] In step S002, the related data of each attribute of each map element is classified to obtain several categories of each attribute of each map element; the distribution relationship of the related data in the categories of different attributes of different map elements in the same geographical position is analyzed to obtain the basic degree of each attribute of each map element, and then the comprehensive basic value of each map element is obtained, and a map element basic sequence is constructed from large to small; and several scales are constructed based on the map element basic sequence from small to large.

[0025] It should be noted that the map elements constructed during data collection include road lines, traffic signs, traffic signs, traffic lighting equipment and other traffic facilities, and the like, which comprehensively constitute the entire road network; the information contained in each map element is different, for example, the information contained in road lines (lane center line, lane edge line) at different positions is different, the lane center line has description of line type, line width, line length, line color, turning information, vehicle type setting, speed limit information, etc., and the lane edge line only has line type, line width, line color, etc., so that the information contained is different, and the position attributes of traffic signs, traffic signs, traffic lighting equipment and the like are determined by the lane line, so it is necessary to determine the reference map element; the expected reference map element can determine the position and other attributes of other map elements, and can also be used to reflect the topology relationship of the intersection for subsequent matching services.

[0026] Preferably, in an embodiment of the present application, the related data of each attribute of each map element is classified to obtain several categories of each attribute of each map element, and the specific method comprises: For any attribute of any map element, if the attribute is a digital attribute, the related data of the attribute is clustered, and in this embodiment, K-means clustering is adopted, the distance measurement adopts the absolute value of the difference between the related data, 6 clustering centers are set, that is, K=6, several clusters are obtained, and the related data in any cluster is taken as a category of the attribute, and then several categories of the attribute of the map element are obtained.

[0027] Furthermore, if the attribute is a text attribute, the text information corresponding to the relevant data of the attribute is numbered, and the same text information has the same number (the number has no actual meaning and is only used for classification). The relevant data of the attribute with the same number is regarded as a category of the attribute, and several categories of the attribute are obtained.

[0028] It should be further explained that, since the latitude and longitude of various map elements are the most basic geographical location elements, comparison is made based on longitude and latitude, that is, distribution comparison is made in similar longitude and latitude areas; a map element contains various attributes, and all attributes reflect all information of the map element. To determine whether this map element is a basic map element, it is necessary to combine the relationship between the attributes and other map elements. If any attribute of the target map element has the same distribution as other map elements, the corresponding map element has a higher degree of basicity.

[0029] Preferably, in one embodiment of the present invention, the distribution relationship of relevant data in categories of different attributes of different map elements in similar geographical locations is analyzed to obtain the basic degree of each attribute of each map element, and then obtain the comprehensive basic value of each map element, and construct a basic sequence of map elements from large to small. The specific method includes: The area is constructed with 1 second of longitude and latitude. For a number of related data of any category, record the areas corresponding to the related data of the category as the distribution area of ​​the category, obtain the related data of any attribute of any other map element in the distribution area except the map element, and calculate the first The joint distribution probability of the attribute of this category and the attribute of other map elements is The mean of the joint distribution probability of each category of the attribute and each attribute of other map elements is used as the first The attribute and the basic factor of the other map elements are used to The average of the attributes and the basic factors of all other map elements except this map element is used as the first It should be noted that the calculation of joint distribution probability is an existing technology in the field of probability theory and will not be described in detail in this embodiment.

[0030] The average value of the basic degree of each attribute of the map element is used as the comprehensive basic value of the map element; the map elements are arranged in descending order of the comprehensive basic value, and the obtained sequence is recorded as the map element basic sequence.

[0031] It is further needed to be explained that different scales contain different information, referring to the main branch idea of the theory of cartography, different scales are constructed with different information performance effects, from simple to complex, the node changes of the analysis possibility are analyzed, in the process of constructing the scale, the small scale is required to be the refinement of the information of the large scale, therefore the first scale is the largest scale, that is, the map element with the largest basic value is taken as the largest scale.

[0032] Preferably, in an embodiment of the present application, a plurality of scales of map elements are constructed based on the basic sequence of map elements from small to large, and the specific method comprises: According to the order of the map elements in the basic sequence of map elements, the first map element is used to construct the first scale, and the first and second map elements are used to construct the second scale, and so on, that is, the first several map elements are used to construct the first several scales, and then a plurality of scales are obtained, and the number of map elements in the scales is from small to large.

[0033] At this point, a plurality of scales containing different numbers of map elements are obtained.

[0034] Step S003, according to the related data of the attributes of the map elements in each scale, a map distribution diagram of each scale is constructed, and the chain code vector of each edge in the map distribution diagram is obtained; the difference in the similarity relationship of the chain code vectors of the nearest edges in the adjacent scales is analyzed, and a plurality of reserved scales are screened; and a plurality of nodes are marked based on the distribution relationship of the edges in the map distribution diagram of the reserved scale with the least number of map elements.

[0035] It is needed to be explained that since some map elements are repeated, there will be repetitive content in the combination of map elements corresponding to the scale, that is, when the nodes are determined subsequently, it is not necessary to calculate the repetitive scale; for any scale, according to the map elements contained in the scale, the latitude and longitude coordinates and the attributes of the corresponding map elements are combined to construct a map distribution diagram (similar to a map CAD diagram).

[0036] Preferably, in an embodiment of the present application, according to the related data of the attributes of the map elements in each scale, a map distribution diagram of each scale is constructed, and the chain code vector of each edge in the map distribution diagram is obtained, and the specific method comprises: For any scale, according to the related data of the attributes of each map element in the scale, the spatial contour information and the position information (latitude and longitude coordinates) are combined to construct a map distribution diagram of the scale (reference to the construction of a map CAD diagram).

[0037] Furthermore, several edges in the map distribution map are obtained. Since the contour information of road lines and traffic facilities is usually a group of parallel edges, two completely parallel edges corresponding to the same map element (with the same length) are regarded as an edge group. For any edge, several key points on the edge, including the starting point, end point and inflection point (the turning point where the direction changes, that is, the direction of the edge changes), are obtained, and the chain code of the edge is obtained based on the key points. The chain code of each edge in the map distribution map of the scale is obtained according to the above method. Based on the chain code of the edge with the largest number of elements in the chain code, the chain code vector of the edge is obtained, that is, the elements in the chain code constitute the chain code vector in sequence, and the number of elements in the chain code vector is equal to the number of elements in the chain code of the edge. Then, the chain code vectors of the chain codes of other edges are also obtained. If the number of chain code elements is insufficient, 0 is padded at the end of the chain code.

[0038] It should be further explained that, by performing similarity analysis on the chain code vectors between edge groups, if the similarity relationship between the same edge group and the chain code vector of the nearest edge does not change after the scale is reduced, then the scale changes without significant changes in the map distribution information, and the reduced scale should be discarded to avoid repeated information analysis.

[0039] Preferably, in one embodiment of the present invention, the difference in similarity between chain code vectors at adjacent scales and the nearest edge is analyzed to screen out several retained scales, including the following specific methods: For the For any edge group of a scale, the chain code vectors of the two edges in the edge group are the same, and are used as the chain code vector of the edge group. The edge closest to the edge group is obtained. Based on the mean Euclidean distance between the key points of the two edges in the edge group and other edges, that is, the mean Euclidean distance between the two edges and any other edge corresponding to a key point, the edge with the smallest mean Euclidean distance is taken as the edge closest to the edge group. The cosine similarity between the chain code vector of the edge group and the chain code vector of the edge closest to the edge group is obtained as the first The neighbor similarity of the edge group at scale ; similarly, Re-obtain the nearest edge of the edge group in the scale, and re-obtain the The nearest neighbor similarity of the edge group at the first scale is obtained; the absolute value of the difference between the nearest neighbor similarities of the edge group at two adjacent scales is obtained as the nearest neighbor similarity of the edge group at the first scale. A scale and The neighbor variation factor of each scale.

[0040] Further, according to the above method, the At each scale, each edge group is A scale and The neighbor change factor of each scale is the average of all neighbor change factors as the first The scale relative to the The information change characteristics of each scale are calculated; a preset change threshold is set. In this embodiment, the change threshold is 0.2. The judgment starts from the information change characteristics of the second scale relative to the first scale. If the information change characteristics are less than the change threshold, the second scale is discarded, and the information change characteristics of the third scale relative to the first scale are calculated. If the third scale is retained, the calculation is subsequently performed based on the information change characteristics of the fourth scale relative to the third scale. And so on. Several scales are discarded based on the information change characteristics, and the remaining scales are retained.

[0041] It should be further explained that since the added nodes are related to the interaction between map elements, the region is divided based on the maximum scale among the retained scales, that is, the retained scale with the least number of map elements. The purpose of region division and node marking is to mark the intersection part through edge extension and subsequent path planning. Grid lines are generated through the corresponding areas, that is, the width of the corresponding edge group is constructed based on the edge group, and the nodes are marked based on the grid.

[0042] Preferably, in one embodiment of the present invention, a number of nodes are marked based on the distribution relationship of the edges in the map distribution graph of the retained scale with the least number of map elements in the retained scale, and the specific method includes: The retained scale with the least number of map elements among the retained scales is used as the maximum retained scale; for each edge group in the map distribution map of the maximum retained scale, the nearest edge of each edge group is obtained, and based on the edge groups corresponding to the nearest edges, the nearest edge groups of each edge group are respectively obtained, thereby obtaining a plurality of nearest edge group combinations; for any nearest edge group combination, the direction difference (the angle of the strike direction, the nearest part, i.e., the part of the edge group corresponding to the smallest distance between the edge segments in the edge group) between the two nearest edge groups is obtained, and a difference threshold is preset. In this embodiment, the difference threshold is described as 15 degrees. If the direction difference is greater than or equal to the difference threshold, the nearest edge group combination is used as a group of regional edge group combinations.

[0043] Further, for any one set of region edge group combination, the region is divided in the set of region edge group combination, that is, two edge groups constitute a divided region; the edges in the two edge groups are extended (extended along the direction of the starting point and the corresponding line segment of the end point) until the edges of the two edge groups intersect, that is, the extensions of the two roads under the intersection in the road network (since there is no corresponding road line in the intersection, the extensions need to intersect), and the intersection part is marked as a node; for the original edges in the two edge groups, since the two edges in the same edge group are parallel, for any one edge group, a preset segmentation ratio is obtained, and the segmentation ratio in this embodiment is 1 / 10, a plurality of straight lines perpendicular to the two edges are obtained, and a grid line in the edge group is obtained from the starting point of each edge to the next key point (knot or end point) (if the two edges are completely parallel, the starting points and key points of the two edges are on the same vertical straight line), and a grid line is extracted every segmentation ratio (edge length between the starting point and the next key point), which is used as the grid line in the edge group. Then a plurality of grid lines are obtained, and the intersection points of the grid lines and the two edges are marked as nodes, and a plurality of nodes are obtained for the set of region edge group combination.

[0044] Up to now, a plurality of nodes are marked in the maximum reserved scale map distribution diagram for subsequent analysis and intersection path planning.

[0045] Step S004, based on the edge distribution in the map distribution diagram of each reserved scale and the related data of the attributes of each map element, the paths of each node of each reserved scale are obtained by combining the Markov model; the differences between the paths of the nodes in adjacent reserved scales are analyzed, and a plurality of reserved paths of each node are screened out, and the high-precision map is generated by combining the map distribution diagram of each reserved scale.

[0046] Preferably, in one embodiment of the present application, based on the edge distribution in the map distribution diagram of each reserved scale and the related data of the attributes of each map element, the paths of each node of each reserved scale are obtained by combining the Markov model, which includes the following specific methods: For a plurality of nodes of any one set of region edge group combination in the maximum reserved scale and the corresponding divided region, a large number of vehicle trajectories of the vehicles in the divided region in the historical data are obtained, the Markov model is constructed by combining the related data of the attributes of each map element (only including the map elements included in the maximum reserved scale) in the divided region, and the vehicle trajectories and the related data are used as input data for training. The nodes in the divided region are judged, and the simulated paths of the nodes are generated; the nodes of each group of region edge groups in the maximum reserved scale are judged based on the Markov model according to the above method, and the simulated paths of the nodes are generated; wherein the Markov model training and path generation based on the vehicle trajectories and the related data are the prior art of the prediction model, and will not be described herein.

[0047] Further, according to the method described above, based on the map distribution diagram of each reserved scale, for each node marked in the maximum reserved scale, according to the corresponding division area of each node in the maximum reserved scale in the map distribution diagram of each reserved scale and the related data of the attributes of the included map elements, the Markov model is trained and the simulation path of each node in the corresponding reserved scale is output, so that the simulation path of each node in each reserved scale is obtained as the path of each node in each reserved scale. It should be noted that the map distribution diagram of the node in different reserved scales is the same, the corresponding division area is also the same based on the map distribution diagram of the maximum reserved scale, and the map elements included in different reserved scales are different, so the training process is affected by the related data of the attributes of the map elements in the corresponding division area.

[0048] Preferably, in an embodiment of the present application, the difference between the paths of the nodes in the adjacent reserved scales is analyzed, and a plurality of reserved paths of each node are screened, and the map distribution diagram of each reserved scale is combined to generate a high-precision map, including the specific method: For any node in two paths of adjacent reserved scales (the reserved scales remaining after discarding part of the scales are still arranged in the original order, here adjacent means sequentially adjacent, and before discarding, they may not be adjacent due to the existence of other discarded scales between them), the DTW distance is obtained by DTW matching of the two paths, the maximum value of the DTW distance between the paths of the node in all reserved scales is obtained, the ratio of the DTW distance corresponding to the adjacent reserved scale to the maximum value of the DTW distance is taken as the path difference factor of the adjacent reserved scale, and the path difference threshold is preset. In this embodiment, the path difference threshold is described as 0.5. If the path difference factor is greater than the path difference threshold, the two paths of the node in the adjacent reserved scale are reserved. Otherwise, if the path difference factor is less than or equal to the path difference threshold, the path of the latter one of the adjacent reserved scales (the reserved scale containing the largest number of map elements among the two reserved scales) is removed. The paths of the node in each reserved scale are taken as a plurality of reserved paths of the node.

[0049] Further, based on the plurality of reserved paths of each node and the plurality of edges in the map distribution diagram corresponding to the reserved scale, the HD lane net and the SD road net are matched. If the topological structures are consistent and the positions are similar, it is considered that the matching is successful. The specific matching method refers to the existing method for matching the SD road net and the HD lane net. This embodiment will not be described again. After obtaining the matching result, the topology is hung according to the arrow of the lane and the traffic regulation common sense, and then the high-precision map is generated. The matching of the HD lane net and the SD road net based on the reserved path and the map distribution is a prior art, and this embodiment will not be described again. Finally, the generation of the high-precision map is realized.

[0050] Thus, the embodiment is completed.

[0051] The above description is only the preferred embodiment of the present application, and is not intended to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the principle of the present application shall be included in the protection scope of the present application.

Claims

1. An intelligent image recognition method for high-precision maps, characterized in that: The method comprises the following steps: Collect various map elements and their outline information required for high-precision maps, and record relevant data on several attributes of each map element; Classify the relevant data of each attribute of each map element to obtain several categories of each attribute of each map element; analyze the distribution relationship of relevant data in the categories of different attributes of different map elements in similar geographical locations to obtain the basic degree of each attribute of each map element, and then obtain the comprehensive basic value of each map element, and construct a basic sequence of map elements from large to small; based on the basic sequence of map elements, construct several scales of map element quantity from small to large; Based on the relevant data of the attributes of the map elements at each scale, a map distribution map of each scale is constructed, and the chain code vector of each edge is obtained. The differences in the similarity relationship between the chain code vectors of the closest edges at adjacent scales are analyzed to screen out several retained scales. Based on the distribution relationship of the edges in the map distribution map of the retained scale with the least number of map elements among the retained scales, several nodes are marked. Based on the edge distribution and related data of each attribute of each map element in the map distribution map of each retention scale, the path of each node at each retention scale is obtained in combination with the Markov model; the difference between the paths of nodes at adjacent retention scales is analyzed, and several retention paths of each node are screened. The high-precision map is generated in combination with the map distribution map of each retention scale.

2. The intelligent image recognition method for high-precision maps according to claim 1, characterized in that: The specific method for obtaining several categories of attributes of each map element includes: For any attribute of any map element, if the attribute is a numeric attribute, cluster the related data of the attribute to obtain several clusters, and take the related data in any cluster as a category of the attribute; If the attribute is a text attribute, the text information corresponding to the relevant data of the attribute is numbered, and the same text information has the same number. The relevant data of the attribute with the same number is regarded as a category of the attribute.

3. The intelligent image recognition method for high-precision maps according to claim 1, characterized in that: The method of obtaining the basic degree of each attribute of each map element, and then obtaining the comprehensive basic value of each map element, and constructing the basic sequence of map elements from large to small, includes the following specific methods: Build regions based on longitude and latitude. For a number of related data of any category, record the areas corresponding to the related data of the category as the distribution area of ​​the category, obtain the related data of any attribute of any other map element in the distribution area except the map element, and calculate the first The joint distribution probability of the attribute of this category and the attribute of other map elements is The mean of the joint distribution probability of each category of the attribute and each attribute of other map elements is used as the first The attribute and the basic factor of the other map elements are used to The average of the attribute and the basic factors of all other map elements except this map element is used as the first The basic level of each attribute; The average value of the basic degree of each attribute of the map element is used as the comprehensive basic value of the map element; the map elements are arranged in descending order of the comprehensive basic value, and the obtained sequence is recorded as the map element basic sequence.

4. The intelligent image recognition method for high-precision maps according to claim 1, characterized in that: The specific method of constructing several scales of map elements from small to large in number based on the basic sequence of map elements includes: According to the order of map elements in the basic sequence of map elements, the first scale is constructed with the first map element, and the second scale is constructed with the first and second map elements, and so on, to obtain several scales.

5. The intelligent image recognition method for high-precision maps according to claim 1, characterized in that: The specific method of constructing the map distribution diagram of each scale and obtaining the chain code vector of each edge therein includes: For any scale, for each map element contained in the scale, the map distribution map of the scale is constructed based on the relevant data of each attribute of each map element, combined with its spatial contour information and location information; Obtain several edges in the map distribution map, and treat two completely parallel edges corresponding to the same map element as an edge group; for any edge, obtain several key points on the edge including the starting point, end point, and inflection point, and obtain the chain code of the edge based on the key points; Get the chain code of each edge in the map distribution map of this scale. Based on the chain code of the edge with the largest number of elements in the chain code, get the chain code vector of this edge. Get the chain code vectors for the chain codes of other edges. If the number of chain code elements is insufficient, add 0 at the end of the chain code.

6. The intelligent image recognition method for high-precision maps according to claim 5, characterized in that: The specific method of obtaining the plurality of retention scales includes: Analyze the differences in the similarity relationship between the chain code vectors of the nearest edge at adjacent scales to obtain the neighbor change factor of each edge group at adjacent scales; Get the At the scale, each edge group is A scale and The neighbor change factor of each scale is taken as the average of all neighbor change factors. The scale relative to the Information change characteristics of each scale; The judgment starts from the information change characteristics of the second scale relative to the first scale. If the information change characteristics are less than the change threshold, the second scale is discarded, and the information change characteristics of the third scale relative to the first scale are calculated. If the third scale is retained, the subsequent calculation is based on the information change characteristics of the fourth scale relative to the third scale. And so on. Several scales are discarded based on the information change characteristics, and the remaining scales are retained.

7. The intelligent image recognition method for high-precision maps according to claim 6, characterized in that: The specific method for obtaining the neighbor change factor of each edge group at adjacent scales includes: For the For any edge group of a scale, the chain code vectors of the two edges in the edge group are the same, and are used as the chain code vector of the edge group. The edge closest to the edge group is obtained. Based on the mean Euclidean distance between the key points of the two edges in the edge group and other edges, the edge with the smallest mean Euclidean distance is used as the edge closest to the edge group. The cosine similarity between the chain code vector of the edge group and the chain code vector of the edge closest to the edge group is obtained as the first The nearest neighbor similarity of the edge group at each scale; In the Re-obtain the edge closest to the edge group in the scale, and re-obtain the The nearest neighbor similarity of the edge group at the first scale is obtained; the absolute value of the difference between the nearest neighbor similarities of the edge group at two adjacent scales is obtained as the nearest neighbor similarity of the edge group at the first scale. A scale and The neighbor variation factor of each scale.

8. The intelligent image recognition method for high-precision maps according to claim 5, characterized in that: The specific method of marking a number of nodes includes: The scale with the least number of map elements among the retained scales is taken as the maximum retained scale; Based on the distribution of edges in each edge group in the map distribution diagram of the maximum retained scale, several groups of regional edge group combinations are obtained; For any set of regional edge group combinations, the region is divided based on the set of regional edge group combinations, and the two edge groups constitute a divided region; each edge in the two edge groups is extended until the edges of the two edge groups intersect, and the intersecting part is marked as a node; For the original edges in the two edge groups, for any of the edge groups, a segmentation ratio is preset, and several straight lines perpendicular to two of the edges are obtained. Starting from the starting points of the two edges, a vertical straight line is extracted between the starting points and the next key point at every segmentation ratio as the grid lines in the edge group. Several grid lines are obtained, and the intersections of the grid lines and the two edges are marked as nodes.

9. The intelligent image recognition method for high-precision maps according to claim 8, characterized in that: The specific method of obtaining a plurality of regional edge group combinations includes: For each edge group in the map distribution map of the maximum retained scale, obtain the edge closest to each edge group, and based on the edge group corresponding to the edge closest to the edge group, obtain the edge group closest to each edge group, and obtain several combinations of the edge groups closest to each other; For any edge group combination with the closest distance, obtain the direction difference between the two edge groups with the closest distance. If the direction difference is greater than or equal to the difference threshold, use the edge group combination with the closest distance as a group of regional edge group combinations.

10. The intelligent image recognition method for high-precision maps according to claim 1, characterized in that: The specific method of screening several retained paths of each node includes: For any two paths of a node at adjacent retained scales, obtain the DTW distance of the two paths through DTW matching. Obtain the maximum DTW distance between the paths of the node at all retained scales. Use the ratio of the DTW distance corresponding to the adjacent retained scale to the maximum DTW distance as the path difference factor of the adjacent retained scale. Preset a path difference threshold. If the path difference factor is greater than the path difference threshold, retain the two paths of the node at the adjacent retained scale. If the path difference factor is less than or equal to the path difference threshold, the path of the next retention scale in the adjacent retention scales is removed; the paths retained by the node at each retention scale are used as the several retained paths of the node.

Images