CN109241069B - Road network rapid updating method and system based on track adaptive clustering - Google Patents
Road network rapid updating method and system based on track adaptive clustering Download PDFInfo
- Publication number
- CN109241069B CN109241069B CN201810965568.5A CN201810965568A CN109241069B CN 109241069 B CN109241069 B CN 109241069B CN 201810965568 A CN201810965568 A CN 201810965568A CN 109241069 B CN109241069 B CN 109241069B
- Authority
- CN
- China
- Prior art keywords
- road
- track
- point
- points
- road network
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
Abstract
The invention provides a road network rapid updating method based on track adaptive clustering, which is characterized by comprising the following steps: judging whether the track points are matched with the original road network or not by the acquired moving track data and the acquired original road network data through a distance constraint condition and a direction constraint condition, acquiring unmatched track points through a matching result, carrying out self-adaptive track clustering on the unmatched track points, carrying out curve fitting on the track points by adopting an optimal main curve fitting method aiming at each track cluster, extracting a road center line, identifying the driving direction and the one/two-way information of the road, and further completing the fusion of the changed road and the original road network. By the method, the problems of rapid identification of the changing area of the urban road network in a large range, self-adaptive extraction and update of the fine geometrical structure of the changing road in a complex situation, and identification and update of semantic information such as the driving direction and single/double directions of the road can be solved.
Description
Technical Field
The invention relates to the field of geographic information systems and intelligent traffic research, in particular to a method and a system for quickly updating a road network based on track adaptive clustering.
Background
The road network is basic data for important applications such as travel navigation, logistics distribution and urban planning of people, and is also one of basic geographic information which is highly valued by surveying and mapping, territorial and planning departments. At present, with the rapid development of urban construction in China, urban road networks are frequently changed. In the face of a rapidly changing urban road network, how to guarantee the situation of road network data and realize rapid updating of road network data is a problem which is paid keen attention by departments of current surveying and mapping, traffic, planning and the like. However, in the existing digital mapping method (such as field measurement, remote sensing image vectorization, laser radar data mapping, etc.), a large amount of manpower and material resources are required to be consumed in the data acquisition process, and the data post-processing process is complex and difficult to realize the timely update of the rapidly changing urban road network. The existing urban road network updating mostly adopts a periodic stage updating means, that is, the whole area is comprehensively acquired and updated once at certain time intervals (such as 5 years, 1 year and the like), and the carpet type data updating mode is difficult to meet the updating requirements of various current intelligent traffic navigation applications on real-time road data.
With the rapid development of mobile positioning technology and sensor technology, devices such as buses, taxis and user mobile phones in cities can record information such as positions, directions and movement speeds of vehicles/pedestrians in real time, and the GPS track data for recording travel routes of the vehicles and the users contains rich road information, so that the updating change of urban roads and the change information of traffic states can be sensed in real time. Compared with manual field measurement data, remote sensing image data and laser radar data, the moving track data is a mass-source derived geographic information, has the advantages of high mass, low cost and strong situational property, becomes an important data source for extracting and updating the road network in real time at present, and has been widely concerned by students in the fields of surveying and mapping, traffic, planning and the like. However, the problems of low positioning accuracy, uneven distribution, more noise and the like of the moving track data also bring some challenges to the extraction and update of the high-accuracy road network. For a single piece of movement track data, because of the influence of GPS positioning error, signal occlusion, noise, etc., it is difficult to reflect the real position and geometry of the road through a piece of movement track data. However, when a large amount of track data is superimposed, the information such as the center position of the road, even the width of the road, etc. can be reflected more accurately.
At present, a road network updating method based on movement track data mainly adopts two research strategies: (1) firstly, reconstructing all road network information according to track data, and then comparing the road network data reconstructed through an algorithm with the existing road network data to detect a newly added road; (2) carrying out map matching on the moving track and the existing road network, and detecting unmatched tracks which are regarded as new or changed roads; and then, partial reconstruction of road information is carried out on the unmatched tracks, and the geometric structure and semantic information of the newly added road, such as information of one-way and two-way roads, steering and the like, are extracted.
The first research strategy focuses on extracting geometric structure features of the whole road network from massive moving track data, and mainly comprises a rasterization method, a clustering-based method and an incremental merging method. The rasterization method comprises the steps of converting track points into raster images, and extracting road center lines by using morphological operators; the method based on clustering mainly divides track point data into different clusters through a clustering algorithm, and takes the centers of the clusters as key points to connect or perform curve fitting to generate a road center line; the incremental merging method is to take a certain track as an initial road network, and further gradually add other tracks into the initial road network by a track-road network matching and graph merging method (such as a gravitational-repulsive force model, a Delaunay triangulation network-based space-time track data fusion model, and the like) to generate final road network data. Although these methods can effectively extract the structural information of the entire road network, the existing road network data is not fully utilized, and reconstruction of a large amount of road network information is not practically necessary, and a large amount of computing resources are required, and it is difficult to process a large amount of movement trajectory data. The second strategy focuses on detecting local change information in the existing road network, and the adopted main method is to match the track with the existing road network, detect the change of the road network through unmatched track data, and further extract the structural information of the changed road through clustering (such as K-means algorithm, PAM clustering algorithm and the like), morphological operators (such as expansion and corrosion operators), curve fitting (such as polynomial fitting and B-spline fitting) and other methods aiming at the unmatched tracks. Compared with the method for reconstructing the whole road network, the local updating methods have higher efficiency, can fully utilize the existing road network data and have higher efficiency of data processing and updating. However, because the track data has the characteristics of high noise, uneven distribution density and the like, the conventional road network local updating method needs to rely on more parameter settings (such as parameters of cluster number, morphological algorithm threshold and the like), can only process relatively simple road change information extraction, and is difficult to process the problem of updating the road network in a wide range.
In general, although the current road network data updating method based on moving track has received wide attention from students and different departments, there are still several key problems to be solved in the current road network updating method based on moving track: (1) and excessive dependence on parameter setting. The existing method needs more parameter settings in the clustering and rasterizing processes, and the analysis result is sensitive to the parameter settings, so that the problem of high subjectivity of the road extraction result is caused; for actual data, how to select proper parameters still needs more manual intervention and debugging; (2) and the analysis processing capacity of the existing algorithm to the complex scene is insufficient. The method is mainly characterized in that due to the fact that distribution density of track data has large difference in space, complete information of changed roads is difficult to extract by an existing clustering and rasterizing-based method, and some locally added roads are easy to lose. In addition, the existing method has a rough road geometry structure generated for some more complicated newly added roads (such as T-shaped intersections and groined road networks), and the requirement for updating a high-precision road network is difficult to meet; (3) the existing method mostly focuses on extracting the center line of the road, and extraction and updating of semantic information such as the road traffic direction and the one-way and two-way roads are relatively lacked.
Disclosure of Invention
The invention aims to provide a method for quickly updating a road network based on track adaptive clustering, which can be used for quickly identifying the changing area of a large-range urban road network, adaptively extracting and updating the fine geometrical structure of a changing road under a complex situation, and identifying and updating semantic information such as the driving direction of the road, single/double directions and the like.
In order to solve the problems, the invention adopts the following technical scheme:
a method for quickly updating a road network based on track adaptive clustering is characterized by comprising the following steps: collecting moving track data and acquiring original road network data; judging whether the track points are matched with the original road network or not according to the distance constraint condition and the direction constraint condition, obtaining unmatched track points according to the matching result, carrying out self-adaptive track clustering on the unmatched track points, carrying out curve fitting on the track points by adopting an optimal main curve fitting method aiming at each track cluster, extracting a road central line, identifying the driving direction and the one/two-way information of the road, and further completing the fusion of the changed road and the original road network.
Preferably, the determining whether the locus point is matched with the original road network according to the distance constraint condition and the direction constraint condition specifically includes: calculating the shortest distance between the track point Pi and all road sections in the original road network RD, selecting the road section with the distance from the track point Pi smaller than a threshold value from the RD as a candidate matching road section C _ RD, and marking the track point Pi as an unmatched track point if the candidate matching road section is empty; otherwise, judging the direction constraint condition.
Preferably, the direction constraint condition is that the road segments in the C _ RD are arranged in an ascending order according to the distance Pi, the direction difference between the candidate road segment ek and the track point is sequentially calculated from the first road segment backward, if the direction difference is smaller than a threshold value, Pi is considered to be matched with ek, otherwise, the next candidate matching road segment is continuously selected, the direction difference between the candidate road segment ek and the track point Pi is calculated, whether the candidate road segment is matched with Pi is judged, all the candidate road segments are traversed according to the steps, if Pi does not find a suitable matching road segment, Pi is marked as an unmatched track point, and the optimal master curve fitting method can be a piecewise linear optimal curve fitting method.
Preferably, the identifying the road driving direction and the one/two-way information specifically includes: according to the obtained track point clustering Cj, counting direction distribution histograms of all track points in the Cj, solving mixed Gaussian distribution parameters by adopting an expectation maximization algorithm, fitting a track point direction distribution density curve, identifying local extreme points of the distribution density curve by a hill climbing algorithm, and judging the direction of the road section by the local extreme points.
Preferably, the determining the road section direction by the local extreme point specifically includes: if only one local extreme point is detected, taking the direction corresponding to the extreme point as the driving direction of the newly added or changed road section, and marking the newly added road section as a unidirectional road section; and if two local extreme points are detected simultaneously, taking the directions corresponding to the two extreme points as the driving directions of the newly added or changed road sections, and marking the newly added road sections as bidirectional road sections.
A system for fast updating of road network based on adaptive clustering of tracks is characterized in that the system comprises: the data preprocessing unit is used for acquiring moving track data; the data acquisition unit is used for acquiring original road network data; the track network matching unit is used for judging whether the track points are matched with the original road network or not according to the distance constraint condition and the direction constraint condition and acquiring unmatched track points according to the matching result; the updating unit is used for carrying out self-adaptive track clustering on unmatched track points, carrying out curve fitting on the track points by adopting an optimal main curve fitting method aiming at each track cluster, extracting a road center line and identifying the driving direction and the one/two-way information of the road; and the road network fusion unit is used for completing the fusion of the changed road and the original road network.
Preferably, the determining whether the locus point is matched with the original road network according to the distance constraint condition and the direction constraint condition specifically includes: calculating the shortest distance between the track point Pi and all road sections in the original road network RD, selecting the road section with the distance from the track point Pi smaller than a threshold value from the RD as a candidate matching road section C _ RD, and marking the track point Pi as an unmatched track point if the candidate matching road section is empty; otherwise, judging the direction constraint condition.
Preferably, the direction constraint condition is that the road segments in the C _ RD are arranged in an ascending order according to the distance Pi, the direction difference between the candidate road segment ek and the track point is sequentially calculated from the first road segment backward, if the direction difference is smaller than a threshold value, Pi is considered to be matched with ek, otherwise, the next candidate matching road segment is continuously selected, the direction difference between the candidate road segment ek and the track point Pi is calculated, whether the candidate road segment is matched with Pi is judged, all the candidate road segments are traversed according to the steps, if Pi does not find a suitable matching road segment, Pi is marked as an unmatched track point, and the optimal master curve fitting method can be a piecewise linear optimal curve fitting method.
Preferably, the identifying the road driving direction and the one/two-way information specifically includes: according to the obtained track point clustering Cj, counting direction distribution histograms of all track points in the Cj, solving mixed Gaussian distribution parameters by adopting an expectation maximization algorithm, fitting a track point direction distribution density curve, identifying local extreme points of the distribution density curve by a hill climbing algorithm, and judging the direction of the road section by the local extreme points.
Preferably, the determining the road section direction by the local extreme point specifically includes: if only one local extreme point is detected, taking the direction corresponding to the extreme point as the driving direction of the newly added or changed road section, and marking the newly added road section as a unidirectional road section; and if two local extreme points are detected simultaneously, taking the directions corresponding to the two extreme points as the driving directions of the newly added or changed road sections, and marking the newly added road sections as bidirectional road sections.
The method has simpler parameter setting, can effectively identify the road change area in the urban area, can process relatively complicated newly added areas such as T-shaped intersections and groined road networks, and can effectively extract the traffic direction information of the roads.
Drawings
FIG. 1 is a process flow provided in an embodiment of the present invention;
FIG. 2 is an overview of data used in an embodiment of the present invention;
FIG. 3 illustrates a road change zone identified in an embodiment of the present invention;
FIG. 4 is a diagram illustrating adaptive clustering results of track points according to an embodiment of the present invention;
FIG. 5 shows road centerline fitting results in an embodiment of the present invention;
FIG. 6 shows the recognition result of semantic information such as driving direction of a road in the embodiment of the present invention;
fig. 7 is an updated urban road network result in an embodiment of the invention.
Detailed Description
The following description of the embodiments of the present invention will be made with reference to the accompanying drawings. The following examples are only for illustrating the technical solutions of the present invention more clearly, and the protection scope of the present invention is not limited thereby.
The flow of the technical method provided by the invention is shown in figure 1. Further, in order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. After reading this disclosure, modifications of various equivalent forms of the present invention by those skilled in the art will fall within the scope of the appended claims.
A large-scale city road network quick updating method based on track adaptive clustering mainly comprises the following steps:
1) rapid identification of wide-range urban road network change area
The method and the device realize the quick matching of the track and the road network according to the hierarchical constraints of the distance and the direction, and are used for finding the change or the area to be updated in the road network of a large-scale city. The adopted track-road network matching method belongs to a local map matching method, firstly, a track is regarded as track points sampled at different time points, the vehicle running direction at the positions of the track points is calculated, and then whether the track points are matched with a road network is judged according to the following two constraint conditions:
①, distance constraint, namely the shortest distance between a track point and a road segment is smaller, the probability that the track point is matched with the road network is higher;
② directional constraints, i.e., the more the direction of motion of the vehicle at a point of the trajectory coincides with the direction of travel of a road segment, indicate a greater likelihood that the trajectory matches the road segment.
The specific algorithm steps are as follows:
given trajectory point data GD ═ P1，P2，P3,…，Pi，…，PnFor each trace point, a triplet P is usedi＝{xi，yi，oiDenotes wherein xiAnd yiRepresenting the coordinates of the points of the track in a plane projection coordinate system, oiRepresenting the motion direction of the vehicle at the locus point, segmenting the original road network data according to straight line segments, and recording as RD { e }1，e2，e3，…，ej，…，em}，ejIndicating the j-th road segment in the road network.
1.1) calculating the current track point PiShortest distance to all road segments in road network RD { d1，d2，d3，…，dt，…，dmSelecting distance trace point P from RDiA link smaller than epsilon (epsilon-50 m) is taken as a candidate matching link C _ RD-ej|dt≤ε}；
1.2) if C _ RD is empty, then mark PiIf the point is an unmatched point, otherwise, continuing the next step;
1.3) sorting the links in the candidate link C _ RD in ascending order according to the distance of the distance Pi, and then sequentially calculating the candidate link (marked as e) backwards from the first linkk) And the track point PiThe calculation formula is as follows:
wherein the content of the first and second substances,
representing candidate links ekIn the direction of (a). If Λ < θ (θ ═ cos (π/4)), then P is considerediAnd ekMatching; otherwise, continuing to select the next candidate matching road section, calculating the direction difference between the candidate matching road section and the track point Pi, and judging whether the candidate matching road section is equal to the track point P or notiAnd (6) matching. Traversing all the candidate road sections according to the steps, if PiIf no suitable matching route section has been found, the label P is assignediUnmatched track points;
1.4) repeating the above processes until all track points are processed, and finishing the matching process.
By the algorithm, unmatched track points with the existing road network are identified, and the unmatched track points indicate changed road regions in the existing road network.
2) Varying road geometry extraction based on adaptive trajectory clustering
In order to overcome the excessive dependence of the existing algorithm on parameter setting, the invention designs a track self-adaptive clustering algorithm based on direction constraint, which is used for detecting the geometric structure information of a changing road. The algorithm needs less parameter setting and can extract the fine geometric structure of the road, and the specific implementation steps are as follows:
2.1) constructing a Delaunay triangulation network DT (cleaning points with the same coordinates in advance) according to the spatial position coordinates of the unmatched track points detected in the step 1), and aiming at each track point PiAnd calculating the integral long edge threshold value:
in the formula, Lmean(DT) represents the mean of all side lengths in the triangular mesh, Lstd(DT) represents the standard deviation of all side lengths, Lmean(Pi) represents the sum of point PiMean value of side lengths of directly connected triangulation networks. And then will be reacted with PiDeleting the connected triangulation network edges with the side length larger than the threshold value of the whole long edge to obtain a trimmed connection graph G;
2.2) sequentially judging each edge e in the G, if the direction difference of two track points connected by the e is larger than ω (ω is cos (pi/12)), deleting the edge e, otherwise, keeping the edge e;
2.3) selecting an unmarked track point P from the graph GiAs a new cluster Ck(k-1, 2, …), label PiClass label of k, cluster CkIs initialized to the track point PiI.e.: o (C)k)＝oi(ii) a Search and PiAdjacent unmarked track points PjIf P isjAnd cluster CkIs different in direction of：
①PjAnd CkIn the same direction, i.e. | cos (o)j-o(CK))|≤θ∧|oj-o(CK) If | is less than π/2, then directly react PjAnd CkMerging the new clusters, and calculating the average moving direction of the track points in the clusters as the direction of the clusters;
②Pjand CkIn the reverse direction, i.e. | cos (o)j-o(CK))|≤θ^|oj-o(CK) If | ≧ π/2, PjAnd CkCan not be merged;
③Pjand CkWith perpendicular or crossed directions, i.e. | cos (o)j-o(CK) H) is > θ, then PjAnd CkCan not be merged; continuing to search unmarked track points connected with the new cluster, and judging merging conditions until no track points which can be merged can be found in the cluster;
2.4) step 2.3) is executed iteratively, and when all track points are traversed, clustering is finished. Finally, cleaning clusters with the number of trace points in the clusters less than r (r is 10) and marking the clusters as noise;
and 2.5) for the track point clusters identified in the above steps, marking as Ck ═ Pk { Pk1, Pk2, …, Pkm } (K ═ 1,2, …, K), obtaining a fitting curve of the track point clusters by adopting an optimal main curve fitting method (such as a piecewise linear optimal curve fitting method), and then the fitting curve is the road center line of the corresponding changed road segment. Because the algorithm adopts a method of firstly clustering and then fitting the track points to obtain the road center, the influences of factors such as uneven distribution of the track points, noise and the like can be processed; meanwhile, accurate geometric structure information of the real road section corresponding to the track points can be obtained through main curve fitting of track point clustering.
3) Semantic information identification of road driving direction, single/double directions and the like
And (3) counting the direction distribution histogram of all track points in the Cj for the track point cluster (such as Cj) identified in the step 2). And then, solving the Gaussian mixture distribution parameters by adopting an expectation maximization algorithm, fitting a distribution density curve in the track point direction, and identifying local extreme points of the distribution density curve by using a hill climbing algorithm. Further, the identification of the link semantic information is performed according to the following steps:
①, if the distribution histogram of the track points in Cj is unimodal distribution, that is, only one local extreme point is detected, the direction corresponding to the extreme point is taken as the driving direction of the newly added or changed road section, and the newly added road section is marked as a unidirectional road section;
② if the distribution histogram of the track points in Cj is bimodal, that is, two local extreme points are detected at the same time, the direction corresponding to the two extreme points is taken as the driving direction of the newly added or changed road section, and the newly added road section is marked as a bidirectional road section;
③ if the distribution histogram of the track points in Cj is multi-peak distribution (more than two peak values), that is, multiple local extreme points are detected at the same time, the cluster Cj is used as input, the track point clustering algorithm proposed in step 2) is adopted to continuously divide Cj into sub-clusters in different single directions, and steps ① - ② are adopted to judge the direction and single/double information of the newly added road section.
4) Fusion update of changed road and existing road network
Some topology problems (such as suspension lines and the like) may exist when the center line of the extracted changed road is fused with the original road network, and therefore, the extracted changed road needs to be further subjected to topology processing so as to be effectively fused with the original road network, and the main operations include broken line connection and node fusion.
①, connecting the broken lines, for simple road structure, the method of the invention will gather it into one kind, and then fit to get a complete road center line, for the road with more complex structure, especially the road with complex arc section (such as the arc auxiliary road of the viaduct), it may divide it into different arc sections, lead to getting many discontinuous curves to connect it, through setting up certain distance tolerance value can merge the section with smaller tolerance of end point distance, and then fit the whole curve to get the final road center line;
② node fusion, for the end points of the changed road central line, some changed road sections can not be fused with the original road network (namely, the problem of hanging line occurs), aiming at the problem, the extracted changed road sections are overlapped with the original road network data, and the changed road sections are extended or trimmed to ensure that the changed road sections can be effectively fused with the original road network data.
Meanwhile, a system for quickly updating the road network based on track adaptive clustering is also provided, and the system comprises: the data preprocessing unit is used for acquiring moving track data; the data acquisition unit is used for acquiring original road network data; the track network matching unit is used for judging whether the track points are matched with the original road network or not according to the distance constraint condition and the direction constraint condition and acquiring unmatched track points according to the matching result; the updating unit is used for carrying out self-adaptive track clustering on unmatched track points, carrying out curve fitting on the track points by adopting an optimal main curve fitting method aiming at each track cluster, extracting a road center line and identifying the driving direction and the one/two-way information of the road, and the road network fusion unit is used for completing fusion of a changed road and an original road network.
The following will specifically describe a specific implementation of the present invention in road extraction by taking the road network data update in the green mountainous area of Wuhan city as an example:
2.1) data preprocessing. The movement track data obtained in the embodiment is GPS track data of 6150 taxis in Wuhan city on the whole day on 5 months and 1 days in 2015 year, the sampling interval of the data is 45-60 seconds, and the data is common low-frequency sampling floating car track data. Because random noise and GPS positioning error exist, a large number of error data records and redundant stagnation points exist in data, and the noise points and the redundant points can influence the extraction of road structure information and increase redundant calculation amount. For this purpose, firstly, trace points with a speed less than 5km/h or more than 100km/h in the data are eliminated through speed constraint to obtain the preprocessed data, as shown in fig. 2.
2.2) rapidly finding the changed road. The discovery of the changed road is mainly realized by adopting a matching algorithm of the track points and the original road network. In the matching process, the threshold of the matching algorithm comprises a distance threshold epsilon and a direction difference threshold theta, the threshold is set according to the requirements of urban road design specifications, and experiments show that an understood result can be obtained when epsilon is 50m and theta is 45 degrees, so that the actual application requirements are met. Fig. 3 shows the matching result of the track point and the original road network, wherein the dot points represent the unmatched track points, i.e. the areas where the roads are changed (e.g. new or shifted).
2.3) adaptive clustering of the track points. And (3) carrying out self-adaptive track clustering on unmatched track points, wherein the track point clustering result is shown in figure 4.
2.4) extracting the center line of the changed road. For each track cluster, curve fitting of the track points is performed by adopting an optimal main curve fitting method, and the center line of the changing road is extracted, and the result is shown in fig. 5.
And 2.5) identifying semantic information such as road driving direction, single/double directions and the like. As shown in fig. 6, a histogram, a distribution density fitting curve, and local extreme points of the direction distribution of the track points in one track cluster in fig. 4 are shown, as well as the identified road direction and uni/bi-directional semantic information.
2.6) fusion of the changed road and the original road network data. The road network in the experimental area updated by the technical method provided by the invention is shown in fig. 7, wherein gray roads are updated road sections, and black roads are original road network data.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.
Claims (10)
1. A method for quickly updating a road network based on track adaptive clustering is characterized by comprising the following steps: judging whether the track points are matched with the original road network or not by the acquired moving track data and the acquired original road network data through a distance constraint condition and a direction constraint condition, acquiring unmatched track points through a matching result, carrying out self-adaptive track clustering on the unmatched track points, carrying out curve fitting on the track points by adopting an optimal main curve fitting method aiming at each track cluster, extracting a road center line, identifying the driving direction and the one/two-way information of the road, and further completing the fusion of the changed road and the original road network, wherein the self-adaptive track is gatheredThe method comprises the following specific steps: 1) constructing a Delaunay triangulation network DT according to the detected space position coordinates of unmatched track points, and aiming at each track point PiAnd calculating the integral long edge threshold value:
in the formula, Lmean(DT) represents the mean of all side lengths in the triangular mesh, Lstd(DT) represents the standard deviation of all side lengths, Lmean(Pi) Representation and point PiThe average of the side lengths of the directly connected triangulation networks will be equal to PiDeleting the connected triangulation network edges with the side length larger than the threshold value of the whole long edge to obtain a connection relation G after trimming; 2) sequentially judging each edge e in the G, if the direction difference of two track points connected by the edge e is larger than omega (omega is cos (pi/12)), deleting the edge e, otherwise, keeping the edge e; 3) selecting an unmarked trace point Pi from G as a new cluster Ck(k-1, 2, …), label PiClass label of k, cluster CkIs initialized to the track point PiI.e.: o (C)k)＝oi(ii) a Search and PiAdjacent unmarked track points PjIf P isjAnd cluster CkThe direction difference of (a) satisfies:
①Pjand CkIn the same direction, i.e. | cos (o)j-o(CK))|≤θ∧|oj-o(CK) If | is less than π/2, then directly react PjAnd CkMerging the new clusters, and calculating the average moving direction of the track points in the clusters as the direction of the clusters;
②Pjand CkIn the reverse direction, i.e. | cos (o)j-o(CK))|≤θ^|oj-o(CK) If | ≧ π/2, PjAnd CkCan not be merged;
③Pjand CkWith perpendicular or crossed directions, i.e. | cos (o)j-o(CK) H) is > θ, then PjAnd CkCan not be merged; continuing to search the unmarked track points connected with the new cluster and judging the merging condition until the cluster can not find the combinable track pointsStopping when the track points; 4) and (3) performing iteration, and finishing clustering when all the track points are traversed, wherein theta is cos (pi/4).
2. The method according to claim 1, wherein the determining whether the locus point matches the original road network according to the distance constraint condition and the direction constraint condition specifically comprises: calculating the shortest distance between the track point Pi and all road sections in the original road network RD, selecting the road section with the distance from the track point Pi smaller than a threshold value from the RD as a candidate matching road section C _ RD, and marking the Pi as an unmatched node if the candidate matching road section is empty; otherwise, judging the direction constraint condition.
3. The method as claimed in claim 2, wherein the direction constraint condition is that the road segments in C _ RD are arranged in an ascending order according to the distance Pi, the direction difference between the candidate road segment ek and the track point is sequentially calculated backwards from the first road segment, if the direction difference is smaller than a threshold value, Pi is considered to be matched with ek, otherwise, the next candidate matching road segment is continuously selected, the direction difference between the candidate road segment ek and the track point Pi is calculated, and whether the candidate road segment is matched with Pi is judged, all the candidate road segments are traversed according to the above steps, if Pi still does not find a suitable matching road segment, Pi is marked as an unmatched track point, and the optimal master curve fitting method can be a piecewise linear optimal curve fitting method.
4. The method according to any one of claims 1-3, wherein identifying road traffic direction, one/two-way information specifically comprises: according to the obtained track point clustering Cj, counting direction distribution histograms of all track points in the Cj, solving mixed Gaussian distribution parameters by adopting an expectation maximization algorithm, fitting a track point direction distribution density curve, identifying local extreme points of the distribution density curve by a hill climbing algorithm, and judging the direction of the road section by the local extreme points.
5. The method of claim 4, wherein determining the link direction from the local extremum point comprises: if only one local extreme point is detected, taking the direction corresponding to the extreme point as the driving direction of the newly added or changed road section, and marking the newly added road section as a unidirectional road section; and if two local extreme points are detected simultaneously, taking the directions corresponding to the two extreme points as the driving directions of the newly added or changed road sections, and marking the newly added road sections as bidirectional road sections.
6. A system for fast updating of road network based on adaptive clustering of tracks is characterized in that the system comprises: the data preprocessing unit is used for acquiring moving track data; the data acquisition unit is used for acquiring original road network data; the track network matching unit is used for judging whether the track points are matched with the original road network or not according to the distance constraint condition and the direction constraint condition and acquiring unmatched track points according to the matching result; the updating unit is used for carrying out self-adaptive track clustering on unmatched track points, carrying out curve fitting on the track points by adopting an optimal main curve fitting method aiming at each track cluster, extracting a road center line and identifying the driving direction and the one/two-way information of the road; the road network fusion unit is used for completing fusion of a changed road and an original road network, and the self-adaptive track clustering specifically comprises the following steps: 1) constructing a Delaunay triangulation network DT according to the detected space position coordinates of unmatched track points, and aiming at each track point PiAnd calculating the integral long edge threshold value:
in the formula, Lmean(DT) represents the mean of all side lengths in the triangular mesh, Lstd(DT) represents the standard deviation of all side lengths, Lmean(Pi) Representation and point PiThe average of the side lengths of the directly connected triangulation networks will be equal to PiDeleting the connected triangulation network edges with the side length larger than the threshold value of the whole long edge to obtain a connection relation G after trimming; 2) sequentially judging each edge e in the G, if the direction difference of two track points connected by the edge e is larger than omega (omega is cos (pi/12)), deleting the edge e, otherwise, keeping the edge e; 3) selecting an unmarked trace point Pi from G as a new cluster Ck(k＝1,2, …), the marker PiClass label of k, cluster CkIs initialized to the track point PiI.e.: o (C)k)＝oi(ii) a Search and PiAdjacent unmarked track points PjIf P isjAnd cluster CkThe direction difference of (a) satisfies:
①Pjand CkIn the same direction, i.e. | cos (o)j-o(CK))|≤θ^|oj-o(CK) If | is less than π/2, then directly react PjAnd CkMerging the new clusters, and calculating the average moving direction of the track points in the clusters as the direction of the clusters;
②Pjand CkIn the reverse direction, i.e. | cos (o)j-o(CK))|≤θ∧|oj-o(CK) If | ≧ π/2, PjAnd CkCan not be merged;
③Pjand CkWith perpendicular or crossed directions, i.e. | cos (o)j-o(CK) H) is > θ, then PjAnd CkCan not be merged; continuing to search unmarked track points connected with the new cluster, and judging merging conditions until no track points which can be merged can be found in the cluster; 4) and (3) performing iteration, and finishing clustering when all the track points are traversed, wherein theta is cos (pi/4).
7. The system of claim 6, wherein the determining whether the locus point matches the original road network according to the distance constraint and the direction constraint specifically comprises: calculating the shortest distance between the track point Pi and all road sections in the original road network RD, selecting the road section with the distance from the track point Pi smaller than a threshold value from the RD as a candidate matching road section C _ RD, and marking the track point Pi as an unmatched track point if the candidate matching road section is empty; otherwise, judging the direction constraint condition.
8. The system of claim 7, wherein the direction constraint condition is that the road segments in C _ RD are arranged in an ascending order according to the distance Pi, the direction difference between the candidate road segment ek and the track point is sequentially calculated from the first road segment backward, if the direction difference is smaller than a threshold value, Pi is considered to be matched with ek, otherwise, the next candidate matching road segment is continuously selected, the direction difference between the candidate road segment ek and the track point Pi is calculated, and whether the candidate road segment is matched with Pi is judged, all the candidate road segments are traversed according to the above steps, if Pi still does not find a suitable matching road segment, Pi is marked as an unmatched track point, and the optimal master curve fitting method can be a piecewise linear optimal curve fitting method.
9. The system according to any one of claims 6-8, wherein identifying road traffic direction, one/two-way information specifically comprises: according to the obtained track point clustering Cj, counting direction distribution histograms of all track points in the Cj, solving mixed Gaussian distribution parameters by adopting an expectation maximization algorithm, fitting a track point direction distribution density curve, identifying local extreme points of the distribution density curve by a hill climbing algorithm, and judging the direction of the road section by the local extreme points.
10. The system of claim 9, wherein determining the link direction from the local extremum point comprises: if only one local extreme point is detected, taking the direction corresponding to the extreme point as the driving direction of the newly added or changed road section, and marking the newly added road section as a unidirectional road section; and if two local extreme points are detected simultaneously, taking the directions corresponding to the two extreme points as the driving directions of the newly added or changed road sections, and marking the newly added road sections as bidirectional road sections.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810965568.5A CN109241069B (en) | 2018-08-23 | 2018-08-23 | Road network rapid updating method and system based on track adaptive clustering |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810965568.5A CN109241069B (en) | 2018-08-23 | 2018-08-23 | Road network rapid updating method and system based on track adaptive clustering |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109241069A CN109241069A (en) | 2019-01-18 |
| CN109241069B true CN109241069B (en) | 2020-01-10 |
Family
ID=65068109
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810965568.5A Active CN109241069B (en) | 2018-08-23 | 2018-08-23 | Road network rapid updating method and system based on track adaptive clustering |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109241069B (en) |
Families Citing this family (26)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109949692B (en) * | 2019-03-27 | 2021-03-26 | 腾讯大地通途（北京）科技有限公司 | Road network matching method and device, computer equipment and storage medium |
| CN111858620B (en) * | 2019-04-29 | 2024-03-01 | 北京四维图新科技股份有限公司 | Map data updating method and device |
| CN112183170B (en) * | 2019-07-05 | 2024-04-09 | 浙江宇视科技有限公司 | Pedestrian road network generation method, device and system |
| CN110633342B (en) * | 2019-07-29 | 2022-06-17 | 武汉光庭信息技术股份有限公司 | Lane topology network generation method |
| CN110619351B (en) * | 2019-08-14 | 2022-04-08 | 浙江工业大学 | Vegetable and bird stager site selection method based on improved k-means algorithm |
| CN110750607B (en) * | 2019-09-17 | 2023-03-14 | 西安工程大学 | Road network matching method based on GNSS vehicle track data |
| CN112584316B (en) * | 2019-09-30 | 2024-02-06 | 沈阳美行科技股份有限公司 | Indoor driving positioning method, device, equipment and storage medium |
| CN110942638A (en) * | 2019-12-18 | 2020-03-31 | 福建工程学院 | Method and system for identifying topological connection edge direction of urban road network |
| CN111597285B (en) * | 2020-05-13 | 2023-09-15 | 汉海信息技术(上海)有限公司 | Road network splicing method and device, electronic equipment and storage medium |
| CN111722252B (en) * | 2020-05-14 | 2023-04-11 | 江苏北斗卫星应用产业研究院有限公司 | High-precision operation surface identification method based on track data |
| CN113672688A (en) * | 2020-05-14 | 2021-11-19 | 阿波罗智联(北京)科技有限公司 | Road network data updating method and device and electronic equipment |
| CN111898648A (en) * | 2020-07-07 | 2020-11-06 | 北京三快在线科技有限公司 | Road network graph generation method and device, electronic equipment and computer readable storage medium |
| CN112013862B (en) * | 2020-07-31 | 2022-06-17 | 深圳大学 | Pedestrian network extraction and updating method based on crowdsourcing trajectory |
| CN112131233B (en) * | 2020-08-28 | 2022-11-15 | 北京百度网讯科技有限公司 | Method, apparatus, device and computer storage medium for identifying updated road |
| CN112101527B (en) * | 2020-09-15 | 2023-09-01 | 北京百度网讯科技有限公司 | Method and device for identifying lane change, electronic equipment and storage medium |
| CN112530158B (en) * | 2020-10-27 | 2022-05-13 | 北方信息控制研究院集团有限公司 | Road network supplementing method based on historical track |
| CN112509356B (en) * | 2020-10-30 | 2022-04-05 | 深圳市跨越新科技有限公司 | Vehicle driving route generation method and system |
| CN112732857B (en) * | 2021-01-20 | 2022-04-22 | 腾讯科技（深圳）有限公司 | Road network processing method, road network processing device, electronic equipment and storage medium |
| CN112987052B (en) * | 2021-04-27 | 2021-08-10 | 中南大学 | Rapid graph matching method based on road network section classification |
| CN113033494B (en) * | 2021-04-28 | 2021-09-24 | 温州中纬测绘有限公司 | Surveying and mapping data acquisition system based on geographic spatial information data surveying and mapping |
| CN113393666A (en) * | 2021-05-14 | 2021-09-14 | 广州方纬智慧大脑研究开发有限公司 | Traffic simulation method, device, equipment and medium for commercial vehicle |
| CN113611115B (en) * | 2021-08-06 | 2022-06-21 | 安徽师范大学 | Vehicle track clustering method based on road network sensitive characteristics |
| CN114485609B (en) * | 2021-12-15 | 2023-08-01 | 武汉中海庭数据技术有限公司 | Road skeleton line optimization method and system based on original track |
| CN114168906B (en) * | 2022-02-14 | 2022-05-06 | 威海市房地产测绘中心有限公司 | Mapping geographic information data acquisition system based on cloud computing |
| CN115984279B (en) * | 2023-03-20 | 2023-11-17 | 银河航天(北京)网络技术有限公司 | Path determination method, system, electronic device and storage medium |
| CN117112867B (en) * | 2023-10-24 | 2024-01-09 | 四川国蓝中天环境科技集团有限公司 | Heavy vehicle emission calculation and visualization method based on GPS track data |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101866545A (en) * | 2010-05-11 | 2010-10-20 | 中国科学院软件研究所 | Method for acquiring road network matching track of mobile object |
| CN102368265A (en) * | 2011-10-19 | 2012-03-07 | 北京世纪高通科技有限公司 | Method and device for updating map data |
| CN105825510A (en) * | 2016-03-17 | 2016-08-03 | 中南大学 | Automatic matching method between point of interest and road network |
| CN106383828A (en) * | 2016-08-22 | 2017-02-08 | 四川九洲电器集团有限责任公司 | Drawing method and electronic equipment |
| CN106840176A (en) * | 2016-12-28 | 2017-06-13 | 济宁中科先进技术研究院有限公司 | GPS space-time datas increment road network real-time update and path matching system |
| CN107403550A (en) * | 2016-05-20 | 2017-11-28 | 高德信息技术有限公司 | Transit network collecting method, device and terminal device |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105806351B (en) * | 2014-12-30 | 2019-06-07 | 阿里巴巴集团控股有限公司 | A kind of road information estimation method and device |
| CN104819726B (en) * | 2015-05-15 | 2018-09-11 | 百度在线网络技术（北京）有限公司 | navigation data processing method, device and navigation terminal |
| CN105843919A (en) * | 2016-03-24 | 2016-08-10 | 云南大学 | Moving object track clustering method based on multi-feature fusion and clustering ensemble |
-
2018
- 2018-08-23 CN CN201810965568.5A patent/CN109241069B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101866545A (en) * | 2010-05-11 | 2010-10-20 | 中国科学院软件研究所 | Method for acquiring road network matching track of mobile object |
| CN102368265A (en) * | 2011-10-19 | 2012-03-07 | 北京世纪高通科技有限公司 | Method and device for updating map data |
| CN105825510A (en) * | 2016-03-17 | 2016-08-03 | 中南大学 | Automatic matching method between point of interest and road network |
| CN107403550A (en) * | 2016-05-20 | 2017-11-28 | 高德信息技术有限公司 | Transit network collecting method, device and terminal device |
| CN106383828A (en) * | 2016-08-22 | 2017-02-08 | 四川九洲电器集团有限责任公司 | Drawing method and electronic equipment |
| CN106840176A (en) * | 2016-12-28 | 2017-06-13 | 济宁中科先进技术研究院有限公司 | GPS space-time datas increment road network real-time update and path matching system |
Non-Patent Citations (3)
| Title |
|---|
| "基于步行GPS轨迹的路网提取方法";欧阳鸿、刘建勋等;《计算机与现代化》;20140217(第2期);论文第124-128页 * |
| "基于行车时间的道路时空自相关模型研究";赵玲、王佳璆等;《中南大学学报（自然科学版）》;20121026;第43卷(第10期);论文第4114-4122 * |
| "道路网络空间中移动对象轨迹聚类算法的研究";王雅琳;《中国优秀硕士学位论文全文数据库--信息科技辑》;20120515(第05期);论文第17-44页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109241069A (en) | 2019-01-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109241069B (en) | Road network rapid updating method and system based on track adaptive clustering | |
| Wang et al. | Automatic intersection and traffic rule detection by mining motor-vehicle GPS trajectories | |
| CN110095127B (en) | Hidden Markov model map matching method based on segmentation | |
| Schroedl et al. | Mining GPS traces for map refinement | |
| CN110909788B (en) | Statistical clustering-based road intersection position identification method in track data | |
| Gikas et al. | A novel geodetic engineering method for accurate and automated road/railway centerline geometry extraction based on the bearing diagram and fractal behavior | |
| CN108961758B (en) | Road junction widening lane detection method based on gradient lifting decision tree | |
| CN108845569A (en) | Generate semi-automatic cloud method of the horizontal bend lane of three-dimensional high-definition mileage chart | |
| US11085791B2 (en) | Method, apparatus, and computer program product for on-street parking localization | |
| CN109147323A (en) | A kind of vehicle GPS data processing method for highway passenger and freight transportation indicator-specific statistics | |
| Qiu et al. | Automatic extraction of road networks from GPS traces | |
| CN115131669A (en) | Multi-source data collaborative man-vehicle road network integrated construction method | |
| CN115435798A (en) | Unmanned vehicle high-precision map road network generation system and method | |
| CN116071722A (en) | Lane geometric information extraction method, system, equipment and medium based on road section track | |
| CN110400461A (en) | A kind of road network alteration detection method | |
| CN110675631A (en) | Traffic flow traceability analysis method and system | |
| Zhao et al. | A new method of road network updating based on floating car data | |
| CN113096389B (en) | Multi-source data-based national highway network topology construction method | |
| CN115081505A (en) | Pedestrian network incremental generation method based on walking track data | |
| CN112309118B (en) | Vehicle trajectory calculation method based on space-time similarity | |
| CN113932821A (en) | Track map matching method based on continuous window average direction features | |
| CN114663992B (en) | Multisource data fusion expressway portal positioning method | |
| Pu et al. | Road intersection detection based on direction ratio statistics analysis | |
| CN115309846B (en) | Road network structure identification method based on parallel coefficients | |
| Kumar et al. | A Survey on Localization for Autonomous Vehicles |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |