CN109405840A

CN109405840A - Map data updating method, server and computer readable storage medium

Info

Publication number: CN109405840A
Application number: CN201710714285.9A
Authority: CN
Inventors: 韦薇
Original assignee: ZTE Corp
Current assignee: ZTE Corp
Priority date: 2017-08-18
Filing date: 2017-08-18
Publication date: 2019-03-01
Anticipated expiration: 2037-08-18
Also published as: CN109405840B

Abstract

The embodiment of the present invention provides a kind of map data updating method, server and computer readable storage medium, by being polymerize to obtain the clustering cluster that can characterize road situation corresponding with track to the driving path comprising road and track, and the service condition of road in clustering cluster is determined according to the number of driving trace and date in each clustering cluster, count the validity of each clustering cluster, so that it is determined that in clustering cluster road validity, and according to the validity of road complete map grid in road update.It is not larger that road track number difference under different conditions is utilized in the program, and the feature that can reflect in time in driving trace data occurs for road condition, the variation of road condition is understood by the number of driving path in Statistical Clustering Analysis cluster, it avoids artificial field data from acquiring brought data and updates lag, the problem of map datum timeliness difference, the user experience is improved, while also reducing the problem of a large amount of human resources brought by map data update expend.

Description

Map data updating method, server and computer readable storage medium

Technical field

The present invention relates to technical field of data processing more particularly to a kind of map data updating methods, server and calculating Machine readable storage medium storing program for executing.

Background technique

With the fast development of network technology, the communication technology and geographic information system technology, airmanship is ripe day by day, When user is in a completely strange position, electronic map can help people to differentiate direction, navigation directions destination.Electricity Sub- digital map navigation relative to the traditional approach asked the way to locals, can effectively solve the problem that aphasis with exchange disturbance of understanding institute band The problem come, while the loss of time energy brought by pathfinding can be also saved, good user experience is provided for users, So nowadays people have been increasingly dependent on electronic map in trip.

But due to the reason of urban construction, almost there is road pause to use daily, also have new road open-minded, and Electronic map data all derives from the field survey of Xian Xia team, map manufacturer substantially.If road or road sign board variation , this information is obtained under map manufacturer is online, arranges special messenger to collection in worksite data, then in backstage manual modification Figure.Then user is reminded to carry out software upgrading, completes map data update.This strength manually detected much is unable to catch up with city The speed of construction, so, it is easy to appear update to lag for map datum.Such as electronic map is recommending the road of some user just Again it is being built, and map datum still continues to use legacy data, then the navigation of electronic map is likely to user to bring into " disconnected Parting ", so that user can not arrive at the destination as expected.After mistake occurs in navigation, user needs to spend the time at double Correcting with energy influences brought by navigational error, this is obviously unfavorable for the modeling of the corporate image of user experience and map manufacturer It makes.On the other hand, it relies on artificial information and acquires realization map data update, the requirement to human resources is relatively high, is unfavorable for providing It distributes rationally in source.

So now it is urgent to provide a kind of new map data update scheme, it is artificial to solve to rely in the prior art Information collection realizes that data update caused map data update lag and human resources and expend high problem.

Summary of the invention

Map data updating method, server and computer readable storage medium provided in an embodiment of the present invention, it is main to solve Certainly the technical issues of is: existing map data update scheme, which relies on artificial information's acquisition, leads to map data update lag, map Real-time property is poor and human resources expend high problem.

In order to solve the above technical problems, the embodiment of the present invention provides a kind of map data updating method, comprising:

The driving path that will acquire is mapped in preset map grid, and the driving path includes road and track；

Clustering processing is carried out according to the distance between path for the driving path in each map grid to obtain comprising at least one The clustering cluster of a driving path；

Path number based on driving path in each clustering cluster counts the validity of each clustering cluster；

The road in the map grid is updated according to statistical result.

The embodiment of the present invention also provides a kind of server, and the server includes processor, memory and communication bus；

The communication bus is for realizing the connection communication between processor and memory；

The processor is for executing one or more program stored in memory, to realize map as described above The step of data-updating method.

The embodiment of the present invention also provides a kind of computer storage medium, and the computer-readable recording medium storage has one Or multiple programs, one or more of programs can be executed by one or more processor, it is as described above to realize The step of map data updating method.

The beneficial effects of the present invention are:

The embodiment of the present invention provides a kind of map data updating method, server and computer readable storage medium, service The driving path that device will acquire is mapped in preset map grid, between the driving path in each map grid according to path Distance carries out clustering processing and obtains the clustering cluster comprising at least one driving path；It is then based on driving path in each clustering cluster Path number counts the validity of each clustering cluster, and completes the road in map grid according to statistical result and be updated.It is logical It crosses and the driving path comprising road and track is polymerize to obtain the clustering cluster that can characterize road situation corresponding with track, and The service condition that road in clustering cluster is determined according to the number of driving trace and date in each clustering cluster, counts each clustering cluster Validity, so that it is determined that in clustering cluster road validity, and road in map grid is completed according to the validity of road It updates.It is not larger that road track number difference under different conditions is utilized in the program, and road condition occurs reflection to arrive in time Feature in driving trace data is understood the variation of road condition by the number of driving path in Statistical Clustering Analysis cluster, avoids people Work field data acquires brought data and updates lag, and the problem of map datum timeliness difference, the user experience is improved, simultaneously Also reduce the problem of a large amount of human resources brought by map data update expend.

Other features of the invention and corresponding beneficial effect are described in the aft section of specification, and should be managed Solution, at least partly beneficial effect becomes apparent from the record in description of the invention.

Detailed description of the invention

Fig. 1 is a kind of flow chart for the map data updating method that the embodiment of the present invention one provides；

Fig. 2 is a kind of schematic diagram for being mapped to map grid in the embodiment of the present invention one to driving path；

Fig. 3 is a kind of schematic diagram of two driving traces shown in the embodiment of the present invention one；

Fig. 4 is that driving path carries out a kind of flow chart of clustering processing in map grid in the embodiment of the present invention one；

Fig. 5 clusters driving path according to k- central point clustering processing scheme for what is provided in the embodiment of the present invention one A kind of schematic diagram of processing；

Fig. 6 is in the embodiment of the present invention one according to the path number of driving path in clustering cluster and date Statistical Clustering Analysis cluster A kind of flow chart of validity；

Fig. 7 is a kind of hardware structural diagram of the server provided in the embodiment of the present invention two；

Fig. 8 is a kind of structural schematic diagram of the map data update system provided in the embodiment of the present invention three.

Specific embodiment

In order to make the objectives, technical solutions, and advantages of the present invention clearer, below by specific embodiment knot Attached drawing is closed to be described in further detail the embodiment of the present invention.It should be appreciated that specific embodiment described herein is only used to It explains the present invention, is not intended to limit the present invention.

Embodiment one:

In order to solve map datum present in the map data update scheme by artificial information's acquisition in the prior art Real-time difference and human resources expend high problem, and the present embodiment provides a kind of map data updating methods, right for convenience Understanding of the invention is introduced the map data updating method below in conjunction with Fig. 1:

S102, the driving path that will acquire are mapped in preset map grid.

In the present embodiment, the sequence that driving path is made of two or more points comprising road and rail Mark.Wherein road is the infrastructure current for various trackless vehicles, directional, and track is then vehicle institute way when current The set of the point of diameter.

For server when carrying out map data update, the driving path that can be will acquire is mapped to preset map grid In, it among these include the mapping to road and the mapping to track.Driving path is mapped in map grid, actually also It is that the band of position according to corresponding to the location information of driving path and map grid determines which of driving path part exists In the map grid.Since the location information of road and track is essentially all to be characterized by longitude and latitude, and map grid institute Corresponding region is usually to be characterized by transverse and longitudinal coordinate.So being needed when server maps driving path The transverse and longitudinal coordinate first latitude and longitude coordinates of driving path being converted into standard vertical coordinate system.The shape of map grid can advise Then can also be irregular, but to being easy to implement the mapping of road and track, the map grid in the present embodiment is regular shape Shape, circle, square, rectangle etc., can simplify the coordinate information in region corresponding to determining map grid in this way.

The mapping process of driving path is simply introduced so that map grid is square as an example below:

As shown in Figure 2, it is assumed that a certain driving trace is by the sequence between five anchor points of A, B, C, D, E and each anchor point Line is constituted, and the location information of the driving trace has had been converted into standard transverse and longitudinal coordinate: wherein five anchor points of A to E Coordinate be (1,1), (2,3), (4,4), (7,5) and (8,4).And the shape of map grid is square, and the map grid Two of them are respectively (0,0) and (4,4) to the coordinate of angular vertex.Map net then can be determined to angular vertex according to the two The range of lattice, on this basis, coordinate position are located at the anchor point within the scope of this and belong to the map grid.Correspondingly, at this Each anchor point within map grid and the sequence line between anchor point then constitute the traveling belonged in the map grid Path.

Since track is all to be distributed according to road, and road is then relied in the distribution of different zones in most cases In resident population in region, economical situations such as.So may and road be not present in certain regions, track is also not present, Such as desert, the mountain ridge, sea area etc..Therefore, it when being updated for a map datum, might have in the map grid of part Not only road is not present, but also track is not present, namely is mapped in these map grids without any driving path.For these Map grid does not need then to carry out subsequent clustering processing etc..

The size of to map grid does not do considered critical in the present embodiment, can be by the customized setting of administrative staff.It is logical Often, the size of map grid should distinguish the difference of road and road enough, such as administrative staff combine a large amount of empirical value true Make suitable map grid size.

S104, the driving path in each map grid is obtained according to the distance progress clustering processing between path comprising extremely The clustering cluster of a few driving path.

After driving path is mapped in map grid by server, can for the driving path in the map grid into Row clustering processing.The purpose of clustering processing is for " similar " object to be placed in same cluster, and the object in different clusters is " phase It is different ".The distance between two driving paths characterize the similarity between this two driving paths.Wherein, apart from smaller, then Similarity is higher, conversely, distance is bigger, then the similarity between two driving paths is lower.

In the present embodiment, server can be using the COS distance between two driving paths, warpage distance and average Any one among distance is as distance.Certainly, server can also using between two driving paths COS distance, Warpage distance and average distance obtain distance after being weighted.It is obtained for example, server is calculated according to formula F=aX+bY+cZ The distance between two driving paths, wherein X, Y and Z characterize COS distance between two driving paths, warpage distance peace respectively Equal distance, and a, b, c then characterize respectively COS distance, warpage distance and average distance three calculate apart from when shared power Weight, the sum of a, b, c three are 1.

For two vectors α and β, angle is smaller between two vectors, and included angle cosine value is bigger, the similarity of two vector It is higher.It, can be in the hope of the relationship between the cosine value and two vectors of its angle by the cosine law.

The COS distance between driving path A and driving path B is calculated, is by being vector come real by driving path tissue Existing.One of the method for calculating COS distance is described as follows: being utilized the least square method based on straight line, is carried out Linear Quasi to track It closes, straight path i.e. two vector of two strip directions of generation calculates its included angle cosine value cos θ.COS distance=1- | cos θ|.The value range of COS distance is [0,1].This method is since straight line fitting processing can filter more information, so being A method of it is more coarse.

Be illustrated below to the details of another method for calculating COS distance: driving path A's is flat in map grid Areal coordinate is expressed as { < x₁,y₁>, < x₂,y₂> ..., < x_m,y_m> }, the plane coordinates table of driving path B in map grid It is shown as { < x '₁,y′₁>, < x '₂,yv₂> ..., < x '_n,y′_n> }, two driving paths are then extracted respectively in x and y two The vector of the sequence composition of value on coordinate system, one shares 4 vectors, if m and n are not etc. (such as m < n), carrying out interpolation ensures The number of elements of 4 vectors is equal, and the method for interpolation is to interleave an average value in the maximum two neighboring number of spacing, repeatedly Iteration has n element, i.e. α 1 until 4 vectors, β 1, α 2, β 2, then calculate separately the value composition on the same coordinate system to Amount (such as α¹And α², β 1 and β 2) cosine value cos^θ1And cos^θ2, COS distance=1-max (| cos^θ1|, | cos^θ2|).Cosine The value range of distance is [0,1].

Frechet distance (Fu Leixie distance) or Hausdorff after the common normalization of warpage distance Distance (Hausdorff distance), value range [0,1].

The average distance between driving path A and driving path B is calculated, one of method is described as follows: driving path A and row The similarity for sailing path B is in all places of driving path A approach, and minimum value is small at a distance from the point of driving path B approach In the number a ' of specified threshold；The similarity of driving path B and driving path A is in all places of driving path B approach, with The number b ' for being less than specified threshold apart from minimum value of the point of driving path A approach.Then, driving path A and traveling road are calculated Similarity between diameter B is (a '+b ')/(a+b), and a is the place number of track A approach, and b is the place of track B approach Number ,/indicate division arithmetic.Average distance=the track 1- similarity, value range [0,1].

For example, as shown in Figure 3.Driving path A is denoted as<a1, a2, a3, and a4>, pass through 4 places on the way；Travel road Diameter B is denoted as<b1,b2,b3,b4,b5>, pass through 5 places on the way, specified threshold is set as 2.B1 on a1 distance travel path B Point distance is 1 recently；B2 point distance on a2 distance travel path B is 2 recently；B3 point on a3 distance travel path B away from It is 3 from nearest；B3 point distance on a4 distance travel path B is 2 recently；So Tdis (A, B)=1+1+0+1=3.B1 away from It with a distance from a1 point on driving path A recently, is 1；A1 point distance on b2 distance travel path A is 1 recently；B3 distance row The a4 point distance sailed on the A of path is nearest, is 2；A4 point distance on b4 distance travel path A is 8 recently；B5 distance travel road A4 point distance on diameter A is 3 recently；So Tdis (B, A)=1+1+1+0+0=3.Similarity between A and B is 0.667, Average distance is 0.333.

The process of clustering processing in the present embodiment is simply introduced below.In the present embodiment, due to finally needing Determine the validity of Roads in Maps, so, the result of server clustering processing is preferably able to the use feelings of clearly every road Condition guarantees in each clustering cluster that is, in the ideal case, each road should be dispersed in different clustering clusters by server Road will not more than one.It is introduced below with reference to process of the Fig. 4 to clustering processing:

S402, the distance and global clustering parameter of each driving path cluster each driving path in grid according to the map Processing.

Cluster includes global clustering, by different level cluster, Local Clustering.In global clustering, all map grids are all adopted With identical global clustering parameter.And if necessary to do specially treated for wherein some or certain map grids, such as Part map grid does hierarchical clustering processing, then hierarchical clustering parameter can be arranged for these map grids.And Local Clustering is joined It is several, it is to be needed poly- to some after carrying out clustering processing for the driving path in some map grid and obtaining clustering cluster Driving path in class cluster does used clustering parameter when further clustering processing.It is different using different clustering algorithms Global clustering parameter.Such as in the cluster based on distance threshold, global clustering parameter is cluster threshold value；It is being based on the center k- In the cluster of point, global clustering parameter is the quantity K of clustering cluster.

In the present embodiment, server can realize the cluster to driving path according to the following two kinds mode:

The first, the clustering algorithm based on distance threshold, server calculates the distance two-by-two of driving trace in map grid, It will be compared apart from same cluster threshold value, then the driving trace that distance is less than or equal to the cluster threshold value is gathered and is clustered for one Cluster.But in the program, the size for depending entirely on cluster threshold value can be clustered between two driving traces, this is easy to cause poly- Class number of clusters amount is excessive or very few, is unfavorable for embodying the number of track on road.

Second, based on the clustering algorithm of k- central point, server is using k- central point clustering schemes to the map grid Middle N driving path carries out the independent clustering processing of n times, due to this n times processing be it is independent, server is eventually To n clustering schemes, then server selects optimal solution from this n independent cluster results.

K- central point clustering schemes refer to that when initial clustering, server travels road from the N item for participating in clustering processing The K initial cluster center (i.e. cluster seed) as clustering cluster is randomly choosed in diameter.The size of K determines final cluster gained The number of clusters of clustering cluster, in general, the value of K be close toInteger or the map grid in road segment number integer Times (such as 2 times), this can the map of grid region according to the map improve situation and regional construction situation carries out difference and sets It sets.After selecting initial cluster center, server travels the N-K other than the K driving path as initial cluster center Path is assigned in corresponding clustering cluster according to the distance between each initial cluster center.Refer to Fig. 5, it is assumed that participate in cluster Driving path has a, b, c, d, e, f, so the value of N is 6, then K value can take 2.Server can be at random from this 6 driving paths Select in the middle two as initial cluster center, it is assumed that server selection be b and e.Then, server is distinguished according to a, c, d and f With the distance between b and e, this remaining 4 driving path is assigned in the corresponding clustering cluster of b and the two initial cluster centers e. In general, being directed to a driving path to be allocated, server is to select that initial cluster nearest apart from driving path distance Center clusters the driving path into that clustering cluster at corresponding initial cluster center.In the present embodiment, server passes through It calculates, in driving path a, c cluster to clustering cluster corresponding to the b of initial cluster center, driving path d, f is clustered to initial In clustering cluster corresponding to the e of cluster center.So driving path { a, b, c } constitutes a clustering cluster, and { d, e, f } then becomes another One clustering cluster.

Although the element number in two clustering clusters handled by initial clustering is equal in the example, Be it will be appreciated by those skilled in the art that, in the present embodiment, either at initial clustering or subsequent center iteration It manages, the number of driving path is all not necessarily equal in obtained each clustering cluster.

After the completion of initial clustering, server needs to carry out the iterative processing of m subcenter to K clustering cluster to obtain the k- central point The corresponding cluster result of clustering schemes.Center iterative processing refers to that server can reselect in new cluster for each clustering cluster The heart replaces initial cluster center with new cluster center, then according to except the excentral N-K driving path of new cluster is the same as in these new clusters The distance between heart re-starts cluster.It is worth noting that, being in shape before when server selects new cluster center At clustering cluster in selected, for example, be directed to above-mentioned example, server is two clustering clusters handled in initial clustering It is selected respectively in { a, b, c } and { d, e, f }.For clustering cluster { a, b, c }, server can be selected and other two traveling roads The smallest that of sum of the distance is used as new cluster center between diameter, and clustering cluster { d, e, f } is similarly handled.It is assumed that server selects The new cluster center for selecting out is a and e respectively, then on this basis, server can re-start tetra- driving paths of b, c, d, f Cluster distribution.It is assumed that two clustering clusters this time obtained after replacement cluster center are { a, b, d } and { c, e, f } respectively.Then, it takes Business device can reselect new cluster center for the two clustering clusters to replace a and e, and it is this again to carry out clustering processing ... again New cluster center is selected, and the process for being clustered distribution again to remaining driving path can carry out m times, m is more than or equal to 1.

After carrying out the iterative processing of m subcenter, server can obtain the corresponding cluster knot of the k- central point clustering schemes Fruit.It should be appreciated that server is selected at random from driving path all in map grid at initial selected cluster center It selects.It will be apparent that the driving path as initial cluster center can largely influence the k- central point cluster side Case it is final as a result, so server can be by the independent carry out n times of k- central point clustering schemes, that is to say, that server can will Process shown in Fig. 5 carries out n times.Then optimal solution is selected from the cluster result of n times k- central point clustering schemes.It is so-called Optimal solution can be selected in this way: the collection in the evaluation resulting clustering cluster of every kind of clustering schemes between driving trace Middle degree, the i.e. size of overall distance are selected apart from a kind of the smallest scheme.

On the other hand, it since the track in driving path is mostly to follow road, may be deposited for a road Between a plurality of track, these tracks and track, perhaps there are some difference, but the gap between every track and road is then Can be smaller, therefore, if selecting road as cluster center, it is easier to the clustering processing effect more concentrated.So this reality It applies in order to promote clustering processing effect in example, server, can be according to preset ratio before selection randomly chooses initial cluster center Weight shared by road in map grid is promoted, to promote the probability that road is selected as initial cluster center.For example, will wherein Road b weight promoted 500 times, then road b is easier to be selected as cluster center relative to track a.

S404, judge whether the number of road in clustering cluster obtained by clustered processing is less than or equal to 1.

It may include one or more clustering cluster in a map grid after clustering processing.In order to protect The subsequent validity for capableing of each road of accurate characterization to the judgement result of clustering cluster validity is demonstrate,proved, so, in the present embodiment, clothes It is engaged in after the completion of device needs guarantee cluster, the number of road is no more than one in resulting each clustering cluster, that is, a map grid Only in this way two kinds of middle clustering cluster: having the clustering cluster and the not no clustering cluster of road of a road.So when by above two After scheme is clustered, server can determine whether resulting cluster result meets condition, and e.g., server judges resulting each In clustering cluster, whether the number of road is less than or equal to 1.If it is judged that be it is no, then server execute S406, if judgement knot Fruit is yes, it may be considered that clustering processing is completed.

Continue to carry out further clustering processing to the driving trace in clustering cluster after S406, setting Local Clustering parameter.

If server determines that not only one, road included in, office is can be set in some resulting clustering cluster Further clustering processing (can be described as sub- cluster) is carried out to the element in the clustering cluster again after portion's clustering parameter.For the first For clustering schemes, Local Clustering parameter includes the son cluster threshold value for the clustering cluster, and server can will cluster threshold value Value reduces, so as to be polymerized to a kind of condition harsher for each element in the clustering cluster.For second of clustering schemes, server K- central point clustering schemes can be executed again in the clustering cluster for being unsatisfactory for condition, Local Clustering parameter just includes the clustering cluster The quantity K ' of lower sub- clustering cluster.

After the clustering cluster that server is directed to the condition that is unsatisfactory for has done further clustering processing, need to re-execute S404 is so recycled, until the number of road in the clustering cluster is less than or equal to 1.

In the corresponding example of Fig. 4, server has only judged each when determining whether cluster result meets condition Whether the number of road is less than or equal to 1 in a clustering cluster, this is primarily to guarantee to use up when final determining road validity Possible accuracy is to each road.But in other examples of the present embodiment, server can also come in conjunction with another condition Judge cluster result: server determines in resulting each clustering cluster whether the number for the clustering cluster that road number is 0 is less than Equal in the first preset number and each clustering cluster, it is default whether the number for the clustering cluster that trace number is 0 is less than or equal to second Number.Condition is not satisfied for institute's cluster result, and server needs to adjust global clustering parameter and clusters again.It is mentioned here heavy New cluster is for all driving paths for participating in clustering processing, that is, there is no appoint for current obtained cluster result What meaning.So global clustering parameter includes the cluster threshold value for all clustering clusters for the first clustering processing scheme, And it is directed to second of clustering processing scheme, global clustering parameter includes the quantity K of clustering cluster.

S106, the path number based on driving path in each clustering cluster count the validity of each clustering cluster.

After clustering processing is completed, server can be determined each poly- based on the path number of driving path in each clustering cluster The validity of class cluster.The validity for determining clustering cluster is actually to determine the validity of road corresponding to clustering cluster, is determined Whether road is within statistics period also by normal use.Such as server carries out the path number of each clustering cluster with normal range (NR) Compare, when the path number for determining the clustering cluster is not belonging to normal range (NR), determines that the clustering cluster is expired cluster.It should be understood that It is that normal range (NR) mentioned here can characterize the numberical range of the clustering cluster possessed driving path under normal circumstances.Just Normal range can be to be determined by a thresholding, can also be determined by two thresholdings.In another example of the present embodiment, Server can first determine under normal circumstances, the normal number for the track that road is possessed in clustering cluster, and then determine the cluster Gap between the path number and normal number of cluster assert that the clustering cluster belonged to after determining that gap reaches pre-determined threshold Phase cluster, road therein also belong to expired road.

In a kind of example of the present embodiment, server can be according to the path number of driving path and day in each clustering cluster Phase counts the validity of each clustering cluster.For example, server can come to each clustering cluster in each clustering cluster in conjunction with expiration date each Path number in minimum statistics unit is judged: server according to clustering cluster the path number of minimum statistics unit can be with Determine whether the corresponding road of the clustering cluster is in normal operating condition in the minimum statistics unit, to understand the cluster It the cluster interim time in normal operating condition and the time that can must normally carry current task in statistics, determines in expiration date Whether road corresponding to the clustering cluster can also normal use after phase.

Road is travelled according in each clustering cluster to server in the present embodiment below with reference to a relatively specific example The process that the path number of diameter and date count the validity of each clustering cluster is introduced, and refers to Fig. 6:

S602, the path number of each clustering cluster each minimum statistics unit within statistics period is obtained to determine for each cluster The normal distribution interim in statistics of cluster path number.

Server determines path number of each clustering cluster within statistics period in each minimum statistics unit first, it is assumed here that Statistics period is on July 31,1 day to 2017 July in 2017；Minimum statistics unit is day.Then server can determine whether each clustering cluster July it is daily in the path number that is possessed.Since driving path includes two kinds of road and track, for the time of track, It is obvious that being exactly the date that it is generated.Such as a certain track is that user drove to generate on certain road July 15, Then the date of the track is exactly this minimum statistics unit in July 15.

But road and track are different, and typically, road can all be used, after coming into operation in July daily It allows within 15th user's passage of driving to produce a track, on the other date, can also allow other users current.So road Road is originally without the date.This originally for determine road it is whether expired there is no what influence because determine road whether mistake When the phase, actually track distribution of the road within statistics period is being determined.So as long as can determine the day of each track Phase, and then determine the distribution of track each minimum unit in measurement period, that is, it can determine that the validity of road.But due to this In embodiment, server is to be carried out according to driving paths all in clustering cluster, also when determining clustering cluster validity It is to say, when server determines whether clustering cluster is effective, not only considers the generation time of track in driving path, further account for The date on road.For the ease of server to track and road " making no exception ", the server can be allowed to be by the way that rule is artificially arranged One date of link allocation.But when server is the link allocation date, the date of road should be avoided to clustering cluster Validity has an impact, so, in the present embodiment, it is earlier by the setting of date of road to can control server, such as The from date or server for being arranged to statistics period can first determine earliest one day of track date in clustering cluster, then The date of road is arranged also more forward than the earliest date.

Due to the trip quantity on every road per diem, by week, monthly be in normal distribution；Similarly, each clustering cluster uplink The path number for sailing path is also in normal distribution, so, server, can be true after the path number for obtaining each clustering cluster Make normal distribution of the driving path within statistics period in the clustering cluster.

S604, the zone of reasonableness that path number is determined according to the normal distribution of each clustering cluster.

Positive normal open of the every road in minimum statistics unit is directly determined based on experience value with what is introduced in aforementioned exemplary The scheme of row amount is different, in this example, server can in determining clustering cluster driving trace within statistics period just After state distribution, the zone of reasonableness of path number is determined based on the normal distribution.Here two kinds of determining path numbers are provided to close Manage the scheme of range:

Scheme one: server determines the mean value and standard deviation of the path number normal distribution of driving path, then will [ Be worth -2 times of standard deviations ,+2 times of standard deviations of mean value] range as zone of reasonableness.

Scheme two: server will determine the quartile of the normal distribution, then incite somebody to action [first quartile -1.5*IQR, Third quartile+1.5*IQR] range as zone of reasonableness, wherein IQR is the very poor of quartile, and value is equal to third Quartile subtracts the value of first quartile.

For standardized normal distribution, the zone of reasonableness that above two scheme determines can include 75% minimum statistics unit Path number, remaining 25% be used as outlier.It should be understood that working as number of path corresponding to a certain minimum statistics unit Mesh is outlier, it may be possible to because of any one in two such reason: first, path corresponding to the minimum statistics unit Number is few, deviates from normal range (NR), and another kind is because the path number of the minimum statistics unit is too many, most so as to cause this Small statistic unit deviates from the zone of reasonableness that the path number of 75% minimum statistics unit is constituted.So determining clustering cluster When validity, server should not directly think road minimum statistics unit corresponding to all outliers of the clustering cluster In be all off and use, for the first case, it may be said that the bright road in the minimum statistics unit does not use substantially, It is in discarded state, but another situation, can only illustrate that the road of the clustering cluster may be in the minimum statistics unit It due to traffic control, opens the external causes such as new road and introduces new vehicle flowrate, so traffic volume increases severely.

For example, it is [800,1500] that server, which is the zone of reasonableness that a certain clustering cluster path number is determined, period is counted It is still July, it is assumed that in July, there are 27 days path numbers all to fall in the zone of reasonableness.But 15~July of July In this three days on the 17th, the path number of the clustering cluster all far surpasses 1500, reaches 3000 or so.According to description above, clothes Business device cannot directly assert that the road of clustering cluster there is a problem during 15~July 17 July, because, in this three days, It may be or because of traffic control, to lead to substitution associated with corresponding road since failure occur in other roads Road travel amount increases severely.Although this is not belonging to the condition of production, it will be apparent that the use state of road is normal.

S606, the minimum statistics unit of date the latest in zone of reasonableness is determined, and using its date as clustering cluster The nearest reasonable date, and a minimum statistics unit using the date in clustering cluster the latest path number as clustering cluster Nearest path number.

After the zone of reasonableness for determining path number, server also it needs to be determined that out in zone of reasonableness the date the latest one A minimum statistics unit, and using the date as the nearest reasonable date.For clustering cluster X, the meaning on nearest rationally date It is, within statistics period, clustering cluster X driving path number reasonable last day, namely within statistics period, the clustering cluster Traffic volume normal last day on road corresponding to C.It is assumed that being directed to clustering cluster X, in July, 2017, date in zone of reasonableness One day the latest is July 26, that is to say, that in remaining 27 days~31, daily driving path number is not just In normal range.The path number of driving path in the past few days may be the minimum value less than zone of reasonableness, it is also possible to be more than The maximum value of zone of reasonableness.

So why not in the reasonable scope for determining each minimum statistics unit not in the reasonable scope, this reality It applies in example, server can also obtain the minimum statistics unit of date the latest in clustering cluster, for the statistics in July, cluster The date, a minimum statistics unit the latest was exactly July 31 in cluster.Server can be by the minimum statistics list of the date the latest For the path number of member as nearest path number, such as July, nearest path number is exactly the path of 31 this day of July Number.

S608, the nearest reasonable date of each clustering cluster is compared with expiration date, and nearest path number is same The minimum value of zone of reasonableness is compared.

It gets recently rationally after date and nearest path number, server can will same expiration date on reasonable date recently It is compared, while nearest path number being compared with the minimum value of zone of reasonableness.It is closed if nearest path number is less than The minimum value for managing range, then illustrate why the minimum statistics unit peels off, and is because its path number is too small.For nearest The reasonable date with expiration date comparison, if rationally the date earlier than expiration date, illustrates, reaches it in expiration date recently Before, the path number of the clustering cluster is with regard to abnormal, not in the reasonable scope；Conversely, then explanation has had been subjected to expiration date, it should Road corresponding to clustering cluster still possesses normal traffic volume.

It is understood that the expiration date in the present embodiment be for measuring one of whether effective factor of clustering cluster, The zone of reasonableness of expiration date and path number, which combination defines, determines the whether effective condition of clustering cluster: if a clustering cluster was Phase cluster, then there are path numbers to be less than zone of reasonableness most in its some or certain minimum statistics unit at least within statistics period The case where small value.And expiration date then defines on the basis of the last one the minimum statistics unit for counting period, needs more When few continuous minimum statistics unit is all satisfied condition of the path number less than zone of reasonableness minimum value, the cluster just can determine that Cluster is expired.For example, the minimum value of zone of reasonableness is 500, statistics period is in July, 2017, and expiration date is July 26.If one A clustering cluster is expired cluster, then it at least should be guaranteed that the path number in this 6 day time on July 26 to July 31 does not surpass Cross 500.

The setting of expiration date can be set as needed by map vendors, for example, if will be set as expiration date The last day of statistical time range, then decision condition is more stringent, if statistical time range last day road occupation state not Normal road is identified as abnormal state；If set expiration date threshold value to one week of statistical time range last day Preceding that day, then only those road occupation state all abnormal roads daily within the last week of statistical time range last day Road is just identified as abnormal state.

S610, expired cluster in map grid and effective cluster are determined according to comparison result.

If rationally the date, path number was less than zone of reasonableness earlier than expiration date, and recently recently for server judgement Minimum value then can be determined that the clustering cluster is expired.For a map grid, wherein may include multiple clustering clusters, needle All clustering clusters in map grid, server can carry out above-mentioned processing respectively, so that it is determined that corresponding clustering cluster has Effect property.After determining the expired cluster in a map grid, so that it may which remaining clustering cluster is determined as effective cluster.

S108, it is updated according to the road in statistical result to map grid.

Finally, server can be updated according to the road in the Usefulness Pair map grid of clustering cluster: by expired cluster Corresponding road is arranged to expired road, characterizes the road and is no longer used, and in discarded state, (certainly, this, which is discarded, may be Temporary, but cut-off statistics period terminates, and which wouldn't provide use).For expired road, server can by its from this It is deleted in map grid.But, in order to embody the difference after map rejuvenation, server can continue to retain in map grid The road, only adds overdue indicator for it, and for example, expired road is configured at the different display color of effective road.This Sample user can recognize that corresponding position is implicitly present in a road when seeing map, but cannot lead at present Row.

Specifically, road update request message content include grid block id, clustering cluster id and clustering cluster tracking quantity and Annual distribution, track identification id, the expired road for modifying type (new added road, expired road), expired road or new added road Or in the per day trace number of new added road, eight grids adjacent with the grid relevant cluster cluster and relative trajectory information Deng convenient for completing section editor and the verifying in grid between grid.

By screening road expired in map grid, electronic map is avoided to continue according to past road For user's planning path, and then enter into some " dead end highways " that can not be arrived at the destination.Electronic map is screening out expired road On the basis of, plan that guidance path, the guidance path cooked up may be than not screening out the navigation road of expired road originally for user More a little further, user needs to take more time and gets to destination diameter, but this is relative to entering into after expired road again Planning path is far better.

In the present embodiment, there is likely to be some clustering clusters in map grid, only track does not have road.This poly- Belong to those of effective cluster clustering cluster in class cluster, be more special one kind, is i.e. " newly-increased cluster ".There is no road in newly-increased cluster, but There is track, this explanation does not represent corresponding road on the electronic map of master, but there is current track here, And road state be it is normal, therefore, corresponding position may be newly to have opened road.So server can be by newly-increased road Road is arranged to " new added road ", and is identified on the electronic map of new edition.It is newly-increased road to allow user to understand the road Road, server the display color of new added road can be arranged and expired road and script existing for road it is all different.

Electronic map is user's planning path on the basis of identifying new added road, can allow meet for user's planning and work as The shortest path of lower reality road conditions, is saved the travel time of user, and then promote user experience.

Map data update scheme provided in this embodiment, due to individually being handled for each map grid, And for a electronic map, data volume to be treated is very big when cluster, availability deciding, road update , therefore, server can use HDFS (Hadoop distributed file system) scheme by the map datum of different map grids Processing, which is assigned on different computing units (such as server or computer), to be executed, to promote the data of electronic map entirety Treatment effeciency.

Map data updating method provided in an embodiment of the present invention, is handled by map grid, by a electronic map Data processing resolve into the data processing of multiple map grids, thus allow multiple computing units with processing capacity simultaneously it is right Different map grids carries out data processing, improves data-handling efficiency, alleviates the place of individual server or calculator Manage pressure.Meanwhile server can be directly based upon in the map grid when carrying out data processing to a certain map grid Driving path carry out, avoid and arrange special messenger to carrying out the brought high-cost problem of data acquisition on the spot.It is prior It is that, for every road, track thereon can most characterize its use state, and track is to update constantly, therefore, server It is updated according to the data that driving path carries out map grid, enables to update result more comprehensively, while real-time is also more preferable, keeps away Exempting from electronic map update lag gives user's bring various troubles, and the user experience is improved, maintains the enterprise of map vendors Image.

Embodiment two:

Since the map data updating method in embodiment one can be realized by computer program, this implementation Example a kind of computer readable storage medium is provided, the computer readable storage medium can store at least one computer program with It is executed for processor, to realize corresponding process flow.In the present embodiment, it is just stored in the computer readable storage medium There is map data update program, which executes the map that can be realized the offer of embodiment one after being read out by the processor, compiling Data-updating method.In some other example of the present embodiment, road can also be stored in computer readable storage medium Data, track data etc..

The server in previous embodiment is introduced below, the hardware configuration for referring to the server of Fig. 7 offer shows It is intended to:

Server 70 includes processor 71, memory 72 and communication bus 73, and wherein communication bus 73 is for realizing place Manage the connection communication between device 71 and memory 72.It is understood that memory 72 is used as a kind of computer-readable storage medium Matter, wherein being stored at least one computer program, these computer programs can read, compile and execute for processor 71, To realize corresponding process flow.For example, in the present embodiment, map data update program is stored in memory 72, locate Reason device 71 can realize the map data updating method introduced in previous embodiment by executing the computer program.

The driving path that processor 71 can will acquire first is mapped in preset map grid.In the present embodiment, row Sailing path is the sequence being made of two or more points comprising road and track.Wherein road is i.e. for various The current infrastructure of trackless vehicle, it is directional, and track is then set of the vehicle in the point of current when institute approach.

For processor 71 when carrying out map data update, the driving path that can be will acquire is mapped to preset map grid In, it among these include the mapping to road and the mapping to track.Driving path is mapped in map grid, actually also It is that the band of position according to corresponding to the location information of driving path and map grid determines which of driving path part exists In the map grid.Since the location information of road and track is essentially all to be characterized by longitude and latitude, and map grid institute Corresponding region is usually to be characterized by transverse and longitudinal coordinate.So being needed when processor 71 maps driving path The transverse and longitudinal coordinate that first latitude and longitude coordinates of driving path are converted into standard vertical coordinate system.The shape of map grid can be with Rule can also be irregular, but to being easy to implement the mapping of road and track, the map grid in the present embodiment is rule Shape, circle, square, rectangle etc., can simplify the coordinate information in region corresponding to determining map grid in this way.

Since track is all to be distributed according to road, and road is then relied in the distribution of different zones in most cases In resident population in region, economical situations such as.So may and road be not present in certain regions, track is also not present, Such as desert, the mountain ridge, sea area etc..Therefore, it when being updated for a map datum, might have in the map grid of part Not only road is not present, but also track is not present, namely is mapped in these map grids without any driving path.For these Map grid, processor 71 do not need then to carry out subsequent clustering processing etc..

It, can be for the driving path in the map grid after driving path is mapped in map grid by processor 71 Carry out clustering processing.The purpose of clustering processing is for " similar " object to be placed in same cluster, and the object in different clusters is " different ".The distance between two driving paths characterize the similarity between this two driving paths.Wherein, distance is got over Small, then similarity is higher, conversely, distance is bigger, then the similarity between two driving paths is lower.

In the present embodiment, processor 71 can be using the COS distance between two driving paths, warpage distance peace Any one among distance is as distance.Certainly, processor 71 can also be using to the cosine between two driving paths Distance, warpage distance and average distance obtain distance after being weighted.For example, processor 71 is according to formula F=aX+bY+cZ It calculates and obtains the distance between two driving paths, wherein X, Y and Z characterize COS distance between two driving paths, warpage respectively Distance and average distance, and a, b, c then characterize respectively COS distance, warpage distance and average distance three calculate apart from when Shared weight, the sum of a, b, c three are 1.

Be illustrated below to the details of another method for calculating COS distance: driving path A's is flat in map grid Areal coordinate is expressed as { < x₁,y₁>, < x₂,y₂> ..., < x_m,y_m> }, the plane coordinates table of driving path B in map grid It is shown as { < x '₁,y′₁>, < x '₂,y′₂> ..., < x '_n,y′_n> }, two driving paths are then extracted respectively in x and y two The vector of the sequence composition of value on coordinate system, one shares 4 vectors, if m and n are not etc. (such as m < n), carrying out interpolation ensures The number of elements of 4 vectors is equal, and the method for interpolation is to interleave an average value in the maximum two neighboring number of spacing, repeatedly Iteration has n element, i.e. α 1 until 4 vectors, β 1, α 2, β 2, then calculate separately the value composition on the same coordinate system to Measure the cosine value cos of (such as α 1 and α 2, β 1 and β 2)^θ1And cos^θ2, COS distance=1-max (| cos^θ1|, | cos^θ2|).Cosine The value range of distance is [0,1].

Frechet distance (Fu Leixie distance) or Hausdorff after the common normalization of warpage distance Distance, value range [0,1].

The process of clustering processing in the present embodiment is simply introduced below.In the present embodiment, due to finally needing Determine the validity of Roads in Maps, so, the result of 71 clustering processing of processor is preferably able to the use of clearly every road Situation guarantees each poly- that is, in the ideal case, processor 71 should will be dispersed in different clustering clusters when each road Road in class cluster will not more than one.The process for carrying out clustering processing to processor 71 below is introduced:

Firstly, processor 71 according to the map in grid the distance and global clustering parameter of each driving path to each driving path Carry out clustering processing.Cluster includes global clustering, by different level cluster, Local Clustering.In global clustering, all map grids All use identical global clustering parameter.And if necessary to do specially treated for wherein some or certain map grids, such as Hierarchical clustering processing is done for part map grid, then hierarchical clustering parameter can be set for these map grids.And part is poly- Class parameter is then to need after carrying out clustering processing for the driving path in some map grid and obtaining clustering cluster to certain Driving path in a clustering cluster does used clustering parameter when further clustering processing.Had using different clustering algorithms Different global clustering parameters.Such as in the cluster based on distance threshold, global clustering parameter is cluster threshold value；It is being based on k- In the cluster of central point, global clustering parameter is the quantity K of clustering cluster.In the present embodiment, server can be according to following two Kind mode realizes the cluster to driving path:

The first, the clustering algorithm based on distance threshold, processor 71 calculates in each map grid driving trace two-by-two Distance will be compared apart from same cluster threshold value (i.e. global clustering parameter), and distance is then less than or equal to the cluster threshold value Driving trace gathers for a clustering cluster.But in the program, it can be clustered between two driving traces and depend entirely on cluster threshold The size of value, this is easy to cause clustering cluster quantity excessive or very few, is unfavorable for embodying the number of track on road.

Second, based on the clustering algorithm of k- central point, processor 71 is using k- central point clustering schemes to the map net N driving path carries out the independent clustering processing of n times in lattice, due to the processing of this n times be it is independent, processor 71 is final N clustering schemes can be obtained, then processor 71 selects optimal solution from this n independent cluster results.

K- central point clustering schemes refer to that when initial clustering, processor 71 is travelled from the N item for participating in clustering processing The K initial cluster center (i.e. cluster seed) as clustering cluster is randomly choosed in path.The size of K determines final cluster institute Clustering cluster number of clusters, in general, the value of K be close toInteger or the map grid in road segment number it is whole Several times (such as 2 times), this can improve situation according to the map of the map grid region and regional construction situation carries out Difference setting.After selecting initial cluster center, processor 71 will be other than the K driving path as initial cluster center N-K driving path is assigned in corresponding clustering cluster according to the distance between each initial cluster center.Refer to Fig. 5, it is assumed that ginseng There are a, b, c, d, e, f with the driving path of cluster, so the value of N is 6, then K value can take 2.Processor 71 can at random from this 6 Select in driving path two as initial cluster center, it is assumed that the selection of processor 71 is b and e.Then, processor 71 According to a, c, d and f respectively with the distance between b and e, this remaining 4 driving path is assigned to the two initial cluster centers b and e In corresponding clustering cluster.In general, being directed to a driving path to be allocated, server is to select apart from driving path distance most That close initial cluster center clusters the driving path into that clustering cluster at corresponding initial cluster center.In this implementation In example, processor 71 in driving path a, c cluster to clustering cluster corresponding to the b of initial cluster center, will be travelled by calculating In path d, f cluster to clustering cluster corresponding to the e of initial cluster center.So driving path { a, b, c } constitutes a clustering cluster, And { d, e, f } then becomes another clustering cluster.

After the completion of initial clustering, processor 71 needs to carry out the iterative processing of m subcenter to K clustering cluster to obtain the center k- The corresponding cluster result of point clustering schemes.Center iterative processing is that finger processor 71 can reselect newly for each clustering cluster Cluster center replaces initial cluster center with new cluster center, then new with these according to the excentral N-K driving path of new cluster is removed The distance between cluster center re-starts cluster.It is worth noting that, being at it when processor 71 selects new cluster center It is selected in preceding established clustering cluster, for example, being directed to above-mentioned example, processor 71 is two handled in initial clustering It is selected respectively in a clustering cluster { a, b, c } and { d, e, f }.For clustering cluster { a, b, c }, processor 71 can select and other The smallest that of sum of the distance is used as new cluster center between two driving paths, and clustering cluster { d, e, f } is similarly handled.It is false Determining the new cluster center that processor 71 chooses is a and e respectively, then on this basis, processor 71 can be to tetra- rows of b, c, d, f It sails path and re-starts cluster distribution.It is assumed that this time after replacement cluster center obtained two clustering clusters be respectively { a, b, d } and {c,e,f}.Then, processor 71 can reselect new cluster center for the two clustering clusters to replace a and e, be gathered again Class processing ... is this to reselect new cluster center, and the process for being clustered distribution again to remaining driving path can be into Row m times, m are more than or equal to 1.

After carrying out the iterative processing of m subcenter, processor 71 can obtain the corresponding cluster of k- central point clustering schemes As a result.It should be appreciated that at initial selected cluster center, processor 71 be from driving path all in map grid with What machine selected.Gather it will be apparent that the driving path as initial cluster center can largely influence the k- central point Class scheme it is final as a result, so processor 71 can be by the independent carry out n times of k- central point clustering schemes, that is to say, that processing Process shown in Fig. 4 can be carried out n times by device 71.Then it is selected most from the cluster result of n times k- central point clustering schemes Excellent solution.So-called optimal solution can be selected in this way: be travelled in the evaluation resulting clustering cluster of every kind of clustering schemes Intensity between track, the i.e. size of overall distance are selected apart from a kind of the smallest scheme.

On the other hand, it since the track in driving path is mostly to follow road, may be deposited for a road Between a plurality of track, these tracks and track, perhaps there are some difference, but the gap between every track and road is then Can be smaller, therefore, if selecting road as cluster center, it is easier to the clustering processing effect more concentrated.So this reality It applies in order to promote clustering processing effect in example, processor 71, can be according to default ratio before selection randomly chooses initial cluster center Example promotes weight shared by road in map grid, to promote the probability that road is selected as initial cluster center.For example, by it In road b weight promoted 500 times, then road b is easier to be selected as cluster center relative to track a.

It may include one or more clustering cluster in a map grid after clustering processing.In order to protect The subsequent validity for capableing of each road of accurate characterization to the judgement result of clustering cluster validity is demonstrate,proved, so, in the present embodiment, place It manages the needs of device 71 and guarantees after the completion of clustering that the number of road is no more than one in resulting each clustering cluster, that is, a map net Only in this way two kinds of clustering cluster in lattice: there are the clustering cluster and the not no clustering cluster of road of a road.So when passing through above-mentioned two After kind scheme is clustered, processor 71 can determine whether resulting cluster result meets condition, and e.g., processor 71 judges institute In each clustering cluster obtained, whether the number of road is less than or equal to 1.If it is judged that be it is no, then processor 71 adjusts part Continue to carry out further clustering processing to the driving trace in clustering cluster after clustering parameter.If the determination result is YES, then can recognize For clustering processing completion.

If processor 71 determine some resulting clustering cluster when included in not only one, road, it is adjustable Further clustering processing (referred to as son clusters) is carried out to the element in the clustering cluster again after Local Clustering parameter.For the first For clustering schemes, Local Clustering parameter includes the son cluster threshold value for the clustering cluster, and processor 71 can will cluster threshold value Value reduce so as to be polymerized to a kind of condition harsher for each element in the clustering cluster.For second of clustering schemes, processing Device 71 can execute k- central point clustering schemes again in the clustering cluster for being unsatisfactory for condition, and Local Clustering parameter is just poly- including this The quantity K ' of sub- clustering cluster under class cluster.

After the clustering cluster that processor 71 is directed to the condition that is unsatisfactory for has done further clustering processing, need to rejudge each Whether the number of road is less than or equal to 1 in a sub- clustering cluster, so recycles, until the number of road is less than or equal in the clustering cluster Until 1.

In above-mentioned example, processor 71 has only judged each when determining whether cluster result meets condition Whether the number of road is less than or equal to 1 in clustering cluster, this primarily to guarantee it is final can to the greatest extent can when determining road validity Each road can be accurate to.But in other examples of the present embodiment, processor 71 can also come in conjunction with another condition Judge cluster result: processor 71 determines in resulting each clustering cluster whether the number for the clustering cluster that road number is 0 is small In being equal in the first preset number and each clustering cluster, it is pre- whether the number for the clustering cluster that trace number is 0 is less than or equal to second If number.Condition is not satisfied for institute's cluster result, and processor 71 needs to adjust global clustering parameter and clusters again.It is referred to herein Again cluster be for it is all participate in clustering processings driving paths for, that is, currently obtained cluster result no longer Have in all senses.So global clustering parameter includes the cluster threshold for all clustering clusters for the first clustering processing scheme Value, and it is directed to second of clustering processing scheme, global clustering parameter includes the quantity K of clustering cluster.

After clustering processing is completed, processor 71 can be determined each based on the path number of driving path in each clustering cluster The validity of clustering cluster.The validity for determining clustering cluster is actually to determine the validity of road corresponding to clustering cluster, is determined Whether road is within statistics period also by normal use.Such as processor 71 is by the same normal range (NR) of the path number of each clustering cluster It is compared, when the path number for determining the clustering cluster is not belonging to normal range (NR), determines that the clustering cluster is expired cluster.It should manage Solution, normal range (NR) mentioned here can characterize the numerical value model of the clustering cluster possessed driving path under normal circumstances It encloses.Normal range (NR) can be to be determined by a thresholding, can also be determined by two thresholdings.In another example of the present embodiment In the middle, processor 71 can first determine under normal circumstances, the normal number for the track that road is possessed in clustering cluster, and then determine Gap between the path number and normal number of the clustering cluster assert the clustering cluster after determining that gap reaches pre-determined threshold Belong to expired cluster, road therein also belongs to expired road.

In a kind of example of the present embodiment, processor 71 can according to the path number of driving path in each clustering cluster and Date counts the validity of each clustering cluster.For example, processor 71 can come to each clustering cluster in conjunction with expiration date in each clustering cluster Path number in each minimum statistics unit is judged: processor 71 according to clustering cluster minimum statistics unit number of path Mesh can determine whether the corresponding road of the clustering cluster is in normal operating condition in the minimum statistics unit, to understand It the clustering cluster interim time in normal operating condition and the time that can must normally carry current task in statistics, determines Whether road corresponding to the clustering cluster can also normal use after expiration date.

Processor 71 in the present embodiment is travelled according in each clustering cluster below with reference to a relatively specific example The process that the path number in path and date count the validity of each clustering cluster is introduced:

Processor 71 determines path number of each clustering cluster within statistics period in each minimum statistics unit first, false here Surely statistics period is on July 31,1 day to 2017 July in 2017；Minimum statistics unit is day.Then processor 71 can determine whether each poly- Class cluster July it is daily in the path number that is possessed.Since driving path includes two kinds of road and track, for track when Between, it is evident that it is exactly the date that it is generated.Such as a certain track is that user drove to generate July 15 on certain road , then the date of the track is exactly this minimum statistics unit in July 15.

But road and track are different, and typically, road can all be used, after coming into operation in July daily It allows within 15th user's passage of driving to produce a track, on the other date, can also allow other users current.So road Road is originally without the date.This originally for determine road it is whether expired there is no what influence because determine road whether mistake When the phase, actually track distribution of the road within statistics period is being determined.So as long as can determine the day of each track Phase, and then determine the distribution of track each minimum unit in measurement period, that is, it can determine that the validity of road.But due to this In embodiment, processor 71 when determining clustering cluster validity, is carried out according to driving paths all in clustering cluster, That is the generation time of track in driving path is not only considered when processor 71 determines whether clustering cluster is effective, also nationwide examination for graduation qualification Consider the date of road.For the ease of processor 71 to track and road " making no exception ", place can be allowed by the way that rule is artificially arranged Reason device 71 is one date of link allocation.But when processor 71 is the link allocation date, the date of road should be avoided The validity of clustering cluster is had an impact, so, in the present embodiment, it can control processor 71 for the date of road and ratio be set It is more forward, such as be arranged to the from date in statistics period or processor 71 and can first determine that the track date is most in clustering cluster Then early one day the date of road is arranged also more forward than the earliest date.

Due to the trip quantity on every road per diem, by week, monthly be in normal distribution；Similarly, each clustering cluster uplink The path number for sailing path is also in normal distribution, so, processor 71, can be with after the path number for obtaining each clustering cluster Determine normal distribution of the driving path within statistics period in the clustering cluster.

Then, processor 71 determines the zone of reasonableness of path number according to the normal distribution of each clustering cluster.And aforementioned exemplary The scheme for directly determining normal pass amount of the every road in minimum statistics unit based on experience value of middle introduction is different, at this In example, processor 71 can for driving trace after the normal distribution in statistics period, being based on should in determining clustering cluster The zone of reasonableness of path number is determined in normal distribution.The scheme of two kinds of determining path number zone of reasonableness is provided here:

Scheme one: processor 71 determines the mean value and standard deviation of the path number normal distribution of driving path, then will [ Be worth -2 times of standard deviations ,+2 times of standard deviations of mean value] range as zone of reasonableness.

Scheme two: processor 71 will determine the quartile of the normal distribution, then by [first quartile -1.5* IQR, third quartile+1.5*IQR] range as zone of reasonableness, wherein IQR is the very poor of quartile, and value is equal to Third quartile subtracts the value of first quartile.

For standardized normal distribution, the zone of reasonableness that above two scheme determines can include 75% minimum statistics unit Path number, remaining 25% be used as outlier.It should be understood that working as number of path corresponding to a certain minimum statistics unit Mesh is outlier, it may be possible to because of any one in two such reason: first, path corresponding to the minimum statistics unit Number is few, deviates from normal range (NR), and another kind is because the path number of the minimum statistics unit is too many, most so as to cause this Small statistic unit deviates from the zone of reasonableness that the path number of 75% minimum statistics unit is constituted.So determining clustering cluster When validity, processor 71 should not directly think road minimum statistics list corresponding to all outliers of the clustering cluster It being all off and uses in member, for the first case, it may be said that the bright road in the minimum statistics unit do not use substantially, It has been in discarded state, but another situation, can only illustrate that the road of the clustering cluster may in the minimum statistics unit It is due to external causes such as traffic controls, so traffic volume increases severely.

For example, it is [800,1500] that processor 71, which is the zone of reasonableness determined of a certain clustering cluster path number, when statistics Phase is still July, it is assumed that in July, has 27 days path numbers all to fall in the zone of reasonableness.But July 15 days~7 In this three days on the 17th moon, the path number of the clustering cluster all far surpasses 1500, reaches 3000 or so.According to description above, Processor 71 cannot directly assert that the road of clustering cluster there is a problem during 15~July 17 July because, this three In it, it may be possible to since failure occur in other roads, or because of traffic control, cause associated with corresponding road Road travel amount is substituted to increase severely.Although this is not belonging to the condition of production, it will be apparent that the use state of road is normal.

After the zone of reasonableness for determining path number, processor 71 is also it needs to be determined that the date is the latest in zone of reasonableness out One minimum statistics unit, and using the date as the nearest reasonable date.For clustering cluster X, the nearest rationally date contains Justice is, within statistics period, clustering cluster X driving path number reasonable last day, namely within statistics period, the cluster Traffic volume normal last day on road corresponding to cluster C.It is assumed that being directed to clustering cluster X, in July, 2017, day in zone of reasonableness Phase the latest one day is July 26, that is to say, that in remaining 27 days~31, daily driving path number does not exist In normal range (NR).The path number of driving path in the past few days may be the minimum value less than zone of reasonableness, it is also possible to be more than The maximum value of zone of reasonableness.

So why not in the reasonable scope for determining each minimum statistics unit not in the reasonable scope, this reality It applies in example, processor 71 can also obtain the minimum statistics unit of date the latest in clustering cluster, for the statistics in July, gather The date, a minimum statistics unit the latest was exactly July 31 in class cluster.Processor 71 can be by the minimum system of the date the latest The path number of unit is counted as nearest path number, such as July, nearest path number is exactly on July 31 this day Path number.

It gets recently rationally after date and nearest path number, processor 71 can will reasonable same expiration date as scheduled recently Phase is compared, while nearest path number being compared with the minimum value of zone of reasonableness.If nearest path number is less than The minimum value of zone of reasonableness, then illustrate why the minimum statistics unit peels off, and is because its path number is too small.For most The nearly rationally date with expiration date threshold value comparison, if rationally the date earlier than expiration date threshold value, illustrates, expired recently Before date threshold reaches, the path number of the clustering cluster is with regard to abnormal, not in the reasonable scope；Conversely, then illustrating Expiration date threshold value is crossed, road corresponding to the clustering cluster still possesses normal traffic volume.

The setting of expiration date threshold value can be set as needed by map vendors, for example, if by expiration date threshold Value is set as the last day of statistical time range, then decision condition is more stringent, as long as the last day road in statistical time range makes Abnormal state is identified as with the abnormal road of state；If by expiration date threshold value be set as statistical time range last That day before it one week, then only those road occupation state is not daily within the last week of statistical time range last day Normal road is just identified as abnormal state.

If processor 71 judges that recently rationally the date is earlier than expiration date threshold value, and nearest path number is less than rationally The minimum value of range then can be determined that the clustering cluster is expired.For a map grid, wherein may include multiple clusters Cluster, for all clustering clusters in map grid, processor 71 can carry out above-mentioned processing respectively, so that it is determined that corresponding cluster The validity of cluster.After determining the expired cluster in a map grid, so that it may be determined as remaining clustering cluster effectively Cluster.

Finally, processor 71 can be updated according to the road in the Usefulness Pair map grid of clustering cluster: will be expired The corresponding road of cluster is arranged to expired road, characterizes the road and is no longer used, and in discarded state, (certainly, this is discarded possible It is temporary, but cut-off statistics period terminates, which wouldn't provide use).For expired road, processor 71 can be by it It is deleted from the map grid.But, in order to embody the difference after map rejuvenation, processor 71 can continue in map grid Middle reservation road, only adds overdue indicator for it, and for example, expired road is configured at the different display of effective road Color.User in this way can recognize that corresponding position is implicitly present in a road when seeing map, but at present It cannot pass through.

In the present embodiment, there is likely to be some clustering clusters in map grid, only track does not have road.This poly- Belong to those of effective cluster clustering cluster in class cluster, be more special one kind, is i.e. " newly-increased cluster ".There is no road in newly-increased cluster, but There is track, this explanation does not represent corresponding road on the electronic map of master, but there is current track here, And road state be it is normal, therefore, corresponding position may be newly to have opened road.So processor 71 can will increase newly Road is arranged to " new added road ", and is identified on the electronic map of new edition.It is newly-increased to allow user to understand the road Road, processor 71 display color of new added road can be arranged and expired road and script existing for road not Together.

Map data update scheme provided in this embodiment, due to individually being handled for each map grid, And for a electronic map, data volume to be treated is very big when cluster, availability deciding, road update , therefore, the map datum processing of different map grids can be assigned to different calculating using HDFS scheme by processor 71 It is executed on unit (such as processor 71 or computer), to promote the data-handling efficiency of electronic map entirety.

Server provided in an embodiment of the present invention can be direct when carrying out data processing to a certain map grid It is carried out based on the driving path in the map grid, avoids and special messenger is arranged to acquire brought high cost to data are carried out on the spot The problem of.Importantly, being directed to every road, track thereon can most characterize its use state, and track be the moment more New, therefore, server is updated according to the data that driving path carries out map grid, enables to update result more comprehensively, together When real-time it is also more preferable, avoid electronic map update lag give user's bring various troubles, the user experience is improved.

Embodiment three:

The present embodiment provides a kind of map data update systems, refer to Fig. 8:

In map data update system 8, including transportation database server 81, track database server 82, more New detection service device 83 and update editing server 84.Wherein, transportation database server 81 and track database server 82 assume responsibility for storing respectively the function of real road data and track data of going on a journey, and update detection service device 83 and be then responsible for basis The data that transportation database server 81 and track database server 82 store have determined that road is expired and which have been increased newly A little roads.Editing server 84 is updated then according to the testing result of update detection service device 83 in transportation database server 81 The road data of storage is updated, to obtain new electronic map data, is then pushed to by modes such as map upgradings User.Both of the aforesaid may be implemented it is understood that updating detection service device 83 and updating the mutual cooperation of editing server 84 The function of server in embodiment.

It should be understood that in the present embodiment, the function of the server of four seed types can be provided simultaneously at one It states and is realized on the server of various functions or the function of two or three servers is desirably integrated on a server.Separately Outside, the data volume if necessary to store or handle is excessive, can also be by the share tasks executed by a server in Fig. 8 It is executed on to one or more other server.For example, in some examples of the present embodiment, in map data update system In, the more update detection service devices and Duo Tai for detecting road update can be disposed and update editing server, these updates Detection service device is respectively used to be updated detection to one or more map grids on electronic map to be upgraded.For more The update detection process of new detection service device 83, and the update editing process of editing server 84 is updated, in previous embodiment In done very detailed introduction, which is not described herein again.

Map data update system provided in this embodiment, is handled by map grid, by the number of a electronic map It is updated according to the data for resolving into multiple map grids are updated, to allow multiple update detection service devices with processing capacity and more New edited server carries out data update to different map grids simultaneously, improves data updating efficiency, alleviates single clothes The processing pressure of business device or calculator.

Obviously, those skilled in the art should be understood that each module of the embodiments of the present invention or each step can be used General computing device realizes that they can be concentrated on a single computing device, or be distributed in multiple computing device institutes On the network of composition, optionally, they can be realized with the program code that computing device can perform, it is thus possible to by them It is stored in computer storage medium (ROM/RAM, magnetic disk, CD) and is performed by computing device, and in some cases, it can With the steps shown or described are performed in an order that is different from the one herein, or they are fabricated to each integrated circuit dies Block, or single integrated circuit module is maked multiple modules or steps in them to realize.So the present invention does not limit It is combined in any specific hardware and software.

The above content is combining specific embodiment to be further described to made by the embodiment of the present invention, cannot recognize Fixed specific implementation of the invention is only limited to these instructions.For those of ordinary skill in the art to which the present invention belongs, Without departing from the inventive concept of the premise, a number of simple deductions or replacements can also be made, all shall be regarded as belonging to the present invention Protection scope.

Claims

1. a kind of map data updating method, comprising:

Clustering processing is carried out according to the distance between path for the driving path in each map grid to obtain comprising at least one row Sail the clustering cluster in path；

The road in the map grid is updated according to statistical result.

2. map data updating method as described in claim 1, which is characterized in that the driving path mapping that will acquire Include: into preset map grid

It determines the apex coordinate of the map grid, and determines the range of the map grid according to the apex coordinate；

The driving path for belonging to the map grid is determined according to the location information of each driving path.

3. map data updating method as described in claim 1, which is characterized in that the traveling in each map grid Path obtains the clustering cluster comprising at least one driving path according to the distance progress clustering processing between path

Each driving path is clustered according to the distance of driving path each in the map grid and global clustering parameter Processing；

Judge whether the number of road in clustering cluster obtained by clustered processing is less than or equal to 1；

If it is not, continue to carry out further clustering processing to the driving trace in the clustering cluster after Local Clustering parameter is then arranged, Until the number of road in the clustering cluster is less than or equal to 1.

4. map data updating method as claimed in claim 3, which is characterized in that described according to row each in the map grid The distance and global clustering parameter for sailing path carry out clustering processing to each driving path and include:

The distance two-by-two for calculating driving trace in the map grid, by the distance with the cluster in the global clustering parameter Threshold value is compared, and the driving trace that distance is less than or equal to the cluster threshold value is gathered for a clustering cluster；

Or,

Using k- central point clustering schemes to driving path N number of in the map grid progress independent clustering processing of n times, and from Optimal solution is selected in n independent cluster results；The k- central point clustering schemes include:

K driving path is randomly choosed from N number of driving path of the map grid as initial cluster center；

Remaining N-K driving path is calculated respectively with the distance at each initial cluster center；

The N-K driving path is respectively divided in clustering cluster corresponding to K initial cluster center according to the distance；

The iterative processing of m subcenter is carried out to K clustering cluster and obtains the corresponding cluster result of k- central point clustering schemes；In described Heart iterative processing includes:

Select in the clustering cluster the smallest driving path of distance between other driving paths poly- as this respectively for each clustering cluster The new cluster center of class cluster；

It calculates and removes as the excentral N-K driving path of new cluster respectively with the new distance at each new cluster center；

The N-K driving path is respectively divided in clustering cluster corresponding to K new cluster centers according to the new distance.

5. map data updating method as claimed in claim 4, which is characterized in that N number of row from the map grid It sails before randomly choosing K driving path in path as initial cluster center, further includes:

According to preset ratio promote road in driving path in the map grid shared by weight.

6. map data updating method as described in any one in claim 1-5, which is characterized in that described based in each clustering cluster The path number of driving path count each clustering cluster validity include: according to the path number of driving path in each clustering cluster and Date counts the validity of each clustering cluster.

7. map data updating method as claimed in claim 6, which is characterized in that according to the road of driving path in each clustering cluster The validity that diameter number and date count each clustering cluster includes:

The path number of each clustering cluster each minimum statistics unit within statistics period is obtained to determine for each cluster The normal distribution interim in the statistics of cluster path number；

The zone of reasonableness of path number is determined according to the normal distribution of each clustering cluster；

Determine the minimum statistics unit of date the latest in the zone of reasonableness, and using its date as the clustering cluster Nearest reasonable date, and the path number of a minimum statistics unit using the date in the clustering cluster the latest are gathered as described The nearest path number of class cluster；

The nearest reasonable date of each clustering cluster is compared with preset expiration date, and by the nearest number of path Mesh is compared with the minimum value of the zone of reasonableness；

Determine that expired cluster and effective cluster in the map grid, the expired cluster are that rationally the date is early recently according to comparison result In the expiration date, and recently, path number is less than the clustering cluster of the minimum value of the zone of reasonableness, and effective cluster is institute State the clustering cluster in map grid in addition to the expired cluster.

8. map data updating method as claimed in claim 7, which is characterized in that it is described according to statistical result to the map It includes: to set expired road for the road in the expired cluster that road in grid, which is updated,.

9. map data updating method as claimed in claim 8, which is characterized in that it is described according to statistical result to the map Road in grid is updated further include:

The effective cluster comprising road number less than 1 is filtered out as newly-increased cluster, and by the newly-increased cluster with other rows It sails the smallest driving path of path distance and is set as new added road.

10. map data updating method as described in any one in claim 1-5, which is characterized in that described to be based on each clustering cluster The validity that the path number of middle driving path counts each clustering cluster includes:

The path number of each clustering cluster is compared with normal range (NR), when the path number for determining the clustering cluster is not belonging to just When normal range, determine that the clustering cluster is expired cluster, the normal range (NR) characterizes the clustering cluster and possessed under normal circumstances The numberical range of driving path.

11. a kind of server, which is characterized in that the server includes processor, memory and communication bus；

The processor is for executing one or more program stored in memory, to realize as in claims 1 to 10 The step of described in any item map data updating methods.

12. a kind of computer readable storage medium, which is characterized in that the computer-readable recording medium storage have one or Multiple programs, one or more of programs can be executed by one or more processor, to realize such as claims 1 to 10 Any one of described in map data updating method the step of.