CN110162586A - A kind of similarity search system and method suitable for mobile intended branch track - Google Patents
A kind of similarity search system and method suitable for mobile intended branch track Download PDFInfo
- Publication number
- CN110162586A CN110162586A CN201910438280.7A CN201910438280A CN110162586A CN 110162586 A CN110162586 A CN 110162586A CN 201910438280 A CN201910438280 A CN 201910438280A CN 110162586 A CN110162586 A CN 110162586A
- Authority
- CN
- China
- Prior art keywords
- track
- similarity
- branch
- branch track
- distance
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/29—Geographical information databases
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Databases & Information Systems (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Remote Sensing (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The present invention relates to a kind of similarity search system and methods suitable for mobile intended branch track, it is difficult to calculate branch track in topological structure for the prior art, the shortcomings that motion track and evolution duration similarity, it proposes track data converting two kinds of data structures of formation sequence figure and path set, pass through the sequence chart DSG editing distance of definition, path set PED distance and path set DT distance, in evolved structure between metric objective track and reference locus, similarity in movement routine and evolution duration, and comprehensive three kinds of distance metrics carry out overall similarity calculating to target trajectory and reference locus, it can be quick based on the similarity, accurately inquire and be matched to the branch track most like with given mobile intended branch track, it excavates and visualizes for space-time trajectory, similar cases reasoning provides support.
Description
Technical field
The present invention relates to a kind of similarity search system and methods suitable for mobile intended branch track, belong to geographical letter
Breath processing and trajectory data mining field (G06F 16/29；G06F 19/24).
Technical background
Spatial data handling and analysis are the core function of GIS-Geographic Information System, the meter including point, line, surface isovector data
It calculates.Track data is a kind of common line vector data, has recorded the spatial position change of mobile target, for example, vehicle, animal,
The track that the mobile target such as typhoon generates.How from a large amount of track data, extracted by data mining means valuable
Space-time trajectory feature and mode have become an important content of geographic information system technology development.The similitude meter of track data
It can be regarded as the key to support the key datas digging technologies such as track classification, cluster, association analysis, there are many correlations to grind in recent years
Study carefully and patent of invention.Trajectory distance measure of Berndt etc. (1994) proposition based on dynamic time warping, but this meter
Calculation method is to data noise-sensitive.In order to inhibit the influence of data noise, Vlachos etc. (2002) uses the public sub- sequence of longest
Column method measures the similitude between track.Chen etc. (2005) proposes a kind of new track similarity measurements based on editing distance
Amount method.The Computer Department of the Chinese Academy of Science disclosed a patent (CN102722541B) in 2002, it is contemplated that track dispersion degree
Difference proposes track similarity calculation method and system based on geometrical characteristic.IBM Corporation disclosed in 2015 one about
The method for measuring similarity (US9747805B2) of mobile target trajectory, by calculate space length between matching line segment and when
Between distance measure two tracks similarity.Hohai University disclosed a patent (CN106960006A) in 2017 and proposes to use
Similarity measurement system and method between a kind of different tracks of most like line segment between longest common subsequence method construct track.
These researchs and primary limitation are the track data for only supporting linear sequence table to show at present, i.e. track is only unique
Start-stop point and unique path.But mobile target of some complexity such as rain group, haze, gloomy fire, vortex etc. is in motion process
It is middle to have the special behaviors such as division, merging, cause its track to have the characteristics that a lot of points, more terminals, multipath.
In the present invention, because the track with multiple sub-trajectories for dividing and merging formation is referred to as branch track.Due to
The matching line segment for having one-to-one correspondence that can calculate distance between existing track similarity measurement technical requirements track, and branch track
Between can have one-to-many or even multi-to-multi a line segment corresponding relationship, time and difference spatially are not only difficult to carry out matching meter
It calculates, the topological structure difference between branch track is not embodied in calculated result even more.So existing track similarity measurements
Amount technology cannot be used directly for measurement branch track similarity measurement, limit trajectory data mining technology application range and
Practical value.
Summary of the invention
Present invention solves the technical problem that: it is directed to limitation of the existing technology, is provided a kind of suitable for mobile target
The similarity search system and method for branch track breaks through skill of the existing track method for measuring similarity on branch's trajectory calculation
Art bottleneck realizes the support to the measuring similarity of the complicated movements target special branch tracks such as gloomy fire, rain group, vortex.
Technical solution of the present invention:
A kind of similarity search system suitable for mobile intended branch track of the present invention, as shown in Figure 1, comprising: data
It is loaded into and format checking module, track data conversion module, distance calculation module and similarity search result output module；It is described
Mobile target refers to that rain group, haze, gloomy fire and vortex can generate the dynamic phenomenon of branch track when moving；The branch track
Refer to complicated track with multiple sub-trajectories of the mobile target because dividing and merging formation；Wherein:
Data are loaded into and format checking module: input data set is branch's track data collection, is read from computer local file
The data set is taken, and whether the format for detecting the data file of input is legal, then prompt mistake if it is illegal and waited stand-by
Family is re-entered, if legal then according to branch's rail most like in branch's track A match query data set of user's selection
Mark, when inquiry will traversal input data is concentrated one by one other branch tracks, if the branch track for being traversed access this moment is B,
Branch track A and B will be output to track data conversion module；
Track data conversion module: the data of branch the track A and B of input are converted respectively and are generated as sequence chart A and sequence
Column figure tetra- kinds of data structures of B and path set A and path set B, and it is output to distance calculation module；
Distance calculation module: the sequence linearization editing distance of sequence of calculation figure A and sequence chart B, abbreviation DSG distance,
With metrictopology structural similarity S1, the set editing distance of path set A and path set B, abbreviation PED distance, with measurement are calculated
Movement routine similarity S2 calculates the time editing distance of path set A and path set B, abbreviation DT distance, to measure evolution duration
Similarity S3, finally according to tri- kinds of distance results of S1, S2 and S3, the overall similarity for calculating target trajectory and reference locus is measured
Value S is passed to similarity measurement result output module；
Similarity search result output module: the overall phase that branch track A is calculated one by one with other branch tracks
Like property metric S by sorting from small to large, choose the smallest branch track be output to inquiry system be labeled as with branch track A most
Similar branch track.
A kind of similarity search method suitable for mobile intended branch track of the present invention, as shown in Fig. 2, including following step
It is rapid:
Step 1, according to the branch track A and branch track B in contrast of user's selection, it is respective to generate track A and B
Sequence chart Gs and path set Ps data, for calculating similarity of the two in topological structure, motion track and evolution duration；
Step 2, for the sequence chart Gs data of branch track A and B, the DSG suitable for branch track for passing through definition is compiled
Distance is collected, the similarity value S1 of branch track A and B on the topology is calculated, for inquiring in track topology and dividing
Branch track A most like branch track；
Step 3, for the path set Ps data of branch track A and B, the PED suitable for branch track for passing through definition is compiled
Distance is collected, similarity value S1 of the branch track A and B in movement routine is calculated, for inquiring in the movement routine of track and dividing
Branch track A most like branch track；
Step 4, for the path set Ps data of branch track A and B, edited by the DT suitable for branch track of definition
Distance calculates similarity measure values S3 of the branch track A and B in evolution duration, for inquire in track evolution duration with
Branch track A most like branch track；
Step 5 is normalized S1, S2 and S3 respectively using extreme value standardized method, and calculates the 2- of three
Overall similarity of the norm value as target trajectory and reference locus in evolved structure, movement routine and evolution duration, is used for
Inquire in topological structure, motion track and evolution duration the generally all most like branch track with branch track A.
The constructing plan of sequence chart is as follows in the step 1,
The sequence chart Gs of track is made of sequence node Vs and connection side Es, GS=< VS,ES>, sequence node is for indicating
Mobile target trajectory is continuous and is free of the sub-trajectory of branch；Division and merging in branch track can generate multiple sequence nodes,
Connection relationship between sequence node is indicated with connection side Es, and the topological structure of track is indicated with sequence chart, is to calculate branch's rail
The precondition of topological structure similarity between mark.
The constructing plan of path set is as follows in the step 1:
The path set Ps of track is made of several path p without branched structure, and each paths are connected to the one of former track
A starting point and a terminal, and be made of the location point of approach in track, p=[v1,v2,…,vn], each location point note
Space coordinate where recording mobile target at that time, the time attribute of path p are expressed as the two of initial time and duration composition
Dimensional vector；The path set for extracting track is calculated between branch track before movement routine similarity and evolution duration similarity
Propose condition.
Topological structure Similarity measures in the step 2 suitable for branch track are as follows:
Firstly, the sequence chart Gs to track carries out uniform enconding: the node that in-degree is 0 since sequence chart, passing through depth
The degree traversal mode of priority, by depth order collating sequence node, the sequence node of same depth merges into set, each node table
It is shown as the bivector of in-degree and out-degree, the sequence chart uniform enconding structure of track is constructed with this；
Then, according to DSG (L defined belowR,LS) editing distance calculate target trajectory and reference locus sequence figure line
Property coding LRAnd LSSimilarity: DSG distance definition is as follows:
Wherein, RiAnd SiRespectively indicate LRAnd LSIn the i-th ordinal position sequence node set, p and q are L respectivelyRAnd LS's
Length, rest (LR) and rest (LS) respectively indicate LRAnd LSRemove the remaining uniform enconding structure of first position sequence node set,
Matched minimum cost between MinCost function calculate node collection, gap indicate the notch node without corresponding position in linear structure
Collection, notch node collection go out the sequence node that in-degree is 0 by one and constitute；The distance between sequence node measurement uses Manhattan
Distance；
In order to make to calculate the similarity measurement for being suitable for branch track, using DSG between sequence of calculation node set
Apart from when, first pass through Kuhn-Munkres combined optimization method the arrangement set of target trajectory and reference locus is carried out it is optimal
Then matching calculates lowest combined cost and represents the similarity S1 of the two on the topology.
Movement routine Similarity measures scheme in the step 3 suitable for branch track is as follows,
Firstly, carrying out center-of-mass coordinate conversion to all paths: path point coordinate (x, y) is converted to the arrow to path mass center
Measure coordinate
Then, the movement routine phase of target trajectory and reference locus is calculated according to PED (P, Q) editing distance defined below
Like degree, path P=[p1,p2,…,pn] and Q=[q1,q2,…,qm] editing distance PED determination it is as follows:
Wherein, piAnd qiI-th point is respectively indicated in P and Q,WithIt is path P and the mass center of Q, rest (P) and rest
(Q) it respectively indicates P and Q and removes the path after first node；
There are mulitpaths for the path set of branch track, in order to make to calculate the similarity measurement for being suitable for branch track,
When calculating the distance between path set, Kuhn-Munkres combined optimization method is first passed through to the road of target trajectory and reference locus
Diameter collection carries out Optimum Matching, then calculates lowest combined cost by PED distance and represents similarity of the two in movement routine
S2。
Evolution duration Similarity measures scheme in the step 4 suitable for branch track is as follows:
Firstly, according to the sampling interval ε and period tau of track data, to the path set data of target trajectory and reference locus
It is converted as follows:
Wherein, ts and △ respectively indicates the initial time in path and the duration is converted, and mod indicates that complementation calculates；
Then, according to the Optimum Matching path for the target trajectory and reference locus being calculated in step 3, by defined below
DT (Pi,Qj) evolution duration similarity of the time editing apart from calculating target trajectory and reference locus, coupling path Pi=
(tsi′,△i') and Qj=(tsj′,△j') time gap DT determination it is as follows:
DT(Pi,Qj)=min | tsi′-tsj′|,τ-|tsi′-tsj′|}+|△i′-△j′|
In order to make to calculate the similarity measurement for being suitable for branch track, using the sum of the time gap DT in all pairing paths
As the similarity measure values S3 of target trajectory and reference locus in evolution duration.
Overall similarity numerical procedure in the step 5 suitable for branch track is as follows:
Firstly, being normalized using extreme value standardized method to S1, S2 and S3, there are three kinds of metrics comparable
Property；
Then, the 2- norm for calculating S1, S2 and S3 indicates the overall similarity of target trajectory and reference locus.
The advantages of the present invention over the prior art are that:
(1) for the existing track similarity measurement technology such as DTW, LCSS, ED can not topological structure to branch track it is poor
Different the shortcomings that being measured, the present invention is abstracted the topological structure of branch track using sequence chart, and passes through uniform enconding and editor
Distance DSG, realize to branch track topological structure difference distance metric.
(2) one-to-many and multi-to-multi line match empty similitude in time can not be supported for technologies such as DTW, LCSS, ED
Limitation is calculated, the present invention carries out optimal to the set of paths in branch track by Kuhn-Munkres combined optimization method
Match, and then define coupling path and be integrated into the distance on room and time, realizes and space-time similitude meter is carried out to branch track
The support of calculation.
(3) for the technologies such as DTW, LCSS, ED can not synthetically measure branch track topological structure, spatial position and
Deficiency in time attribute, when the present invention respectively defines topological structure distance S1, movement routine distance S2 and develops over long distances
S3, and measured by a kind of comprehensive distance that branch track is established in extreme value standardization and 2- norm to improve above-mentioned deficiency.
(4) it defines the distance between branch track respectively in terms of evolved structure, movement routine and evolution duration three, fills
The Optimum Matching and distance metric problem between branch track of determining in topological structure, spatial position and time attribute are decomposed, is dashed forward
Technical bottleneck of the existing track method for measuring similarity on branch's trajectory calculation is broken, is realized to complexity such as gloomy fire, rain group, vortexs
The similarity calculation of mobile target special branch track is supported.
Detailed description of the invention
Fig. 1 is the composition block diagram of present system；
Fig. 2 is main flow chart of the invention；
Fig. 3 is the diagram that the present invention implements 1 Zhong Liangtiao branch track of use-case；
Fig. 4 is the uniform enconding result diagram for implementing sequence chart in use-case 1；
Fig. 5 is the 8 track diagrams implemented in use-case 2；
Fig. 6 is the similarity result expression diagram for implementing 8 tracks and the track P1 in use-case 2；
Fig. 7 is the cold whirlpool in Dongsha for implementing to be matched in use-case 3 by similarity search.
Specific embodiment
The following describes the present invention in detail with reference to the accompanying drawings and embodiments.
Similarity calculation between the embodiment branch track 1: Liang Tiao
The present embodiment is input with Liang Tiao branch shown in Fig. 3 track Tra1 and Tra2, branch track through the invention
Similarity search system can be passed through existing with similarity degree of the measurement in topological structure, spatial position and time attribute
There are other methods then and cannot achieve the similitude matching of the two.The specific implementation step of the embodiment of the present invention is as follows:
The first step, input Tra1 and Tra2 data set to branch's track similarity query system of the invention select Tra1
For target trajectory, the similarity of inquiry and calculating and the matched branch track Tra1.
Branch's track data of Tra1 and Tra2 is input to track data conversion module by second step, inquiry system, is generated
Liang Tiao branch track corresponding sequence chart Gs1, Gs2 and path set Ps1, Ps2；
Sequence chart Gs1, Gs2 is input to sequence chart DSG apart from computational submodule, to sequence chart by third step, inquiry system
Gs1 and Gs2 carry out uniform enconding as shown in figure 4, after coding, and topology diagram representated by Gs1 and Gs2 is converted into two
Group linear order, on the basis of this linear order, by DSG apart from available Gs1 and Gs2 on the topology similar
Property metric S1；
Path set Ps1, Ps2 is input to path set PED apart from computational submodule, by path set by the 4th step, inquiry system
Path in Ps1 and Ps2 carries out coordinate conversion by centroid position, is then calculated between any two paths using PED distance
Distance, and minimum cost distance is calculated as path set Ps1 and Ps2 by Kuhn-Munkres combinatorial optimization algorithm and is being moved
Similarity measure values S2 on dynamic path；
Path set Ps1, Ps2 is input to path set DT apart from computational submodule, calculates step 4 by the 5th step, inquiry system
The DT distance of middle coupling path, and the similarity measure values using DT sum of the distance as path set Ps1 and Ps2 in evolution duration
S3；
Distance metric value S1, S2 and S3 are input to similarity search result output module, pass through extreme value mark by final step
S1, S2 and S3 are transformed into [0,1] range by quasi-ization method, take multidimensional of the 2- norm value of three as Liang Tiao branch track
Similarity measure values S, and return result to user.
Embodiment 2: inquiry and the most like branch track in intended branch track in a plurality of branch track
In embodiments of the present invention, the data set of input includes the 8 different track data of temporal-spatial evolution structure such as Fig. 5 institutes
Show, the track in figure is indicated by node and line, and node indicates the x of mobile target, y-coordinate and specific Time of Day.Its
In, P1 to P4 is branchiess track, and P5 to P8 is branch track, and the present embodiment is by branch's track similarity through the invention
Inquiry system, inquiry and track P1 most like track, and the similarity measurement for not passing through track and P1 is provided, and by existing
The similitude matching degree of P1 Yu branch track P5 to P8 can not be measured by having other methods then, thus can not also inquire these phases
As branch track.The specific implementation process of the present embodiment is as follows:
Data set including this 8 tracks is input in inquiry system of the invention by the first step, and track P1 is selected to make
For target trajectory, match query and P1 most like track from data；
Track P1 and remaining 7 track data are input to track data conversion module, turned respectively by second step, inquiry system
It is changed to sequence chart and path set data set；
The sequence diagram data of all tracks is input to sequence chart DSG apart from computational submodule by third step, inquiry system, by
One calculates the DSG distance value S1 of P1 and other tracks；
The sequence diagram data of all tracks is input to path set PED apart from computational submodule by the 4th step, inquiry system, by
One calculates the PED distance value S2 of P1 and other tracks；
The sequence diagram data of all tracks is input to path set DT apart from computational submodule by the 5th step, inquiry system, by
One calculates the DT distance value S3 of P1 and other tracks；
S1, S2 and S3 are input to similarity search result output module by the 6th step, inquiry system, calculate P1 and other rails
The overall similarity S of mark, and sort and obtain the track most like with P1, according to the similarity S1, S2 of P1 and remaining 7 track and
S3, available P1 and the similarity measure such as Fig. 6 of remaining track on topological structure, motion track and evolution duration.Such as Fig. 6
Shown, what the point in figure positioned at origin (0,0,0) indicated is the track P1, and what the point of remaining different colours represented is other 7 rails
Mark, each point the coordinate representation in figure be the point represent track and the track P1 between topological structure, motion track and
Distance in three dimensions of evolution duration, by the method for visualizing can be intuitive to see with P1 relatively be P2, P4
And P5, with P1 it is apart from each other be P6 and P8.
Embodiment 3: inquiry and target swirl branch, the most like Dongsha track Leng Wo, track
In embodiments of the present invention, the branch's track similarity search system proposed through the invention, can from for a long time with
In the swirling trajectory data that track obtains, inquire with the consistent swirling trajectory in the cold whirlpool in target Dongsha, and pass through existing other technologies
Real corresponding swirling trajectory therewith cannot then be inquired.The specific implementation process of the embodiment is as follows:
The mesoscale eddy track data that South China Sea Dongsha surrounding waters in 2001 tracks is input to the present invention by the first step
Similarity search system in, in the data Integrated query, (Grey Point linear list shows in Fig. 7 with the target Dongsha track cold whirlpool d
Track) most consistent swirling trajectory；
All swirling trajectories that the track of the cold whirlpool d in Dongsha and data are concentrated are input to track number by second step, inquiry system
According to conversion module, it is respectively converted into sequence chart and path set data set；
The sequence diagram data of all tracks is input to sequence chart DSG apart from computational submodule by third step, inquiry system, by
One calculates the track of the cold whirlpool d in Dongsha and the DSG distance value S1 of other tracks；
The sequence diagram data of all tracks is input to path set PED apart from computational submodule by the 4th step, inquiry system, by
One calculates the track of the cold whirlpool d in Dongsha and the PED distance value S2 of other tracks；
The sequence diagram data of all tracks is input to path set DT apart from computational submodule by the 5th step, inquiry system, by
One calculates the track of the cold whirlpool d in Dongsha and the DT distance value S3 of other tracks；
S1, S2 and S3 are input to similarity search result output module by the 6th step, inquiry system, calculate the cold whirlpool d in Dongsha
Track and other tracks overall similarity S, and sort obtain the track most like with the track of the cold whirlpool d in Dongsha.It is opened up in Fig. 7
Two tracks P12122 and P12233 most like with the track of the cold whirlpool d in Dongsha in result are shown.Wherein, looking into through the invention
It is P12122 that system retrieval, which is ask, to the corresponding track the cold whirlpool d in Dongsha, and P12122 is a branch track.And P12233 although
In all quite similar, but its spatial position not phases with the cold whirlpool d in Dongsha that is vortexed in starting and end time and movement routine
Meet, so, if without proposed by the present invention to branch's track similarity measurement inquiry system, it may be difficult in the track number of magnanimity
According to middle match query to most identical branch track.
The foregoing is only a preferred embodiment of the present invention, is not intended to restrict the invention, for the skill of this neighborhood
For art personnel, the present invention can change and change.All within the spirits and principles of the present invention, it is made it is any modification, etc.
With replacement, improvement etc., should be included within scope of the presently claimed invention.
Claims (8)
1. a kind of similarity search system suitable for mobile intended branch track, it is characterised in that: include: that data are loaded into dative
Formula checks module, track data conversion module, distance calculation module and similarity search result output module；The mobile target
Refer to that rain group, haze, gloomy fire and vortex can generate the dynamic phenomenon of branch track when moving；The branch track refers to movement
Complicated track with multiple sub-trajectories of the target because dividing and merging formation；Wherein:
Data are loaded into and format checking module: input data set is branch's track data collection, and reading from computer local file should
Data set, and whether the format for detecting the data file of input is legal, then prompts mistake if it is illegal and waits user's weight
New input, is then looked into according to branch track most like in branch's track A match query data set of user's selection if legal
By other branch tracks that traversal input data is concentrated one by one when inquiry, if the branch track for being traversed access this moment is B, branch's rail
Mark A and B will be output to track data conversion module；
Track data conversion module: the data of branch the track A and B of input are converted respectively and are generated as sequence chart A and sequence chart
Tetra- kinds of data structures of B and path set A and path set B, and it is output to distance calculation module；
Distance calculation module: the sequence linearization editing distance of sequence of calculation figure A and sequence chart B, abbreviation DSG distance, with degree
Topological structure similarity S1 is measured, the set editing distance of path set A and path set B, abbreviation PED distance, to measure movement are calculated
Similarity of paths S2, calculates the time editing distance of path set A and path set B, and abbreviation DT distance is similar to measure evolution duration
S3 is spent, finally according to tri- kinds of distance results of S1, S2 and S3, calculates the overall similarity metric S of target trajectory and reference locus,
Incoming similarity measurement result output module；
Similarity search result output module: the overall similarity that branch track A and other branch tracks are calculated one by one
Metric S chooses the smallest branch track and is output to inquiry system labeled as most like with branch track A by sorting from small to large
Branch track.
2. a kind of similarity search method suitable for mobile intended branch track, it is characterised in that: the following steps are included: described
Mobile target refers to that rain group, haze, gloomy fire and vortex can generate the dynamic phenomenon of branch track when moving；The branch track
Refer to complicated track with multiple sub-trajectories of the mobile target because dividing and merging formation；
Step 1, according to the branch track A and branch track B in contrast of user's selection, generate the respective sequence of track A and B
Gs and path set Ps data are schemed, for calculating similarity of the two in topological structure, motion track and evolution duration；
Step 2, for the sequence chart Gs data of branch track A and B, by the DSG suitable for branch track of definition edit away from
From branch track A and B similarity value S1 on the topology being calculated, for inquiring in track topology and branch's rail
Mark A most like branch track；
Step 3, for the path set Ps data of branch track A and B, by the PED suitable for branch track of definition edit away from
From similarity value S1 of the branch track A and B in movement routine being calculated, for inquiring in the movement routine of track and branch's rail
Mark A most like branch track；
Step 4, for the path set Ps data of branch track A and B, by the DT suitable for branch track of definition edit away from
From similarity measure values S3 of the calculating branch track A and B in evolution duration, for inquiring in track evolution duration and dividing
Branch track A most like branch track；
Step 5 is normalized S1, S2 and S3 respectively using extreme value standardized method, and calculates the 2- norm of three
It is worth the overall similarity as target trajectory and reference locus in evolved structure, movement routine and evolution duration, for inquiring
The generally all most like branch track with branch track A in topological structure, motion track and evolution duration.
3. a kind of similarity search method suitable for mobile intended branch track according to claim 2, feature exist
In: the constructing plan of sequence chart is as follows in the step 1:
The sequence chart Gs of track is made of sequence node Vs and connection side Es, GS=< VS,ES>, sequence node is for indicating mobile
Target trajectory is continuous and is free of the sub-trajectory of branch；Division and merging in branch track can generate multiple sequence nodes, sequence
Connection relationship between node is indicated with connection side Es, and the topological structure of track is indicated with sequence chart, be calculate branch track it
Between topological structure similarity precondition.
4. a kind of similarity search method suitable for mobile intended branch track according to claim 2, feature exist
In: the constructing plan of path set is as follows in the step 1:
The path set Ps of track is made of several path p without branched structure, and each paths are connected to one of former track
Point and a terminal, and be made of the location point of approach in track, p=[v1,v2,…,vn], each location point record moves
Moving-target at that time where space coordinate, the time attribute of path p be expressed as two dimension that initial time and duration are constituted to
Amount；The path set for extracting track is the premise item for calculating movement routine similarity and evolution duration similarity between branch track
Part.
5. a kind of similarity search method suitable for mobile intended branch track according to claim 2, feature exist
In: the topological structure Similarity measures in the step 2 suitable for branch track are as follows:
Firstly, the sequence chart Gs to track carries out uniform enconding: the node that in-degree is 0 since sequence chart, passing through depth time
The mode of priority is gone through, by depth order collating sequence node, the sequence node of same depth merges into set, and each node is expressed as
The bivector of in-degree and out-degree constructs the sequence chart uniform enconding structure of track with this；
Then, according to DSG (L defined belowR,LS) editing distance calculates target trajectory and the sequence chart of reference locus is linearly compiled
Code LRAnd LSSimilarity: DSG distance definition is as follows
Wherein, RiAnd SiRespectively indicate LRAnd LSIn the i-th ordinal position sequence node set, p and q are L respectivelyRAnd LSLength
Degree, rest (LR) and rest (LS) respectively indicate LRAnd LSRemove the remaining uniform enconding structure of first position sequence node set,
Matched minimum cost between MinCost function calculate node collection, gap indicate the notch node without corresponding position in linear structure
Collection, notch node collection go out the sequence node that in-degree is 0 by one and constitute；The distance between sequence node measurement uses Manhattan
Distance；
In order to make calculate be suitable for branch track similarity measurement, using DSG between sequence of calculation node set away from
From when, first pass through Kuhn-Munkres combined optimization method and optimal carried out to the arrangement set of target trajectory and reference locus
Match, then calculates lowest combined cost and represent the similarity S1 of the two on the topology.
6. a kind of similarity search method suitable for mobile intended branch track according to claim 2, feature exist
In: the movement routine Similarity measures scheme in the step 3 suitable for branch track is as follows,
Firstly, carrying out center-of-mass coordinate conversion to all paths: path point coordinate (x, y) is converted to the vector seat to path mass center
Mark
Then, similar to the movement routine of reference locus according to PED (P, Q) editing distance calculating target trajectory defined below
Degree, path P=[p1,p2,…,pn] and Q=[q1,q2,…,qm] editing distance PED determination it is as follows:
Wherein, piAnd qiI-th point is respectively indicated in P and Q,WithIt is path P and the mass center of Q, rest (P) and rest (Q) divide
Not Biao Shi P and Q remove the path after first node；
There are mulitpaths for the path set of branch track, in order to calculate calculating suitable for the similarity measurement of branch track
When the distance between path set, Kuhn-Munkres combined optimization method is first passed through to the path set of target trajectory and reference locus
Optimum Matching is carried out, lowest combined cost is then calculated by PED distance and represents similarity S2 of the two in movement routine.
7. a kind of similarity search method suitable for mobile intended branch track according to claim 2, feature exist
In: the evolution duration Similarity measures scheme in the step 4 suitable for branch track is as follows:
Firstly, being carried out according to the sampling interval ε and period tau of track data to the path set data of target trajectory and reference locus
Following conversion:
Wherein, ts and △ respectively indicates the initial time in path and the duration is converted, and mod indicates that complementation calculates；
Then, according to the Optimum Matching path for the target trajectory and reference locus being calculated in step 3, by DT defined below
(Pi,Qj) evolution duration similarity of the time editing apart from calculating target trajectory and reference locus, coupling path Pi=(tsi′,
△i') and Qj=(tsj′,△j') time gap DT determination it is as follows:
DT(Pi,Qj)=min | tsi′-tsj′|,τ-|tsi′-tsj′|}+|△i′-△j′|
In order to make to calculate the similarity measurement for being suitable for branch track, using the sum of the time gap DT in all pairing paths conduct
The similarity measure values S3 of target trajectory and reference locus in evolution duration.
8. a kind of similarity search method suitable for mobile intended branch track according to claim 2, feature exist
In: the overall similarity numerical procedure in the step 5 suitable for branch track is as follows:
Firstly, being normalized using extreme value standardized method to S1, S2 and S3, it is comparable three kinds of metrics；
Then, the 2- norm for calculating S1, S2 and S3 indicates the overall similarity of target trajectory and reference locus.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910438280.7A CN110162586B (en) | 2019-05-24 | 2019-05-24 | Similarity query system and method suitable for branch track of moving target |
| AU2020100419A AU2020100419A4 (en) | 2019-05-24 | 2020-03-19 | A system of querying similar branched trajectory of moving objects and a method thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910438280.7A CN110162586B (en) | 2019-05-24 | 2019-05-24 | Similarity query system and method suitable for branch track of moving target |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110162586A true CN110162586A (en) | 2019-08-23 |
| CN110162586B CN110162586B (en) | 2021-10-19 |
Family
ID=67632552
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910438280.7A Active CN110162586B (en) | 2019-05-24 | 2019-05-24 | Similarity query system and method suitable for branch track of moving target |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN110162586B (en) |
| AU (1) | AU2020100419A4 (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110555061A (en) * | 2019-09-06 | 2019-12-10 | 北京百度网讯科技有限公司 | method and device for determining track similarity |
| CN111651437A (en) * | 2020-04-17 | 2020-09-11 | 北京嘀嘀无限科技发展有限公司 | Data cleaning method and device, electronic equipment and storage medium |
| CN111695299A (en) * | 2020-06-04 | 2020-09-22 | 哈尔滨工程大学 | Mesoscale vortex trajectory prediction method |
| CN112037245A (en) * | 2020-07-22 | 2020-12-04 | 杭州海康威视数字技术股份有限公司 | Method and system for determining similarity of tracked target |
| CN113496317A (en) * | 2020-03-18 | 2021-10-12 | 深圳云天励飞技术有限公司 | Multi-track collision analysis method and device, electronic equipment and storage medium |
| CN113515982A (en) * | 2020-05-22 | 2021-10-19 | 阿里巴巴集团控股有限公司 | Track restoration method and equipment, equipment management method and management equipment |
| CN115870974A (en) * | 2022-11-15 | 2023-03-31 | 华南理工大学 | Multi-axis motion system time optimal trajectory planning method based on cosine similarity comparison |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113126132B (en) * | 2021-04-09 | 2022-11-25 | 内蒙古科电数据服务有限公司 | Method and system for calibrating and analyzing track in mobile inspection |
| CN114285754B (en) * | 2021-12-27 | 2023-05-12 | 中国联合网络通信集团有限公司 | Method, device, equipment and storage medium for generating network topology |
| CN115907159B (en) * | 2022-11-22 | 2023-08-29 | 应急管理部国家减灾中心 | Method, device, equipment and medium for determining typhoons in similar paths |
| CN116227907B (en) * | 2023-05-06 | 2023-07-28 | 安徽思高智能科技有限公司 | Multi-dimensional RPA flow similarity calculation method based on alignment |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103593361A (en) * | 2012-08-14 | 2014-02-19 | 中国科学院沈阳自动化研究所 | Movement space-time trajectory analysis method in sense network environment |
| CN106776482A (en) * | 2016-12-01 | 2017-05-31 | 河海大学 | A kind of track similarity calculating method |
| CN106960006A (en) * | 2017-02-24 | 2017-07-18 | 河海大学 | Measuring similarity system and its measure between a kind of different tracks |
| WO2017143144A1 (en) * | 2016-02-19 | 2017-08-24 | Ohio State Innovation Foundation | Systems, methods and devices for geo-localization |
| CN107784311A (en) * | 2016-08-24 | 2018-03-09 | 中国海洋大学 | Global mesoscale eddy space-time hierarchical topology path construction technology |
-
2019
- 2019-05-24 CN CN201910438280.7A patent/CN110162586B/en active Active
-
2020
- 2020-03-19 AU AU2020100419A patent/AU2020100419A4/en not_active Ceased
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103593361A (en) * | 2012-08-14 | 2014-02-19 | 中国科学院沈阳自动化研究所 | Movement space-time trajectory analysis method in sense network environment |
| WO2017143144A1 (en) * | 2016-02-19 | 2017-08-24 | Ohio State Innovation Foundation | Systems, methods and devices for geo-localization |
| CN107784311A (en) * | 2016-08-24 | 2018-03-09 | 中国海洋大学 | Global mesoscale eddy space-time hierarchical topology path construction technology |
| CN106776482A (en) * | 2016-12-01 | 2017-05-31 | 河海大学 | A kind of track similarity calculating method |
| CN106960006A (en) * | 2017-02-24 | 2017-07-18 | 河海大学 | Measuring similarity system and its measure between a kind of different tracks |
Non-Patent Citations (3)
| Title |
|---|
| HUIMENG WANG等: "A New Method for Measuring Topological Structure Similarity between Complex Trajectories", 《IEEE TRANSACTION ON KNOWLEDGE AND DATA ENGINEERING》 * |
| YANG MO等: "Application of trajectory clustering and regionalization to occean eddies in the South China Sea", 《2015 2ND IEEE INTERNATIONAL CONFERENCE ON SPATIAL DATA MINING AND GEOGRAPHICAL KNOWLEDGE SERVICES(ICSDM)》 * |
| 吕晓聪等: "基于GIS地图的轨迹相似度分析***实现", 《电子技术与软件工程》 * |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110555061A (en) * | 2019-09-06 | 2019-12-10 | 北京百度网讯科技有限公司 | method and device for determining track similarity |
| CN110555061B (en) * | 2019-09-06 | 2022-04-05 | 北京百度网讯科技有限公司 | Method and device for determining track similarity |
| CN113496317A (en) * | 2020-03-18 | 2021-10-12 | 深圳云天励飞技术有限公司 | Multi-track collision analysis method and device, electronic equipment and storage medium |
| CN111651437A (en) * | 2020-04-17 | 2020-09-11 | 北京嘀嘀无限科技发展有限公司 | Data cleaning method and device, electronic equipment and storage medium |
| CN113515982A (en) * | 2020-05-22 | 2021-10-19 | 阿里巴巴集团控股有限公司 | Track restoration method and equipment, equipment management method and management equipment |
| CN113515982B (en) * | 2020-05-22 | 2022-06-14 | 阿里巴巴集团控股有限公司 | Track restoration method and equipment, equipment management method and management equipment |
| CN111695299A (en) * | 2020-06-04 | 2020-09-22 | 哈尔滨工程大学 | Mesoscale vortex trajectory prediction method |
| CN111695299B (en) * | 2020-06-04 | 2022-12-13 | 哈尔滨工程大学 | Mesoscale vortex trajectory prediction method |
| CN112037245A (en) * | 2020-07-22 | 2020-12-04 | 杭州海康威视数字技术股份有限公司 | Method and system for determining similarity of tracked target |
| CN112037245B (en) * | 2020-07-22 | 2023-09-01 | 杭州海康威视数字技术股份有限公司 | Method and system for determining similarity of tracked targets |
| CN115870974A (en) * | 2022-11-15 | 2023-03-31 | 华南理工大学 | Multi-axis motion system time optimal trajectory planning method based on cosine similarity comparison |
| CN115870974B (en) * | 2022-11-15 | 2024-05-17 | 华南理工大学 | Multi-axis motion system time optimal track planning method based on cosine similarity comparison |
Also Published As
| Publication number | Publication date |
|---|---|
| AU2020100419A4 (en) | 2020-04-16 |
| CN110162586B (en) | 2021-10-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110162586A (en) | A kind of similarity search system and method suitable for mobile intended branch track | |
| Ying et al. | Semantic trajectory mining for location prediction | |
| Motro et al. | Estimating the quality of databases | |
| Do et al. | The places of our lives: Visiting patterns and automatic labeling from longitudinal smartphone data | |
| Richter et al. | Semantic trajectory compression: Representing urban movement in a nutshell | |
| CN110008300A (en) | Determination method, apparatus, computer equipment and the storage medium of POI alias | |
| Torres-Sospedra et al. | A comprehensive and reproducible comparison of clustering and optimization rules in wi-fi fingerprinting | |
| KR101945749B1 (en) | Method of searching a data base, navigation device and method of generating an index structure | |
| CN103729580A (en) | Method and device for detecting software plagiarism | |
| Li et al. | Automatically and accurately matching objects in geospatial datasets | |
| CN110162997B (en) | Anonymous privacy protection method based on interpolation points | |
| CN116415430B (en) | Customized SysML model conversion method oriented to space science task demonstration | |
| Dalvi et al. | Deduplicating a places database | |
| Taniar et al. | On mining movement pattern from mobile users | |
| CN110196923B (en) | Underwater detection-oriented multi-source heterogeneous data preprocessing method and system | |
| CN103488525A (en) | Determination of user preference relevant to scene | |
| CN117312578B (en) | Construction method and system of non-genetic carrier spectrum | |
| Shafique et al. | Recommending most popular travel path within a region of interest from historical trajectory data | |
| JP2022040063A (en) | Method and apparatus for extracting spatial relationship between geographic location points | |
| Tang et al. | A trajectory partition method based on combined movement features | |
| Wang et al. | Mining personal frequent routes via road corner detection | |
| US8115619B2 (en) | Information processing apparatus, document management system and method, and computer program | |
| US20130138635A1 (en) | Search method using a plurality of space of interest objects | |
| CN110263108A (en) | A kind of keyword Skyline fuzzy query method and system based on road network | |
| Riede et al. | Automated annotation of Orthoptera songs: first results from analysing the DORSA sound repository |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |