CN109029446A - A kind of pedestrian position prediction technique, device and equipment - Google Patents
A kind of pedestrian position prediction technique, device and equipment Download PDFInfo
- Publication number
- CN109029446A CN109029446A CN201810651020.3A CN201810651020A CN109029446A CN 109029446 A CN109029446 A CN 109029446A CN 201810651020 A CN201810651020 A CN 201810651020A CN 109029446 A CN109029446 A CN 109029446A
- Authority
- CN
- China
- Prior art keywords
- pedestrian
- poi
- track
- predicted
- pattern track
- 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
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
- G01C21/20—Instruments for performing navigational calculations
- G01C21/206—Instruments for performing navigational calculations specially adapted for indoor navigation
Landscapes
- Engineering & Computer Science (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Automation & Control Theory (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Traffic Control Systems (AREA)
- Navigation (AREA)
Abstract
The embodiment of the invention provides a kind of pedestrian position prediction technique, device and equipment, wherein this method comprises: obtaining the position locus information for being predicted pedestrian, position locus information includes the location information for being predicted pedestrian's different moments；According to position locus information, determine the motor pattern track for being predicted pedestrian, wherein motor pattern track includes point of interest (the Point of Interest for being predicted pedestrian and passing through in different periods, POI) region, the region POI indicate that the accessed frequency reaches the region of preset condition；From the multiple historical pattern tracks pre-saved, the matching historical pattern track with motor pattern path matching is searched, wherein historical pattern track includes history pedestrian in the region POI that different periods pass through；According to matching historical pattern track, by Prediction of Markov mode, prediction is predicted the position of pedestrian.It so can be improved the accuracy of prediction.
Description
Technical field
The present invention relates to technical field of data processing, more particularly to pedestrian position prediction technique in a kind of indoor environment,
Device and equipment.
Background technique
With the sustainable development of domestic economy level, shopping and the accounting gone on a journey in the scene of people's daily life are gradually
Increase.The large size indoor scene such as airport, station, market remains biggish flow of the people, the pass between pedestrian and retail shop for a long time
System, it is general only there are two types of relationship, the strong relationship of the retail shop interested to it is actively gone to one is pedestrian, another kind is that retail shop is wide
Broadcast weak relationship of the formula to pedestrian's interspersing advertisements.In weak relationship, retail shop lacks precisely push to pedestrian, causes retail shop in dealing environment
In be in a disadvantageous position, actively with pedestrian can not establish strong relationship.If predicted indoor pedestrian's mobile trend, quotient can be helped
Paving finds potential target client in time, realizes the information such as accurate advertisement service or preferential policy.In addition, in public peace
In all directions, produced in recent years repeatedly because crowd density it is excessive caused by generation that is crowded, trampling accident, accident is all ninety-nine times out of a hundred
The serious consequence that will cause more people's casualties can differentiate room area if predicted indoor pedestrian movement's trend in time
Crowd density, find the potentially possible high crowd density region that can assemble formation ahead of time, issue the measures such as early warning in advance, avoid
Trample the generation of the tragedies such as accident.In addition, in areas such as airport, stations, situations many for the navigation Service of indoor pedestrian are
Route as short as possible the time required to needing to provide facilitates passenger to rush for security check or ticketing spot, and compared to outdoor, it is indoor
Environment is small, crowd density changing greatly in a short time, it is easy to crowd occurs and flock together close-packed scene, at this point, such as
Fruit predicts indoor pedestrian's mobile trend, it can be determined that crowd density on the route of path planning can be accomplished dynamic in time
State programme path improves pedestrian and goes out line efficiency.As can be seen that the research for indoor pedestrian's mobile trend, can be applied to reality
In many scenes of border life.
In the prior art, the prediction to pedestrian's mobile trend in indoor environment is realized by Prediction of Markov method.Tool
Body, multiple pedestrians, which are obtained, in the position data of multiple historical junctures, such as t=1 moment is located at position a1, the t=2 moment is located at position
Set a2..., the t=n moment is located at position anEtc.；According to the position data, the probability transfer matrix based on Markov is calculated:Wherein, element in probability transfer matrix, Pij=P { Xn+1=aj|Xn=ai, indicate row
People is located at position a at the n momentiWhen, the n+1 moment is located at position ajProbability, i=1,2 ..., m；J=1,2 ..., m；Known
When the position at pedestrian's current time, according to the probability transfer matrix, the position of pedestrian's subsequent time, e.g., current time position are predicted
In position a1When, selecting maximum element in the probability transfer matrix the first row is P1j, in this way, pedestrian's subsequent time can be determined
Position be aj。
Understandable to be, the position of subsequent time in pedestrian's moving process is not only related with the position at current time, and
And it is related with the position of multiple historical junctures.And in the prior art, by Prediction of Markov method to pedestrian's mobile trend
During being predicted, the position of subsequent time is predicted in the position that is at current time according only to pedestrian, so that right
The accuracy of pedestrian position prediction is lower.
Summary of the invention
The embodiment of the present invention is designed to provide a kind of pedestrian position prediction technique, device and equipment, to improve prediction
Accuracy.Specific technical solution is as follows:
In a first aspect, the embodiment of the invention provides a kind of pedestrian position prediction techniques, comprising:
The position locus information for being predicted pedestrian is obtained, when the position locus information is predicted pedestrian's difference including described in
The location information at quarter；
According to the position locus information, the motor pattern track of pedestrian is predicted described in determination, wherein the movement mould
Formula track is predicted pedestrian in the region point of interest POI that different periods pass through described in including, and the region POI indicates accessed frequency
The secondary region for reaching preset condition；
From the multiple historical pattern tracks pre-saved, the matching history mould with the motor pattern path matching is searched
Formula track, wherein the historical pattern track includes history pedestrian in the region POI that different periods pass through；
According to the matching historical pattern track position of pedestrian is predicted by Prediction of Markov mode described in prediction
It sets.
Optionally, what the position locus information that multiple historical pattern tracks are respectively directed to each history pedestrian determined；
Pre-save the historical pattern track, comprising:
For each history pedestrian, the position locus information of history pedestrian is obtained；
According to the position locus information of history pedestrian, the location of history pedestrian in preset time range is determined
Constant position is first stop；
According to density-based algorithms, the first stop is clustered, and the class cluster that cluster is obtained is true
It is set to the region POI；
Each first stop is mapped into the region POI, obtains the corresponding historical pattern rail of history pedestrian
Mark；
Save the historical pattern track.
Optionally, described according to the position locus information, determine described in be predicted the motor pattern track of pedestrian, packet
It includes:
According to the position locus information for being predicted pedestrian, determination is described in preset time range to be predicted pedestrian institute
The constant position in the position at place is the second dwell point；
Each second dwell point is mapped into the region POI, obtains the motor pattern track.
Optionally, described from the multiple historical pattern tracks pre-saved, it searches and the motor pattern path matching
Matching historical pattern track, comprising:
According to Euclidean distance calculation, the first of the motor pattern track and multiple historical pattern tracks is calculated separately
Similarity；
According to the sequence of first similarity from high to low, preset quantity historical pattern track is chosen, as primary election
Historical pattern track；
According to longest common subsequence LCS calculation, the motor pattern track and the primary election historical pattern are calculated
Second similarity of track；
According to the sequence of second similarity from high to low, preset quantity primary election historical pattern track is chosen,
As the matching historical pattern track.
Optionally, described to be predicted described pre- according to the matching historical pattern track by Prediction of Markov mode
Survey the position of pedestrian, comprising:
Determine the candidate region POI for including in the matching historical pattern track；
Pedestrian is predicted described in calculating and is moved to by second order Prediction of Markov formula according to the candidate region POI
The probability in the different candidate regions POI；
Determine the corresponding candidate region POI when the probability meets condition, to be predicted pedestrian's described in prediction
Position.
Optionally, the region POI that the motor pattern track includes, by including the movement shape for being predicted pedestrian
The POI regional model of state information forms；The region POI that the matching historical pattern track includes, by including history pedestrian
Movement state information POI regional model composition；
The determination probability corresponding candidate region POI when meeting condition, to be predicted to go described in prediction
The position of people, comprising:
Multiple candidate regions POI are corresponded to when the probability highest, calculate separately the POI that the motor pattern track includes
The matching degree for the movement state information that the running state information and multiple candidate regions POI for including in region include；
The corresponding candidate region POI when by the matching degree highest is determined as the position for being predicted pedestrian of prediction.
Optionally, described according to Euclidean distance calculation, calculate separately the motor pattern track and multiple history moulds
First similarity of formula track, comprising:
Determine the POI sequence of all region POI compositions；
Determine corresponding first track vector in the motor pattern track；The value of element is successively in first track vector
Indicate in the motor pattern track whether include corresponding position in the POI sequence the region POI；
Corresponding second track vector in multiple historical pattern tracks, element in second track vector are determined respectively
Value indicate the historical pattern track whether include corresponding position in the POI sequence the region POI；
Calculate separately the Europe of corresponding second track vector of first track vector and multiple historical pattern tracks
Family name's distance；
According to the Euclidean distance, pass through formula:Calculate the movement mould
Formula track the first similarity with multiple historical pattern tracks respectively；
Wherein, sim (αi,αj) it is the first track vector αiWith the second track vector αjThe first similarity,
dis(αi,αj) it is the first track vector αiWith the second track vector αjEuclidean distance, the dim (α) is described
The dimension of POI sequence α.
Optionally, described according to longest common subsequence LCS calculation, calculate the motor pattern track and it is described just
Select the second similarity of historical pattern track, comprising:
Calculate the longest common subsequence length of the motor pattern track Yu primary election historical pattern track；
Determine the minimum length in the motor pattern track and primary election historical pattern track；
According to the longest common subsequence length and the minimum length, pass through formula:Calculate second similarity；
Wherein, the simi2(Trjm,Trjn) it is the motor pattern track Trj that sequence identifier is mmAnd sequence identifier
For the primary election historical pattern track Trj of nnThe second similarity, the L (m, n) be the longest common subsequence length,
Min (the Lm,Ln) it is the minimum length.
Second aspect, the embodiment of the invention provides a kind of pedestrian position prediction meanss, comprising:
Module is obtained, for obtaining the position locus information for being predicted pedestrian, the position locus information includes the quilt
Predict the location information of pedestrian's different moments；
Determining module, for being predicted the motor pattern track of pedestrian described according to the position locus information, determining,
In, the motor pattern track includes described being predicted pedestrian in the region point of interest POI that different periods pass through, the area POI
Domain representation is accessed the region that the frequency reaches preset condition；
Searching module, for searching and the motor pattern track from the multiple historical pattern tracks pre-saved
The matching historical pattern track matched, wherein the historical pattern track includes the POI that history pedestrian passes through in different periods
Region；
Prediction module, for predicting the quilt by Prediction of Markov mode according to the matching historical pattern track
Predict the position of pedestrian.
The third aspect, the embodiment of the invention provides a kind of pre- measurement equipment of pedestrian position, including processor, communication interface,
Memory and communication bus, wherein the processor, the communication interface, the memory are completed by the communication bus
Mutual communication；
The memory, for storing computer program；
The processor when for executing the program stored on the memory, realizes method described in first aspect
Step.
Pedestrian position prediction technique, device and equipment provided in an embodiment of the present invention, can be according to the position of multiple pedestrians
Trace information determines the region point of interest (Point of Interest, POI) frequently accessed by pedestrian, and according to different pedestrians
Position locus information, determine include pedestrian the region point of interest POI that different periods pass through motor pattern track；And it selects
The matching historical pattern track of motor pattern path matching corresponding with pedestrian is predicted, and according to matching historical pattern track pair
It predicts the position for being predicted pedestrian.In this way, consider pedestrian movement during purpose and pedestrian between motion profile
Similitude, the position for being predicted pedestrian is predicted, can be improved the accuracy of prediction.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of invention for those of ordinary skill in the art without creative efforts, can be with
It obtains other drawings based on these drawings.
Fig. 1 is the flow chart of pedestrian position prediction technique provided in an embodiment of the present invention；
Fig. 2 is the flow chart that the region POI is determined in the embodiment of the present invention；
Fig. 3 is the schematic diagram that historical pattern track is determined in the embodiment of the present invention；
Fig. 4 is the flow chart that motor pattern track is determined in the embodiment of the present invention；
Fig. 5 is the structural schematic diagram of pedestrian position prediction meanss provided in an embodiment of the present invention；
Fig. 6 is the structural schematic diagram of the pre- measurement equipment of pedestrian position provided in an embodiment of the present invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
Currently, realized to the prediction of pedestrian position by single order Prediction of Markov mode in indoor environment, i.e. only root
The position of subsequent time is predicted according to the position at pedestrian's current time, and according to moving process common sense it is found that pedestrian's moving process
The position of middle subsequent time, it is not only related with the position at current time but also related with the position of multiple historical junctures.So
Meeting is so that predict that the accuracy of the position of pedestrian is lower by current single order Prediction of Markov mode.In order to improve prediction
Accuracy it is contemplated that being predicted by position of the high-order Prediction of Markov mode to pedestrian, namely considers multiple
The influence of historical juncture pedestrian position, the i.e. position of pedestrian's subsequent time and the position at current time are related, and current time
Position is related with the position of previous moment, and so on, realize the prediction to pedestrian position.
But during being realized by high-order Prediction of Markov mode to pedestrian position prediction, so that Markov
Exponentially type trend growth, computation complexity increase in process searched space, further such that the efficiency calculated is lower.And to interior
When pedestrian position is predicted in environment, computational efficiency is very important importance.Therefore, in order to accurate in raising prediction
Property while, guarantee computational efficiency, the embodiment of the present invention considers the similitude between track based on the thought frequently accessed, right
Pedestrian position is predicted.The process of pedestrian's prediction technique provided in an embodiment of the present invention is described in detail below.
The embodiment of the invention provides a kind of pedestrian position prediction techniques, as shown in Figure 1, comprising:
S101, obtains the position locus information for being predicted pedestrian, and position locus information includes being predicted pedestrian's different moments
Location information.
The embodiment of the present invention can be applied to electronic equipment, wherein electronic equipment may include desktop computer, portable
Computer, mobile terminal etc..
Being predicted pedestrian can be anyone moved in indoor environment.Position locus information such as timestamp,
XCoordinate, YCoordinate }, timestamp be predicted pedestrian movement to coordinate be (XCoordinate,
YCoordinate position).
During being predicted pedestrian movement, record is predicted the location information of pedestrian in real time, such as in t0Moment is predicted
Pedestrian movement to coordinate be (X0, Y0) position, in t1Moment, be predicted pedestrian movement to coordinate be (X1, Y1) position, etc.
Deng.In this way, need the position for being predicted pedestrian's subsequent time is predicted when, the current time of available record and
Multiple historical junctures before current time, this is predicted the position of pedestrian, that is, to obtain the position locus information for being predicted pedestrian.
S102 determines the motor pattern track for being predicted pedestrian according to position locus information.
Wherein, motor pattern track includes being predicted pedestrian in the region POI that different periods pass through, and the region POI indicates quilt
Visitation frequency reaches the region of preset condition.
The region POI is the key concept in GIS-Geographic Information System, indicates the region that pedestrian can be attracted to access.The present invention is real
The region that can be accessed by historical data analysis pedestrian's high density in example is applied, for example, the accessed frequency reaches preset condition
Region determines the region POI.Specifically, it is determined that the process in the region POI, as shown in Figure 2, comprising:
S201 obtains the position locus information of multiple pedestrians.
When pedestrian moves in environment indoors, the pedestrian is recorded in real time in the location information of different moments.I.e., it is possible to record
The location information of the multiple pedestrians moved in environment indoors, in this way, the position locus information of available multiple pedestrians.
S202 chooses dwell point.
Dwell point is pedestrian's significant location point during the motion, for example, it may be in preset time range not
The location point of change, or variation is less than the location point, etc. of threshold value in preset time range.Specifically, to dwell point
It chooses, a kind of mode, can choose by the position locus information of multiple pedestrians and continue motionless position in preset time range
It sets a little；Another way can be the point chosen and moved in preset range, specifically can use the location track of multiple pedestrians
The variance that preset time range internal coordinate changes in information extracts variance in the point set of preset range, and by the center of mass point of point set
As dwell point.
S203, dwell point cluster.
According to density-based algorithms, for example, DBSCAN algorithm, clusters dwell point, and cluster is obtained
Class cluster be determined as the region POI.
S204 obtains the region POI.
After determining the region POI, the position locus information for being predicted pedestrian can analyze, selection is predicted pedestrian movement's process
In significant location point, for example, the position that pedestrian's visitation frequency meets condition is predicted, e.g., by 5 times within 10 minutes
Location point；Determine the region POI, each position, which can correspond to, falls into the region POI, by the significant location point selected point
The region POI is not mapped to.It so, it is possible to obtain and be predicted pedestrian in the region POI that different periods pass through to get to being predicted to go
The motor pattern track of people.
S103 searches the matching history with motor pattern path matching from the multiple historical pattern tracks pre-saved
Mode trace, wherein historical pattern track includes history pedestrian in the region POI that different periods pass through.
Multiple historical pattern tracks are pre-saved in the embodiment of the present invention, multiple historical pattern tracks are respectively according to multiple
What the position locus information of history pedestrian determined.Specifically, historical pattern rail is determined according to the position locus information of history pedestrian
The process of mark, the position locus information of pedestrian is predicted with above-mentioned basis, and determination is predicted the corresponding motor pattern track of pedestrian
Process it is similar.In this way, the historical pattern track of the motor pattern path matching of pedestrian can be searched and is predicted, track is considered
Between similitude, and then according to matching historical pattern track the position for being predicted pedestrian is predicted.
Specifically, it can be searched and movement according to the motor pattern track similarity with multiple historical pattern tracks respectively
The matched matching historical pattern track of mode trace.Wherein it is possible to by calculating longest common subsequence (Longest Common
Subsequence, LCS) mode, calculate the motor pattern track similarity with historical pattern track respectively, LCS passes through dynamic
The thought of state planning searches for the identical subsequence of longest sequence in two or more sequences, in the embodiment of the present invention, by searching
Rope motor pattern track subsequence identical as the longest sequence of multiple historical pattern tracks, calculate motor pattern track respectively with go through
The similarity of logotype track.Alternatively, can be calculated in the embodiment of the present invention motor pattern track respectively with multiple historical patterns
The Euclidean distance of track determines whether motor pattern track matches with historical pattern track according to the difference of Euclidean distance, specifically
Ground determines motor pattern when can be the Euclidean distance between motor pattern track and historical pattern track less than preset threshold
Track and the historical pattern path matching, that is, determine that the historical pattern track is the matching historical pattern track found.Alternatively,
LCS mode can also be combined with Euclidean distance calculation, by way of layering, from the multiple history pre-saved
In mode trace, the matching historical pattern track with motor pattern path matching is searched, e.g., is first screened by Euclidean distance first
Partial history mode trace out, then matching historical pattern track is filtered out from partial history mode trace by LCS mode.
S104, according to matching historical pattern track, by Prediction of Markov mode, prediction is predicted the position of pedestrian.
Determine the candidate region POI for including in matching historical pattern track；Then by Prediction of Markov mode, to quilt
It predicts in the region POI that prediction pedestrian may access.Specifically, Ma Erke can be passed through according to all candidate regions POI
It is general that husband's mode calculates the transfer of the high-orders such as single order transition probability matrix, second order transition probability matrix or three rank transition probability matrixs
Rate matrix, then in single order transition probability matrix selection on the basis of the region POI at current time, current time it is next
Moment accesses in the probability in the different regions POI, the corresponding region POI is the position of prediction when probability highest；Alternatively, turning in second order
It moves in probability matrix and selects on the basis of the region POI at current time and the region POI of the last moment at current time,
The subsequent time at current time accesses in the probability in the different regions POI, the corresponding region POI is the position of prediction when probability highest
It sets；Alternatively, POI region, the POI area of the last moment at current time of the selection at current time in three rank transition probability matrixs
On the basis of the region POI of the last moment of domain and last moment, the subsequent time at current time accesses the different regions POI
Probability in, probability highest when the corresponding region POI be prediction position.
Alternatively, can directly calculate on the basis of the region POI at current time, the subsequent time access at current time is not
With the probability in the region POI, the position that the corresponding region POI when probability highest is prediction then can choose；Alternatively, can be direct
It calculates on the basis of the region POI at current time and the region POI of the last moment at current time, under current time
One moment accessed the probability in the different regions POI, then can choose the position that the corresponding region POI when probability highest is prediction；
Alternatively, can directly calculate in the region POI at current time, the region POI of the last moment at current time and last moment
Last moment the region POI on the basis of, the subsequent time at current time accesses the probability in the different regions POI, then can be with
The corresponding region POI is the position, etc. of prediction when select probability highest.
In the embodiment of the present invention, according to the position locus information of multiple pedestrians, the area POI frequently accessed by pedestrian is determined
Domain, and according to the position locus information of different pedestrians, determine the fortune for including pedestrian in the region point of interest POI that different periods pass through
Dynamic mode trace；And the matching historical pattern track of selection motor pattern path matching corresponding with pedestrian is predicted, and according to
Matching historical pattern track is predicted by the position for being predicted pedestrian.In this way, consider pedestrian movement during purpose with
And between pedestrian motion profile similitude, the position for being predicted pedestrian is predicted, can be improved the accuracy of prediction.
In order to allow to from the multiple historical pattern tracks pre-saved, search and motor pattern path matching
Matching historical pattern track, in the embodiment of the present invention, pre-save multiple historical pattern tracks.Wherein, multiple historical patterns
What the position locus information that track is respectively directed to each history pedestrian determined.Specifically, historical pattern track is pre-saved, is wrapped
It includes:
B1 obtains the position locus information of history pedestrian for each history pedestrian.
History pedestrian can be the pedestrian moved in historical time, in the indoor environment for being predicted pedestrian movement.It goes through
The position locus information of history pedestrian includes the location information of history pedestrian's different moments.
B2 is determined locating for history pedestrian in preset time range according to the position locus information of history pedestrian
The constant position in position is first stop.
B3 clusters first stop according to density-based algorithms, and the class cluster that cluster is obtained determines
For the region POI.
Above-mentioned steps B1-B3, it is similar with step S201-S204 in Fig. 2 in above-described embodiment, in above-described embodiment into
It has gone and has described in detail, just repeated no more here.
Each first stop is mapped to the region POI by B4, obtains the corresponding historical pattern track of history pedestrian.
After determining the region POI, each first stop correspondence is fallen into the region POI, and then available first stops
Stationary point to POI point to region mapping relations:
POIi∈{Ps1,Ps2,…,Psn}
Wherein, POIiFor i-th of POI, Ps1For the 1st first stop, Ps2For the 2nd first stop, PsnIt is n-th
A first stop.
Since intersection is empty characteristic, the POI that first stop can be determined by mapping relations between class cluster in cluster
Region is unique.
In this way, each first stop correspondence is passed through to the region POI, available history pedestrian in different periods
The region POI, it can obtain the corresponding historical pattern track of history pedestrian.As shown in figure 3, history pedestrian A passes through the region POI:
POIn-1, POIn, POIn+1, wherein n-1, n, n+1 are the label in the region POI, so can then determine that history pedestrian A is corresponding
Historical pattern track are as follows:
And the region POI in historical pattern track including, movement state information including history pedestrian etc. can be corresponded to
POI regional model composition.Specifically, such as pedestrian A, POI regional model can be indicated are as follows:
Wherein,It is pedestrian A in tiThe corresponding POI regional model in the region POI locating for moment,oThis is left for pedestrian
At the time of the region POI,Yi AIt is pedestrian A in tiThe location information at moment, POInFor the region POI belonging to the location information,It is pedestrian A in tiMovement state information before moment, movement state information may include into the speed before the region POI
The factors such as degree, deflection and the speed behind the region POI, deflection are left, as described by following formula:
Wherein,For pedestrian A enter the region POI before flashy speed,Indicate that pedestrian A is leaving this
Flashy speed behind the region POI,For pedestrian A enter the region POI before flashy deflection,Indicate row
People A flashy deflection, t after leaving the region POIstayStay time for pedestrian A in the region POI can indicate
For to-ti。
Specifically, the corresponding POI in each region POI for including in historical pattern track can be determined in the embodiment of the present invention
Regional model obtains the historical pattern track comprising richer information.
B5 saves historical pattern track.
The corresponding history of each history pedestrian can be obtained in the process that above-mentioned B1-B4 is carried out for each history pedestrian
Mode trace.In this way, the corresponding multiple historical pattern tracks of multiple history pedestrians can be saved.
On the basis of the above embodiments, according to position locus information, the motor pattern track for being predicted pedestrian is determined, such as
Shown in Fig. 4, comprising:
S401, according to the position locus information for being predicted pedestrian, determination is predicted in preset time range locating for pedestrian
The constant position in position be the second dwell point.
Step B2 in specifically process, with above-described embodiment: it according to the position locus information of history pedestrian, determines pre-
If the constant position of the location of history pedestrian is that the process of first stop is similar in time range, step is being executed
Referring to above-mentioned steps B2 when S401, just repeat no more here.
Each second dwell point is mapped to the region POI, obtains motor pattern track by S402.
Each first stop is mapped to the region POI, is somebody's turn to do by step B4 in specifically process, with above-described embodiment
The process of the corresponding historical pattern track of history pedestrian is similar, when executing step S402 referring to above-mentioned steps B4, here
Just repeat no more.
It is obtaining being predicted the corresponding motor pattern track of pedestrian, and is pre-saving the basis of multiple historical pattern tracks
On, the matching historical pattern rail with motor pattern path matching can be searched from the multiple historical pattern tracks pre-saved
Mark.On the basis of above-mentioned embodiment illustrated in fig. 1, specifically, may include:
C1 calculates separately the first of motor pattern track and multiple historical pattern tracks according to Euclidean distance calculation
Similarity.
Specifically, comprising:
C10 determines the POI sequence of all region POI compositions.
Indoor environment important difference compared with outdoor environment is that indoor environment range is small, in this way, the region POI determined
It is limited and can determine.The embodiment of the present invention is based on indoor environment, and the number in the determining region POI is it is believed that so may be used
To determine dimensional vector for the building of the region POI, that is, determine POI sequence.
For example, POI sequence are as follows: α={ POI1→POI2...→POIn→…}。
C11 determines corresponding first track vector in motor pattern track.
Whether the value of element successively indicates in motor pattern track to include corresponding position in POI sequence in first track vector
The region POI at place.
For example, motor pattern track be Tr=... → POIn-1→POIn→POIn+1→... }, can indicate by 0 should
It indicates to pass through 1 time or multiple POI in the motor pattern track without the region POI in motor pattern track, 1, can determine
First track vector αi={ 0,0 ..., 1,1,1 }, wherein first 0 expression motor pattern track is without POI1, second
A 0 indicates the motor pattern track without POI2, POI is passed through in first 1 expression motor pattern trackn-1, second 1 table
Show the motor pattern track by POIn, POI is passed through in 1 expression motor pattern track of thirdn+1。
In the embodiment of the present invention, can also using first track vector as the expansion POI of motor pattern track access to
Amount.
Herein in motor pattern track (POIE→POIC→POIA→POIB→POIF) in expand POI access vector α, it is as follows
Formula indicates:
(POIE→POIC→POIA→POIB→POIF, αi)
Wherein, POI accesses vector αiIt is following to indicate:
Wherein,For the first track vector αiIn element, i be indicate element numerals change
Amount.
C12 determines corresponding second track vector in multiple historical pattern tracks respectively.
The value of element indicates whether historical pattern track includes corresponding position in POI sequence in second track vector
The region POI.
Specifically, it is determined that the process of the second track vector, the process class of the first track vector is determined with above-mentioned steps C11
Seemingly, it referring to above-mentioned C11 process when executing step C12, just repeats no more here.
C13, calculate separately the Euclidean of corresponding second track vector of the first track vector and multiple historical pattern tracks away from
From.
Specifically, following formula can be passed through:
Calculate Euclidean distance, wherein dis (αi,αj) it is the first track vector αiWith the second track vector αjEuclidean distance,For the first track vector αiIn element,For the second track vector αjIn
Element.
In order to calculate simplicity, the first track vector αiWith the second track vector αjIn element value be only 0 or 1, can recognize
It is made of for vector element the variable of Boolean type, in this way, calculating process, which can simplify, makees every after inner product be added for vector
Process, following formula:
It is so available:Wherein, k is the variable for indicating vector element label,ForWithMake the result of inner product.
C14 passes through formula according to Euclidean distance:Calculate motor pattern rail
Mark the first similarity with multiple historical pattern tracks respectively.
Wherein, sim (αi,αj) it is the first track vector αiWith the second track vector αjThe first similarity, dis (αi,αj)
For the first track vector αiWith the second track vector αjEuclidean distance, dim (α) be POI sequence α dimension.
C2 chooses preset quantity historical pattern track, goes through as primary election according to the sequence of the first similarity from high to low
Logotype track.
Preset quantity can select according to actual needs, for example, 20,10, etc..
Because Euclidean distance calculation be with sequence it is unrelated, as long as the element between vector is consistent, Ji Keji
Calculate the similarity between vector, computation complexity is low, and computational efficiency is high, in this way, in the embodiment of the present invention first according to it is European away from
From calculation, selected part can be improved computational efficiency as primary election historical pattern track from multiple historical pattern tracks,
Calculation amount can also be reduced simultaneously.
C3 calculates the second similarity of motor pattern track and primary election historical pattern track according to LCS calculation.
Specifically, comprising:
C31 calculates the longest common subsequence length of motor pattern track and primary election historical pattern track.
By following recurrence formula, determine that sequence identifier is the motor pattern track Trj of mmThe primary election for being n with sequence identifier
Historical pattern track TrjnLongest common subsequence length L (m, n):
Wherein, trjmFor motor pattern track TrjmIn element, trjnFor primary election historical pattern track TrjnIn member
Element.
C32 determines the minimum length in motor pattern track and primary election historical pattern track.
C33 passes through formula according to longest common subsequence length and minimum length:Calculate the second similarity.
Wherein, simi2(Trjm,Trjn) it is the motor pattern track Trj that sequence identifier is mmIt is the first of n with sequence identifier
Select historical pattern track TrjnThe second similarity, L (m, n) be longest common subsequence length, min (Lm,Ln) it is minimum long
Degree.
C4 chooses preset quantity primary election historical pattern track according to the sequence of the second similarity from high to low, as
With historical pattern track.
Preset quantity can select according to actual needs, for example, 10,5, etc..
Motor pattern track is the very strong data of Ordered Dependency, according only to the Euclidean distance calculation unrelated with sequence
Motor pattern corresponding with pedestrian is predicted track is chosen, the accuracy of calculating can be restricted.The embodiment of the present invention passes through layering
The mode of calculating, i.e., according to Euclidean distance calculation, selected part is as primary election history from multiple historical pattern tracks
On the basis of mode trace, matching historical pattern track is chosen from primary election historical pattern track by LCS calculation.So
It can guarantee the accuracy calculated while improving computational efficiency, can guarantee the high efficiency and accuracy calculated simultaneously.
On the basis of the above embodiments, quilt is predicted by Prediction of Markov mode according to matching historical pattern track
The position for predicting pedestrian may include:
D1 determines the candidate region POI for including in matching historical pattern track.
It determines that matching historical pattern track can be 1, or is also possible to multiple.For example, matching historical pattern track
For (POIE→POIC→POIA→POIB→POIF), then it can determine the candidate region POI are as follows: POIE、POIC、POIA、POIB、
POIF。
D2, according to the candidate region POI, by second order Prediction of Markov formula, calculating is predicted pedestrian and is moved to difference
The probability in the candidate region POI.
In the embodiment of the present invention, determine that the POI regional ensemble that cluster obtains is state space, such as:
{Xn: n=POI1,POI2,POI3,POI4..., POIn}
By following second order Prediction of Markov formula, calculating is predicted pedestrian and is moved to the general of the different candidate regions POI
Rate.
ByIt can obtain:
D3 determines the corresponding candidate region POI, the position for being predicted pedestrian for prediction when probability meets condition.
Probability, which meets condition, can be probability greater than preset threshold, alternatively, probability is in preset threshold range etc..
Corresponding unique candidate region POI when probability meets condition, it is determined that unique candidate region POI is prediction
The position for being predicted pedestrian.
But for the position prediction of indoor environment, since the region POI is less, cause the state of Markov model empty
Between dimension it is smaller, therefore it is relatively more the equal conflict situations of probability between transfer mode occur, when there is the conflict of this probability,
A kind of mode is to randomly select a transfer mode as prediction result.In a kind of preferred embodiment of the embodiment of the present invention,
It determines the corresponding candidate region POI, the position for being predicted pedestrian for prediction when probability meets condition, may include:
D31 corresponds to multiple candidate regions POI when probability highest, calculates separately the region POI that motor pattern track includes
In include running state information and multiple candidate regions POI movement state information for including matching degree.
In the embodiment of the present invention, the region POI that motor pattern track includes, by including the motion state letter for being predicted pedestrian
The POI regional model of breath forms；The region POI that matching historical pattern track includes is believed by the motion state for including history pedestrian
The POI regional model of breath forms.
In the embodiment of the present invention, in the basic unit POI model of component movement mode trace and historical pattern track, include
Pedestrian passes through row in POI model in the case where probability of occurrence conflict in the movement state information of the region the POI dwell phase
The movement state information of people is associated with movement state information of the collision candidates under the POI model and matches, specific to match
It is as follows to spend calculation formula:
Wherein,tstayA is to be predicted to move in the corresponding POI regional model of pedestrian
The element of status information,tstayP is to move in the corresponding POI regional model of history pedestrian
The element of status information,The running state information for including in the region POI for including for motor pattern track and candidate
The matching degree for the movement state information that the region POI includes.
D32, the corresponding candidate region POI when by matching degree highest are determined as the position for being predicted pedestrian of prediction.
In above-mentioned matching degree calculation formula, it is corresponding with history pedestrian that each element in the movement state information of pedestrian will be predicted
Movement state information carry out apart from calculating, and be normalized to obtain that maximum value is 5 using each element maximum value
With degree score, all matching degree scores are ranked up, when highest scoring the corresponding candidate region POI, that is, can be identified as pre-
Survey the predicted position of pedestrian.
The embodiment of the invention provides a kind of pedestrian position prediction meanss, as shown in Figure 5, comprising:
Module 501 is obtained, for obtaining the position locus information for being predicted pedestrian, position locus information includes being predicted to go
The location information of people's different moments；
Determining module 502, for determining the motor pattern track for being predicted pedestrian according to position locus information, wherein fortune
Dynamic mode trace includes being predicted pedestrian in the region point of interest POI that different periods pass through, and the region POI indicates that the accessed frequency reaches
To the region of preset condition；
Searching module 503, for searching and motor pattern path matching from the multiple historical pattern tracks pre-saved
Matching historical pattern track, wherein historical pattern track includes history pedestrian in the region POI that different periods pass through；
Prediction module 504, for according to matching historical pattern track, by Prediction of Markov mode, prediction to be predicted
The position of pedestrian.
In the embodiment of the present invention, according to the position locus information of multiple pedestrians, the area POI frequently accessed by pedestrian is determined
Domain, and according to the position locus information of different pedestrians, determine the fortune for including pedestrian in the region point of interest POI that different periods pass through
Dynamic mode trace；And the matching historical pattern track of selection motor pattern path matching corresponding with pedestrian is predicted, and according to
Matching historical pattern track is predicted by the position for being predicted pedestrian.In this way, consider pedestrian movement during purpose with
And between pedestrian motion profile similitude, the position for being predicted pedestrian is predicted, can be improved the accuracy of prediction.
Optionally, what the position locus information that multiple historical pattern tracks are respectively directed to each history pedestrian determined；
The device further include: preserving module；
Preserving module, comprising:
Acquisition submodule obtains the position locus information of history pedestrian for being directed to each history pedestrian；
First determines submodule, for the position locus information according to history pedestrian, determines in preset time range
The constant position of the location of history pedestrian is first stop；
Second determines submodule, for clustering to first stop, and will gather according to density-based algorithms
The class cluster that class obtains is determined as the region POI；
It is corresponding to obtain history pedestrian for each first stop to be mapped to the region POI for first mapping submodule
Historical pattern track；
Submodule is saved, for saving historical pattern track.
Optionally, determining module 502, comprising:
Third determines submodule, for determining in preset time range according to the position locus information for being predicted pedestrian
Being predicted the constant position of the location of pedestrian is the second dwell point；
Second mapping submodule obtains motor pattern track for each second dwell point to be mapped to the region POI.
Optionally, searching module 503, comprising:
First computational submodule, for calculating separately motor pattern track and being gone through with multiple according to Euclidean distance calculation
First similarity of logotype track；
First chooses submodule, for the sequence according to the first similarity from high to low, chooses preset quantity history mould
Formula track, as primary election historical pattern track；
Second computational submodule, for according to longest common subsequence LCS calculation, calculating motor pattern track and just
Select the second similarity of historical pattern track；
Second chooses submodule, for the sequence according to the second similarity from high to low, chooses preset quantity primary election and goes through
Logotype track, as matching historical pattern track.
Optionally, prediction module 504, comprising:
4th determines submodule, for determining the candidate region POI for including in matching historical pattern track；
Third computational submodule, for being calculated pre- according to the candidate region POI by second order Prediction of Markov formula
Survey the probability that pedestrian is moved to the different candidate regions POI；
5th determines submodule, and the corresponding candidate region POI when for determining that probability meets condition is predicted for prediction
The position of pedestrian.
Optionally, the region POI that motor pattern track includes, by including the POI for being predicted the movement state information of pedestrian
Regional model composition；The region POI that matching historical pattern track includes, by the POI of the movement state information including history pedestrian
Regional model composition；
5th determines submodule, comprising:
First computing unit calculates separately motor pattern track for corresponding to multiple candidate regions POI when probability highest
Including the region POI in include running state information and multiple candidate regions POI movement state information for including matching degree；
First determination unit, the corresponding candidate region POI when for by matching degree highest, be determined as predicting is predicted to go
The position of people.
Optionally, the first computational submodule, comprising:
Second determination unit, for determining the POI sequence of all region POI compositions；
Third determination unit, for determining corresponding first track vector in motor pattern track；It is first in first track vector
The value of element successively indicate in motor pattern track whether include corresponding position in POI sequence the region POI；
4th determination unit, for determining corresponding second track vector in multiple historical pattern tracks, the second track respectively
In vector element value indicate historical pattern track whether include corresponding position in POI sequence the region POI；
Second computing unit, for calculating separately corresponding second track of the first track vector and multiple historical pattern tracks
The Euclidean distance of vector；
Third computing unit, for passing through formula according to Euclidean distance:
Calculate motor pattern track the first similarity with multiple historical pattern tracks respectively；
Wherein, sim (αi,αj) it is the first track vector αiWith the second track vector αjThe first similarity, dis (αi,αj)
For the first track vector αiWith the second track vector αjEuclidean distance, dim (α) be POI sequence α dimension.
Optionally, the second computational submodule, comprising:
4th computing unit, it is long for calculating motor pattern track and the longest common subsequence of primary election historical pattern track
Degree；
5th determination unit, for determining the minimum length in motor pattern track and primary election historical pattern track；
5th computing unit, for passing through formula according to longest common subsequence length and minimum length:Calculate the second similarity；
Wherein, simi2(Trjm,Trjn) it is the motor pattern track Trj that sequence identifier is mmIt is the first of n with sequence identifier
Select historical pattern track TrjnThe second similarity, L (m, n) be longest common subsequence length, min (Lm,Ln) it is minimum long
Degree.
It should be noted that pedestrian position prediction meanss provided in an embodiment of the present invention, are pre- using above-mentioned pedestrian position
The device of survey method, then all embodiments of above-mentioned pedestrian position prediction technique are suitable for the device, and can reach identical
Or similar beneficial effect.
The embodiment of the invention also provides a kind of pre- measurement equipments of pedestrian position, as shown in fig. 6, including processor 601, communication
Interface 602, memory 603 and communication bus 604, wherein processor 601, communication interface 602, memory 603 are total by communication
Line 604 completes mutual communication,
Memory 603, for storing computer program；
Processor 601 when for executing the program stored on memory 603, realizes above-mentioned pedestrian position prediction technique
Method and step.
In the embodiment of the present invention, according to the position locus information of multiple pedestrians, the area POI frequently accessed by pedestrian is determined
Domain, and according to the position locus information of different pedestrians, determine the fortune for including pedestrian in the region point of interest POI that different periods pass through
Dynamic mode trace；And the matching historical pattern track of selection motor pattern path matching corresponding with pedestrian is predicted, and according to
Matching historical pattern track is predicted by the position for being predicted pedestrian.In this way, consider pedestrian movement during purpose with
And between pedestrian motion profile similitude, the position for being predicted pedestrian is predicted, can be improved the accuracy of prediction.
The communication bus that the above-mentioned pre- measurement equipment of pedestrian position is mentioned can be Peripheral Component Interconnect standard (Peripheral
Component Interconnect, PCI) bus or expanding the industrial standard structure (ExtendedIndustry Standard
Architecture, EISA) bus etc..The communication bus can be divided into address bus, data/address bus, control bus etc..For just
It is only indicated with a thick line in expression, figure, it is not intended that an only bus or a type of bus.
Communication interface is for the communication between the pre- measurement equipment of above-mentioned pedestrian position and other equipment.
Memory may include random access memory (Random Access Memory, RAM), also may include non-easy
The property lost memory (Non-Volatile Memory, NVM), for example, at least a magnetic disk storage.Optionally, memory may be used also
To be storage device that at least one is located remotely from aforementioned processor.
Above-mentioned processor can be general processor, including central processing unit (Central Processing Unit,
CPU), network processing unit (Network Processor, NP) etc.；It can also be digital signal processor (Digital Signal
Processing, DSP), it is specific integrated circuit (Application Specific Integrated Circuit, ASIC), existing
It is field programmable gate array (Field-Programmable Gate Array, FPGA) or other programmable logic device, discrete
Door or transistor logic, discrete hardware components.
The embodiment of the invention provides a kind of computer readable storage medium, meter is stored in computer readable storage medium
Calculation machine program realizes the method and step of above-mentioned pedestrian position prediction technique when computer program is executed by processor.
In the embodiment of the present invention, according to the position locus information of multiple pedestrians, the area POI frequently accessed by pedestrian is determined
Domain, and according to the position locus information of different pedestrians, determine the fortune for including pedestrian in the region point of interest POI that different periods pass through
Dynamic mode trace；And the matching historical pattern track of selection motor pattern path matching corresponding with pedestrian is predicted, and according to
Matching historical pattern track is predicted by the position for being predicted pedestrian.In this way, consider pedestrian movement during purpose with
And between pedestrian motion profile similitude, the position for being predicted pedestrian is predicted, can be improved the accuracy of prediction.
It should be noted that, in this document, relational terms such as first and second and the like are used merely to a reality
Body or operation are distinguished with another entity or operation, are deposited without necessarily requiring or implying between these entities or operation
In any actual relationship or order or sequence.Moreover, the terms "include", "comprise" or its any other variant are intended to
Non-exclusive inclusion, so that the process, method, article or equipment including a series of elements is not only wanted including those
Element, but also including other elements that are not explicitly listed, or further include for this process, method, article or equipment
Intrinsic element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that
There is also other identical elements in process, method, article or equipment including the element.
Each embodiment in this specification is all made of relevant mode and describes, same and similar portion between each embodiment
Dividing may refer to each other, and each embodiment focuses on the differences from other embodiments.Especially for device,
For equipment and computer readable storage medium embodiment, since it is substantially similar to the method embodiment, so the comparison of description
Simply, the relevent part can refer to the partial explaination of embodiments of method.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the scope of the present invention.It is all
Any modification, equivalent replacement, improvement and so within the spirit and principles in the present invention, are all contained in protection scope of the present invention
It is interior.
Claims (10)
1. a kind of pedestrian position prediction technique characterized by comprising
It obtains and is predicted the position locus information of pedestrian, the position locus information includes described being predicted pedestrian's different moments
Location information；
According to the position locus information, the motor pattern track of pedestrian is predicted described in determination, wherein the motor pattern rail
Mark is predicted pedestrian in the region point of interest POI that different periods pass through described in including, and the region POI indicates that the accessed frequency reaches
To the region of preset condition；
From the multiple historical pattern tracks pre-saved, the matching historical pattern rail with the motor pattern path matching is searched
Mark, wherein the historical pattern track includes history pedestrian in the region POI that different periods pass through；
According to the matching historical pattern track position of pedestrian is predicted by Prediction of Markov mode described in prediction.
2. the method according to claim 1, wherein multiple historical pattern tracks are respectively to be directed to each history row
What the position locus information of people determined；
Pre-save the historical pattern track, comprising:
For each history pedestrian, the position locus information of history pedestrian is obtained；
According to the position locus information of history pedestrian, determine that the location of history pedestrian is constant in preset time range
Position be first stop；
According to density-based algorithms, the first stop is clustered, and the class cluster that cluster obtains is determined as
The region POI；
Each first stop is mapped into the region POI, obtains the corresponding historical pattern track of history pedestrian；
Save the historical pattern track.
3. according to the method described in claim 2, determining the quilt it is characterized in that, described according to the position locus information
Predict the motor pattern track of pedestrian, comprising:
According to the position locus information for being predicted pedestrian, described in preset time range be predicted locating for pedestrian is determined
The constant position in position is the second dwell point；
Each second dwell point is mapped into the region POI, obtains the motor pattern track.
4. the method according to claim 1, wherein described from the multiple historical pattern tracks pre-saved,
Search the matching historical pattern track with the motor pattern path matching, comprising:
According to Euclidean distance calculation, it is similar to the first of multiple historical pattern tracks to calculate separately the motor pattern track
Degree；
According to the sequence of first similarity from high to low, preset quantity historical pattern track is chosen, as primary election history
Mode trace；
According to longest common subsequence LCS calculation, the motor pattern track and primary election historical pattern track are calculated
The second similarity；
According to the sequence of second similarity from high to low, preset quantity primary election historical pattern track is chosen, as
The matching historical pattern track.
5. passing through horse the method according to claim 1, wherein described according to the matching historical pattern track
Er Kefu prediction mode is predicted the position of pedestrian described in prediction, comprising:
Determine the candidate region POI for including in the matching historical pattern track；
Pedestrian is predicted described in calculating and is moved to difference by second order Prediction of Markov formula according to the candidate region POI
The probability in the candidate region POI；
Determine the corresponding candidate region POI, the position to be predicted pedestrian described in prediction when the probability meets condition.
6. according to the method described in claim 5, it is characterized in that, the region POI that the motor pattern track includes, by
POI regional model composition including the movement state information for being predicted pedestrian；The matching historical pattern track includes
The region POI is made of the POI regional model of the movement state information including history pedestrian；
The determination probability corresponding candidate region POI when meeting condition, to be predicted pedestrian's described in prediction
Position, comprising:
Multiple candidate regions POI are corresponded to when the probability highest, calculate separately the region POI that the motor pattern track includes
In include running state information and multiple candidate regions POI movement state information for including matching degree；
The corresponding candidate region POI when by the matching degree highest is determined as the position for being predicted pedestrian of prediction.
7. according to the method described in claim 4, calculating separately institute it is characterized in that, described according to Euclidean distance calculation
State the first similarity of motor pattern track Yu multiple historical pattern tracks, comprising:
Determine the POI sequence of all region POI compositions；
Determine corresponding first track vector in the motor pattern track；The value of element successively indicates in first track vector
In the motor pattern track whether include the POI sequence in corresponding position the region POI；
Corresponding second track vector in multiple historical pattern tracks is determined respectively, the value of element in second track vector
Indicate the historical pattern track whether include corresponding position in the POI sequence the region POI；
Calculate separately the Euclidean of corresponding second track vector of first track vector and multiple historical pattern tracks away from
From；
According to the Euclidean distance, pass through formula:Calculate the motor pattern rail
Mark the first similarity with multiple historical pattern tracks respectively；
Wherein, sim (αi,αj) it is the first track vector αiWith the second track vector αjThe first similarity, dis (αi,
αj) it is the first track vector αiWith the second track vector αjEuclidean distance, the dim (α) be the POI sequence α
Dimension.
8. according to the method described in claim 4, it is characterized in that, described according to longest common subsequence LCS calculation, meter
Calculate the second similarity of the motor pattern track Yu primary election historical pattern track, comprising:
Calculate the longest common subsequence length of the motor pattern track Yu primary election historical pattern track；
Determine the minimum length in the motor pattern track and primary election historical pattern track；
According to the longest common subsequence length and the minimum length, pass through formula:
Calculate second similarity；
Wherein, the simi2(Trjm,Trjn) it is the motor pattern track Trj that sequence identifier is mmIt is n's with sequence identifier
Primary election historical pattern track TrjnThe second similarity, the L (m, n) be the longest common subsequence length, it is described
min(Lm,Ln) it is the minimum length.
9. a kind of pedestrian position prediction meanss characterized by comprising
Module is obtained, for obtaining the position locus information for being predicted pedestrian, the position locus information includes described is predicted
The location information of pedestrian's different moments；
Determining module, for being predicted the motor pattern track of pedestrian described according to the position locus information, determining, wherein
The motor pattern track is predicted pedestrian in the region point of interest POI that different periods pass through, the region POI table described in including
Show that the accessed frequency reaches the region of preset condition；
Searching module, for from the multiple historical pattern tracks pre-saved, searching and the motor pattern path matching
Match historical pattern track, wherein the historical pattern track includes history pedestrian in the area POI that different periods pass through
Domain；
Prediction module, for being predicted described in prediction according to the matching historical pattern track by Prediction of Markov mode
The position of pedestrian.
10. a kind of pre- measurement equipment of pedestrian position, which is characterized in that including processor, communication interface, memory and communication bus,
Wherein, the processor, the communication interface, the memory complete mutual communication by the communication bus；
The memory, for storing computer program；
The processor when for executing the program stored on the memory, realizes any side claim 1-8
Method step.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810651020.3A CN109029446B (en) | 2018-06-22 | 2018-06-22 | Pedestrian position prediction method, device and equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810651020.3A CN109029446B (en) | 2018-06-22 | 2018-06-22 | Pedestrian position prediction method, device and equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109029446A true CN109029446A (en) | 2018-12-18 |
| CN109029446B CN109029446B (en) | 2020-11-20 |
Family
ID=64610058
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810651020.3A Active CN109029446B (en) | 2018-06-22 | 2018-06-22 | Pedestrian position prediction method, device and equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109029446B (en) |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109991641A (en) * | 2019-04-10 | 2019-07-09 | 广东工业大学 | A kind of acquisition methods, the apparatus and system of position of mobile equipment information |
| CN110245554A (en) * | 2019-04-29 | 2019-09-17 | 中科智城(广州)信息科技有限公司 | A kind of method, system platform and the storage medium of the early warning of pedestrian movement's trend |
| CN110442662A (en) * | 2019-07-08 | 2019-11-12 | 清华大学 | A kind of method and information-pushing method of determining customer attribute information |
| CN110708658A (en) * | 2019-09-05 | 2020-01-17 | 中国联合网络通信集团有限公司 | Data transmission method and base station |
| CN110726418A (en) * | 2019-10-10 | 2020-01-24 | 北京百度网讯科技有限公司 | Method, device and equipment for determining interest point region and storage medium |
| CN110781415A (en) * | 2019-09-16 | 2020-02-11 | 腾讯科技（深圳）有限公司 | Interest point recommendation method, device, equipment and medium |
| CN111132029A (en) * | 2019-10-16 | 2020-05-08 | 张苏 | Positioning method and device based on terrain constraint |
| CN111239727A (en) * | 2020-02-26 | 2020-06-05 | 深圳雷研技术有限公司 | Passenger counting method and communication equipment |
| CN111401531A (en) * | 2020-04-24 | 2020-07-10 | 中国人民解放军国防科技大学 | Trajectory prediction method and system |
| CN112805730A (en) * | 2019-12-30 | 2021-05-14 | 华为技术有限公司 | Trajectory prediction method and related equipment |
| CN114312829A (en) * | 2021-12-06 | 2022-04-12 | 广州文远知行科技有限公司 | Pedestrian trajectory prediction method and device, electronic equipment and storage medium |
| CN114511999A (en) * | 2020-11-17 | 2022-05-17 | 郑州宇通客车股份有限公司 | Pedestrian behavior prediction method and device |
| CN115408452A (en) * | 2022-11-02 | 2022-11-29 | 中南大学 | Urban facility association pattern mining method and related equipment |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103326903A (en) * | 2013-07-05 | 2013-09-25 | 华北电力大学 | Hidden-Markov-based Internet network delay forecasting method |
| US20140274109A1 (en) * | 2013-03-12 | 2014-09-18 | Qualcomm Incorporated | Positioning performance prediction |
| CN105488213A (en) * | 2015-12-11 | 2016-04-13 | 哈尔滨工业大学深圳研究生院 | LBS-oriented individual recommendation method based on Markov prediction algorithm |
| CN105825297A (en) * | 2016-03-11 | 2016-08-03 | 山东大学 | Markov-model-based position prediction method |
| CN106488405A (en) * | 2016-12-29 | 2017-03-08 | 电子科技大学 | A kind of position predicting method merging individuality and neighbour's movement law |
| CN107784046A (en) * | 2016-11-14 | 2018-03-09 | 平安科技（深圳）有限公司 | POI treating method and apparatus |
| CN107977734A (en) * | 2017-11-10 | 2018-05-01 | 河南城建学院 | A kind of Forecasting Methodology based on mobile Markov model under space-time big data |
-
2018
- 2018-06-22 CN CN201810651020.3A patent/CN109029446B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140274109A1 (en) * | 2013-03-12 | 2014-09-18 | Qualcomm Incorporated | Positioning performance prediction |
| CN103326903A (en) * | 2013-07-05 | 2013-09-25 | 华北电力大学 | Hidden-Markov-based Internet network delay forecasting method |
| CN105488213A (en) * | 2015-12-11 | 2016-04-13 | 哈尔滨工业大学深圳研究生院 | LBS-oriented individual recommendation method based on Markov prediction algorithm |
| CN105825297A (en) * | 2016-03-11 | 2016-08-03 | 山东大学 | Markov-model-based position prediction method |
| CN107784046A (en) * | 2016-11-14 | 2018-03-09 | 平安科技（深圳）有限公司 | POI treating method and apparatus |
| CN106488405A (en) * | 2016-12-29 | 2017-03-08 | 电子科技大学 | A kind of position predicting method merging individuality and neighbour's movement law |
| CN107977734A (en) * | 2017-11-10 | 2018-05-01 | 河南城建学院 | A kind of Forecasting Methodology based on mobile Markov model under space-time big data |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109991641A (en) * | 2019-04-10 | 2019-07-09 | 广东工业大学 | A kind of acquisition methods, the apparatus and system of position of mobile equipment information |
| CN110245554A (en) * | 2019-04-29 | 2019-09-17 | 中科智城(广州)信息科技有限公司 | A kind of method, system platform and the storage medium of the early warning of pedestrian movement's trend |
| CN110245554B (en) * | 2019-04-29 | 2021-06-29 | 中科智城(广州)信息科技有限公司 | Pedestrian movement trend early warning method, system platform and storage medium |
| CN110442662A (en) * | 2019-07-08 | 2019-11-12 | 清华大学 | A kind of method and information-pushing method of determining customer attribute information |
| CN110708658B (en) * | 2019-09-05 | 2020-12-15 | 中国联合网络通信集团有限公司 | Data transmission method and base station |
| CN110708658A (en) * | 2019-09-05 | 2020-01-17 | 中国联合网络通信集团有限公司 | Data transmission method and base station |
| CN110781415A (en) * | 2019-09-16 | 2020-02-11 | 腾讯科技（深圳）有限公司 | Interest point recommendation method, device, equipment and medium |
| CN110726418A (en) * | 2019-10-10 | 2020-01-24 | 北京百度网讯科技有限公司 | Method, device and equipment for determining interest point region and storage medium |
| US11740102B2 (en) | 2019-10-10 | 2023-08-29 | Beijing Baidu Netcom Science Technology Co., Ltd. | Method, apparatus, device and storage medium for determining point of interest area |
| CN111132029A (en) * | 2019-10-16 | 2020-05-08 | 张苏 | Positioning method and device based on terrain constraint |
| CN112805730A (en) * | 2019-12-30 | 2021-05-14 | 华为技术有限公司 | Trajectory prediction method and related equipment |
| CN111239727A (en) * | 2020-02-26 | 2020-06-05 | 深圳雷研技术有限公司 | Passenger counting method and communication equipment |
| CN111239727B (en) * | 2020-02-26 | 2020-12-29 | 深圳雷研技术有限公司 | Passenger counting method and communication equipment |
| CN111401531A (en) * | 2020-04-24 | 2020-07-10 | 中国人民解放军国防科技大学 | Trajectory prediction method and system |
| CN114511999A (en) * | 2020-11-17 | 2022-05-17 | 郑州宇通客车股份有限公司 | Pedestrian behavior prediction method and device |
| CN114511999B (en) * | 2020-11-17 | 2023-09-01 | 宇通客车股份有限公司 | Pedestrian behavior prediction method and device |
| CN114312829A (en) * | 2021-12-06 | 2022-04-12 | 广州文远知行科技有限公司 | Pedestrian trajectory prediction method and device, electronic equipment and storage medium |
| CN114312829B (en) * | 2021-12-06 | 2024-04-23 | 广州文远知行科技有限公司 | Pedestrian track prediction method and device, electronic equipment and storage medium |
| CN115408452A (en) * | 2022-11-02 | 2022-11-29 | 中南大学 | Urban facility association pattern mining method and related equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109029446B (en) | 2020-11-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109029446A (en) | A kind of pedestrian position prediction technique, device and equipment | |
| Wang et al. | SPORE: A sequential personalized spatial item recommender system | |
| Rong et al. | Du-parking: Spatio-temporal big data tells you realtime parking availability | |
| Chen et al. | Mining moving patterns for predicting next location | |
| Zheng et al. | Detecting collective anomalies from multiple spatio-temporal datasets across different domains | |
| CN106912015B (en) | Personnel trip chain identification method based on mobile network data | |
| Lu et al. | Personalized trip recommendation with multiple constraints by mining user check-in behaviors | |
| US10068024B2 (en) | Method and apparatus for correlating and viewing disparate data | |
| CN104123398B (en) | A kind of information-pushing method and device | |
| Lian et al. | Collaborative filtering meets next check-in location prediction | |
| Chen et al. | A system for destination and future route prediction based on trajectory mining | |
| Chen et al. | A personal route prediction system based on trajectory data mining | |
| US20160125307A1 (en) | Air quality inference using multiple data sources | |
| Chen et al. | Predicting next locations with object clustering and trajectory clustering | |
| CN106709606A (en) | Personalized scene prediction method and apparatus | |
| CN110267206B (en) | User position prediction method and device | |
| Li et al. | Robust inferences of travel paths from GPS trajectories | |
| CN110442662B (en) | Method for determining user attribute information and information push method | |
| CN108986453A (en) | A kind of traffic movement prediction method based on contextual information, system and device | |
| Huang et al. | Unsupervised interesting places discovery in location-based social sensing | |
| CN109190656B (en) | Indoor semantic track marking and complementing method under low-sampling positioning environment | |
| Lin et al. | Successive poi recommendation with category transition and temporal influence | |
| CN113486927A (en) | Unsupervised track access place labeling method based on prior probability | |
| Chen et al. | Origin-aware location prediction based on historical vehicle trajectories | |
| CN110619090B (en) | Regional attraction assessment method and device |
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 |