CN108133185A - The method and system of pedestrian's relationship is judged based on track data - Google Patents

The method and system of pedestrian's relationship is judged based on track data Download PDF

Info

Publication number
CN108133185A
CN108133185A CN201711397361.4A CN201711397361A CN108133185A CN 108133185 A CN108133185 A CN 108133185A CN 201711397361 A CN201711397361 A CN 201711397361A CN 108133185 A CN108133185 A CN 108133185A
Authority
CN
China
Prior art keywords
individual
pedestrian
distance
relationship
track data
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
Application number
CN201711397361.4A
Other languages
Chinese (zh)
Other versions
CN108133185B (en
Inventor
李焱
杜萍
郑向伟
刘弘
张建新
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shandong Normal University
Original Assignee
Shandong Normal University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Shandong Normal University filed Critical Shandong Normal University
Priority to CN201711397361.4A priority Critical patent/CN108133185B/en
Publication of CN108133185A publication Critical patent/CN108133185A/en
Application granted granted Critical
Publication of CN108133185B publication Critical patent/CN108133185B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/50—Context or environment of the image
    • G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
    • G06V20/53—Recognition of crowd images, e.g. recognition of crowd congestion
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/22—Matching criteria, e.g. proximity measures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/20—Analysis of motion
    • G06T7/246—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/40—Scenes; Scene-specific elements in video content
    • G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items
    • G06V20/42—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items of sport video content
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10016—Video; Image sequence
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30196—Human being; Person
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30241—Trajectory

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Evolutionary Computation (AREA)
  • Evolutionary Biology (AREA)
  • General Engineering & Computer Science (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Computational Linguistics (AREA)
  • Software Systems (AREA)
  • Traffic Control Systems (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses a kind of method and system that pedestrian's relationship is judged based on track data, the method includes：Establish evacuation simulated scenario；It reads the track data of pedestrian and is pre-processed；Based on the track data, candidate set is determined according to the distance between pedestrian and direction character；Determine the relative position of pedestrian's individual in the scene in candidate set；Relative position based on pedestrian's individual in the scene carries out cluster analysis, and group is divided according to the close and distant distance of relationship.The present invention can judge the groupuscule of close relation in crowd, and for crowd evacuation, rehearsal provides important foundation safely, can also detect building evacuation performance, optimize practical evacuation process and improve evacuation efficiency.

Description

The method and system of pedestrian's relationship is judged based on track data
Technical field
The invention belongs to the behavior understanding fields of moving target more particularly to one kind to judge pedestrian's relationship based on track data Method and system.
Background technology
Although current traffic convenience, there are many alternative mode of transportation, and diversification shape is also presented in the trip mode of people State, but manner of walking is still essential, and is irreplaceable in many situations.For example, most feelings The beginning and end gone on a journey under condition is completed with walking, for another example many public places as subway station, stadiums, automobile or Railway station, market, campus etc. not only need walking, and the congestion of pedestrian also takes place frequently.And the public environment of walking, mostly It is crowded, especially often there is the situation that flow of the people is big, density of stream of people is high in peak period or festivals or holidays, this just hides very big Security risk.Therefore, carry out the behavioral study of pedestrian group, behavioral mechanism, the feature of pedestrian group are understood, to pedestrian is instructed to hand over Logical, improvement building safety facility, formulation emergency evacuation strategy etc. are significant.
With the rapid growth of urban population, people also increasingly pay close attention to the public safeties such as stream of people's evacuation；It calculates simultaneously The rapid development of machine technology also provides more easily method for this kind of research.Stream of people's video is shot with camera, passes through video Track data is extracted, computerized algorithm analysis track data is recycled, is exactly current master to study behavioural characteristic of people etc. Stream method.
In practical applications suitable evacuation prediction scheme is obtained usually using the mode of evacuation experiment.Which has specific aim By force, the features such as informative.However, the problem of big etc. inevitable is put into due to that can not ensure, test there are personnel safety, Computer Simulation becomes the most efficient method of crowd evacuation under research accident.But, by analyzing track data, to push away The research of row interpersonal relation is fewer in disconnected crowd, and with clustering method, simple to analyze data to judge interpersonal relationships Research with regard to less.
But sociologist's research shows that, during crowd movement, pedestrian is often according to the close and distant knot of social relationships It advances jointly into groupuscule one by one；Moreover, these groupuscules in close relations (are mostly family, lovers, classmate, colleague, trip Row group etc.) mostly it is parallel in a row or in modes such as multichannel columns (Main Basiss number how much to form traveling mode) to advance；Together When, during crowd advances, the behavior of groupuscule can usually influence the speed of crowd's entirety.So, pass through the meters such as cluster Calculation machine technology judges pedestrian's relationship in crowd, divides groupuscule, just can be more true to nature utilize Computer Simulation crowd evacuation.
The groupuscule of relationship and formation in the method for existing emulation crowd evacuation between non-consideration crowd is to fortune Dynamic influence.Even if some documents possess some special knowledge to groupuscule or microcommunity phenomenon, but only simple this kind of phenomenon are ground Study carefully, also without and emulation crowd evacuation algorithm combine.For example, current mainstream intends crowd evacuation in microcosmic upper mold The social force model that best algorithm model --- Helbing and Molnar is proposed, the model is according to the simulation of Newtonian mechanics formula Crowd behaviour, although some phenomenons can be reappeared well, such as the arch phenomenon in " more thinking fast slower instead " and exit. But the behavior of pedestrian of its over-simplification, the social relationships between pedestrian and the groupuscule being consequently formed are not accounted for fortune The interpersonal relationships how judged between pedestrian is not more discussed in dynamic influence.
Therefore, it in order to solve the problems, such as simulation crowd evacuation more true to nature, needs to consider a kind of method, it can be from track number The such as close and distant degree of general relationship between pedestrian is analyzed in, so that the later stage can more accurately study group behavior feature, is reached To optimization evacuation process and the purpose of raising evacuation efficiency.
Invention content
To overcome above-mentioned the deficiencies in the prior art, the present invention provides a kind of sides that pedestrian's relationship is judged based on track data Method and system according to crowd's track data from video extraction, judge pedestrian's relationship with clustering algorithm, carry out crowd's grouping, Crowd evacuation is emulated, the natural situation such as crowd's grouping during being evacuated so as to simulation more true to nature.
To achieve the above object, the present invention adopts the following technical scheme that：
A kind of method that pedestrian's relationship is judged based on track data, is included the following steps：
Step 1：Establish evacuation simulated scenario；
Step 2：It reads the track data of pedestrian and is pre-processed；
Step 3：Based on the track data, candidate set is determined according to the distance between pedestrian and direction character；
Step 4：Determine the relative position of pedestrian's individual in the scene in candidate set；
Step 5：Relative position based on pedestrian's individual in the scene carries out cluster analysis, is divided according to the close and distant distance of relationship Group.
Further, the step 3 includes：
Step 3.1：Computing unit is divided according to the corresponding coordinate range of track data；
Step 3.2：For each computing unit, candidate individual is obtained according to the Euclidean distance between individual；
Step 3.3：Calculate the tight ness rating between candidate individual pair；
Step 3.4：The track angle of candidate individual pair in each computing unit is calculated, candidate is determined according to the angle Group；
Step 3.5：For the candidate set of acquisition, merger is carried out according to relation transmission.
Further, the step 3.3 includes：
Step 3.3.1：The Euclidean distance between each two candidate individual is calculated, if less than distance threshold, labeled as effective Distance；
Step 3.3.2：It calculates between described two candidate individuals in the average effective distance of whole computing units, it is taken to fall Number is the tight ness rating of each two candidate individual after normalization.
Further, the step 3.4 includes：
Step 3.4.1：Make the geometric locus of the candidate individual of each computing unit；
Step 3.4.2：For each candidate individual pair, segmentation calculates the angle of line-to-line；
Step 3.4.3：The average angle of entire computing unit is calculated, if no more than movement threshold in the same direction, is labeled as having Imitate angle；
Step 3.4.4：Ask this to individual other units effective angle, if effectively angle sum is more than unit sum Half, then it is candidate set to individual to mark this.
Further, the step 4 includes：
Step 4.1：The enhancing Hausdorff distance between pedestrian's individual in each computing unit is calculated, if no more than threshold value, It is effective computing unit to determine the computing unit；
Step 4.2：Calculating enhances the sum of Hausdorff distance in each computing unit, then divided by effective unit number, if small In threshold value, then the relative position of each pedestrian's individual in the scene is denoted as the pedestrian and coordinate is recorded in the middle position of the computing unit, no Then, labeled as adjustment group；
Step 4.3：Adjustment group with neighbours' candidate set of average distance minimum is merged, step 4.1-4.2 is repeated, if again The secondary adjustment group that is marked as then chooses the minimum enclosed rectangle center of whole position curve of the group membership in the unit as this The position of member in the scene.
Further, the step 5 includes：
Step 5.1：According to the position of pedestrian's individual in the scene, the distance between pedestrian's individual table is established；
Step 5.2：The local density centered on each pedestrian's individual is calculated, and a group spacing is determined according to local density From；
Step 5.3：According to local density and component distance, the clustering cluster heart and its number are automatically determined；
Step 5.4：Group is divided according to the close and distant distance of relationship.
Further, the step 5.2 includes：
Step 5.2.1：Setting calculates the individual amount in the range of pedestrian's individual certain radius centered on pedestrian's individual, That is the local density of pedestrian individual；
Step 5.2.2：For individual bigger than itself local density there are 2 and above in radius, select from this The distance of individual nearest individual is as group distance；Otherwise the distance of farthest individual is selected as group distance.
Further, the step 5.3 includes：
Step 5.3.1：The local density of individual and group distance are subjected to product, and descending arranges；
Step 5.3.2：It draws by ordinate value, serial number abscissa value of product, calculates the oblique of adjacent point-to-point transmission successively Rate and the angle for calculating two line segments, ask for the ratio with next angle, when the ratio is not more than 1/2, form at this time larger The angular vertex of angle is the last one cluster heart point, records its serial number.
Further, the step 5.4 includes：Divide the non-cluster heart individual behind two-stage merger, be according to intimate distance first Radius merger close relationship group, then more than intimate distance and less than or equal to the individual of personal distance, be classified as conventional relationship group, remain Remaining individual unifying identifier is isolated individual.
Second purpose according to the present invention is based on what track data judged pedestrian's relationship the present invention also provides a kind of System including memory, processor and stores the computer program that can be run on a memory and on a processor, the processor The method is realized when performing described program.
Third purpose according to the present invention, the present invention also provides a kind of computer readable storage mediums, are stored thereon with Computer program performs the method that pedestrian's relationship is judged based on track data when the program is executed by processor.
Beneficial effects of the present invention
1st, judge that the method for pedestrian's relationship has thought better of the distance and rail of pedestrian track point the present invention is based on track data Reaction of the similitude of mark to interpersonal relationships in itself, by analysis of key frame and prolonged data comparison, has judged pedestrian Between relationship and accordingly with the clustering algorithm based on distance and density, the groupuscule of close relation has been marked off, for crowd Evacuation safety rehearsal provides important foundation, can also detect building evacuation performance, optimizes practical evacuation process and improves and dredges Dissipate efficiency；
2nd, pedestrian's relationship judgment method of the invention is suitable for the regions such as large-scale plane venue, and is more suitably applied to big Scale crowd can expand to the plane domains such as the square of multiple outlets.
Description of the drawings
The accompanying drawings which form a part of this application are used for providing further understanding of the present application, and the application's shows Meaning property embodiment and its explanation do not form the improper restriction to the application for explaining the application.
Fig. 1 is the flow diagram of the method that pedestrian's relationship is judged based on track data of the present invention；
Fig. 2 is sectional drawing (x coordinate and the y seats for the two individual partial traces data extracted in the slave video of the present invention Mark)；
Fig. 3 is 3 individual partial traces coordinate diagrams in crowd of the invention；
Fig. 4 is two individual movement angular separation schematic diagrames in crowd of the invention；
Fig. 5 is local density's schematic diagram of the individual I of the present invention；
Fig. 6 is the group distance schematic diagram of the present invention；
Fig. 7 be the present invention automatically determine cluster heart schematic diagram；
Fig. 8 is the schematic diagram before the crowd of the present invention divides；
Fig. 9 is the schematic diagram of the groupuscule of the present invention.
Specific embodiment
It is noted that described further below be all exemplary, it is intended to provide further instruction to the application.It is unless another It indicates, all technical and scientific terms used herein has usual with the application person of an ordinary skill in the technical field The identical meanings of understanding.
It should be noted that term used herein above is merely to describe specific embodiment, and be not intended to restricted root According to the illustrative embodiments of the application.As used herein, unless the context clearly indicates otherwise, otherwise singulative It is also intended to include plural form, additionally, it should be understood that, when in the present specification using term "comprising" and/or " packet Include " when, indicate existing characteristics, step, operation, device, component and/or combination thereof.
In the absence of conflict, the feature in the embodiment and embodiment in the application can be combined with each other.
Embodiment one
Present embodiment discloses a kind of method that pedestrian's relationship is judged based on track data, as shown in Figure 1, including following step Suddenly：
Step 1：Establish evacuation simulated scenario；
Specifically, the rectangle plane region of long 30m, width 20m are built as evacuation simulated scenario, which goes out including 3 Mouth (in order to which effect is more true to nature, establishing threedimensional model).
Step 2：It reads the track data of pedestrian and is pre-processed；
Step 2.1：It analyzes and reads in track data file (as shown in Figure 2, it is shown that the partial traces number of two pedestrians According to), reference number of a document is 1~N；
Step 2.2：Track data is defined to integrate as DP={ dpi, i=1,2 ..., N }, dpi=(dpi1,dpi2,…,dpin) table Registration is according to No. i-th individual data of concentration, dpijNo. i-th individual jth item data is represented, including serial number, coordinate, flag bit etc..
Demographic data collection P={ pi, i=1,2 ..., N }, pi=(pi1,pi2,…,pim) represent i-th of individual in data set, pijRepresent j-th of attribute of i-th of individual.
Step 3：Based on the track data, candidate set is determined according to the distance between pedestrian and direction character；
Step 3.1：Computing unit is divided according to the corresponding coordinate range of track data；
Data set is divided, the trajectory coordinates read in step (2), (is recorded according to total step number from first position The last item position records, and every record regards a step as) it is divided into multiple computing units.Preferably, it is divided according to every 5 percent The rule of one computing unit is divided into 20 computing units, it is assumed that a total of 4000 step, then every 200 step, one unit, and accordingly The data of each file are divided into 20 parts, and 1~ns of number (ns=20), then No. i-th individual is in the 1st article of kth computing unit The position of record such as formula (1)：
Step 3.2：For each computing unit, candidate individual is obtained according to the Euclidean distance between individual；
Euclidean distance between asking for individual with the middle position coordinate of each computing unit calculates the distance (two two-by-two between individual The distance of individual is onlyed demand once, and individual sum is N here), with dist (dpi,dpj) represent；Overall average distance is sought again, With(such as shown in formula (2)) represents；WhenWhen, by giving meter digital, put 1 (meter digital just Value is zero) to mark this to individual, i.e. dpio=1, dpjo=1；The meter digital and d of each individual is counted againisumIf disum More than the half (ns/2) of unit sum, then the individual mark is candidate individual.
Step 3.3：Calculate the tight ness rating between candidate individual pair；
The serial number of candidate individual is added in candidate individual serial number collection ORD={ ordi, i=1,2 ..., mm }, mm is candidate The number of body；The Euclidean distance dist (dp between each two candidate individuali,dpj) same distance threshold value d1(intimate distance, according to after The inner regulation in face " social distance of people ", d1Take 0.3m) it compares, if dist (dpi,dpj) ＜ d1, then the distance for effectively away from From, and by giving meter digital, 1 (initial value of meter digital is zero) is put to mark this to candidate individual, then count each candidate The meter digital and ds of bodyiAnd the sum of effective distance, then calculate average effective distance total in whole units (such as shown in formula (3)) then takes its inverse, is finally normalized, obtains the tight ness rating between each two individualWith rel (pi,pj) represent (such as shown in formula (4))；Descending arrangement tight ness rating again, and update serial number collection ORD。
The social distance of people：What American Anthropologist Edward's Hall doctor published in 1966《Hiding dimension》One In book, four kinds of distances are divided so that relationship is close and distant, have been intimate distance, personal distance, sociodistance and public distance respectively.This four The value range of kind distance is as follows：
Step 3.4：The track angle of candidate individual pair in each computing unit is calculated, candidate is determined according to the angle Group；
The geometric locus (as shown in Figure 3) of the candidate individual of each unit is made, from serial number collection ORD (according to tight Density descending arranges) individual is chosen successively to serial number, (line, is defined as one between the adjacent two positions of i.e. same individual for segmentation Section) the angle a that calculates between two lines (i.e. a pair of of adjacent body, synchronization line segment or extended line) (asks angle, such as formula using slope (5) shown in, schematic diagram is as shown in Figure 4), then angle is averaging to the angle of entire unit, if no more than 3 degree (movements in the same direction Threshold value, Clark Mcphail [U.S.], 1982,《Using Film to Analyze Pedestrian Behavior》In carry Go out), then labeled as effective angle, and count, then ask this to individual other units effective angle, if effectively angle sum More than the half (ns/2) of unit sum, then it is candidate set to individual to mark this.
Tan (a)=(k1-k2)/(1+k1 × k2) (5)
Step 3.5：For the candidate set of acquisition, merger is carried out according to relation transmission；
According to relation transmission merger candidate set, such as individual i and individual j is in same candidate set, and individual j and individual k In same candidate set, then two groups are merged into a candidate set.
Step 4：Determine the relative position of candidate set individual in the scene；
Step 4.1：Effective computing unit is determined according to the distance between candidate set individual；
It determines effective record count, takes first item record of the candidate set in each unit successively, calculate the person of outstanding talent of enhancing one by one Si Duofu (hausdorff) distances (such as formula (6) shown in), if the value no more than threshold value (average Euclidean between the group individual away from From), then it is assumed that the relative position of the group is constant, and the unit is marked to add 1 for effective unit and effective meter digital, is otherwise not added with.
Step 4.2：According to the relationship between candidate set and neighbours' candidate set, candidate set individual in the scene opposite is judged Position；
Determine adjustment group, the Hausdorff distance of the group obtained in every unit summed, then divided by effective unit number, if The middle position of each member of the group then being taken to record less than threshold value (Hausdorff distance with the enhancing of neighbours' group), (middle position record determines such as Formula (9)) coordinate for the position of corresponding member in the scene, otherwise, labeled as adjustment group.
Adjustment group is merged with neighbours' candidate set of average distance minimum, recalculate Hao Siduofu (hausdorff) away from From, same threshold value comparison again, whole position curve of the group membership in the unit is chosen if adjustment group is still marked as Position of the minimum enclosed rectangle center as the member in the scene.
Step 5：Groupuscule is divided based on the clustering algorithm of distance and density；
Step 5.1：According to the position of individual in the scene, establish person-to-person apart from table；
According to the position of the individual determined in step 4 in the scene, calculate the distance between they and arranged by ascending order It is stored as apart from table.
Step 5.2：The local density centered on each individual is calculated, and group distance is determined according to local density；
Centered on the individual in table, dr (personal distance takes dr=0.8m here) be radius, calculate its radius model Interior individual amount, i.e. local density are enclosed, uses deniIt represents (such as formula (10), schematic diagram such as attached drawing 5)；It resettles close with part The descending table of descending arrangement is spent, then takes the individual in table successively, if surrounding (in radius) is there are 2 and closeer than it above Big individual is spent, selects to use dist as group distance with a distance from the individual nearest from itiRepresent (such as formula (11), schematic diagram Such as attached drawing 6), otherwise select the distance of farthest individual (can ensure the highest individual of local density in this way as group distance Group distance is also maximum).
Wherein,
Step 5.3：Automatically determine the clustering cluster heart and its number；
The local density den for portion's candidate individual of first demanding perfectioniWith its group distance distiProduct, proi=deni× disti, and descending arrangement update serial number collection ORD, then manifold PRO={ proi, i ∈ ORD }, it meets sequence pro1≥pro2≥ pro3≥…≥pron(n is the number of candidate individual)；Using PRO as the longitudinal axis, 1~n of serial number is horizontal axis, they can be mapped For a series of point in one quadrant of rectangular coordinate system, there are one apparent jumps between the cluster heart and the non-cluster heart.Then in reference axis The line segment slope of 2 points of compositions, as shown in formula (12)：
If sequentially seeking the slope of line segment between consecutive points, such as shown in formula (13)：
kij=proj-proi(j-i=1) (13)
Slope deposit manifold K={ kij, j=i+1, i ∈ ORD }, k is taken successivelyijItem k adjacent theretojj+1Calculate two line segments If angle (as shown in Figure 7), and ask adjacent angle poor current poor when being not less than adjacent poor 2 times, records j values, then the time before j It is the cluster heart to select individual, and cluster calculation mesh is j.
Step 5.4：Divide the non-cluster heart individual behind two-stage merger, be first radius merger close relationship according to intimate distance Group, then more than intimate distance and less than or equal to the individual of personal distance, it is classified as conventional relationship group, remaining individual unifying identifier For isolated individual.
Embodiment two
The purpose of the present embodiment is to provide a kind of computing system.
A kind of system that pedestrian's relationship is judged based on track data, including memory, processor and storage on a memory And the computer program that can be run on a processor, the processor realize following steps when performing described program, including：
Step 1：Establish evacuation simulated scenario；
Step 2：It reads the track data of pedestrian and is pre-processed；
Step 3：Based on the track data, candidate set is determined according to the distance between pedestrian and direction character；
Step 4：Determine the relative position of pedestrian's individual in the scene in candidate set；
Step 5：Relative position based on pedestrian's individual in the scene carries out cluster analysis, is divided according to the close and distant distance of relationship Group.
Embodiment three
The purpose of the present embodiment is to provide a kind of computer readable storage medium.
A kind of computer readable storage medium, is stored thereon with computer program, which performs when being executed by processor Following steps：
Step 1：Establish evacuation simulated scenario；
Step 2：It reads the track data of pedestrian and is pre-processed；
Step 3：Based on the track data, candidate set is determined according to the distance between pedestrian and direction character；
Step 4：Determine the relative position of pedestrian's individual in the scene in candidate set；
Step 5：Relative position based on pedestrian's individual in the scene carries out cluster analysis, is divided according to the close and distant distance of relationship Group.
Each step involved in the device of above example two and three is corresponding with embodiment of the method one, specific embodiment It can be found in the related description part of embodiment one.Term " computer readable storage medium " is construed as including one or more The single medium or multiple media of instruction set；Any medium is should also be understood as including, any medium can be stored, be compiled Code carries the instruction set for being performed by processor and processor is made to perform the either method in the present invention.
Experimental result
As shown in attached drawing 8~9, the environmental information according to extraction establishes the planar rectangular region of 30m*20m, in order to which effect is forced Very, the 3 dimension simulated scenarios with enclosure wall are constructed, there are 50 people, (people is also 3 dimension modules, and color is unified when initial, is all in movement Red can make different colors into after being divided into groupuscule).Comprising 3 outlets in scene, the width each exported is 2m. Fig. 8 is the schematic diagram before crowd divides, and Fig. 9 is the groupuscule schematic diagram after dividing.We judge row with based on track data The method of relationship, by data read in and handle, screening candidate set, determine individual relative position, finally be based on away from From the clustering algorithm with density, groupuscule has been marked off, by being compared with video, it is found that the visual determination with human eye is consistent 's.
The present invention can mark off the groupuscule of close relation automatically using computer, without human intervention, for crowd Evacuation safety rehearsal provides important foundation, can also detect building evacuation performance, optimizes practical evacuation process and improves and dredges Dissipate efficiency；Pedestrian's relationship judgment method of the present invention is suitable for the regions such as large-scale plane venue, and is more suitably applied to advise greatly Mould crowd can expand to the plane domains such as the square of multiple outlets.
It will be understood by those skilled in the art that each module or each step of the invention described above can be filled with general computer It puts to realize, optionally, they can be realized with the program code that computing device can perform, it is thus possible to which they are stored In the storage device by computing device come perform either they are fabricated to respectively each integrated circuit modules or by they In multiple modules or step be fabricated to single integrated circuit module to realize.The present invention is not limited to any specific hardware and The combination of software.
Above-mentioned, although the foregoing specific embodiments of the present invention is described with reference to the accompanying drawings, not protects model to the present invention The limitation enclosed, those skilled in the art should understand that, based on the technical solutions of the present invention, those skilled in the art are not Need to make the creative labor the various modifications or changes that can be made still within protection scope of the present invention.

Claims (10)

  1. A kind of 1. method that pedestrian's relationship is judged based on track data, which is characterized in that include the following steps：
    Step 1：Establish evacuation simulated scenario；
    Step 2：It reads the track data of pedestrian and is pre-processed；
    Step 3：Based on the track data, candidate set is determined according to the distance between pedestrian and direction character；
    Step 4：Determine the relative position of pedestrian's individual in the scene in candidate set；
    Step 5：Relative position based on pedestrian's individual in the scene carries out cluster analysis, and group is divided according to the close and distant distance of relationship Body.
  2. 2. the method for pedestrian's relationship is judged based on track data as described in claim 1, which is characterized in that the step 3 is wrapped It includes：
    Step 3.1：Computing unit is divided according to the corresponding coordinate range of track data；
    Step 3.2：For each computing unit, candidate individual is obtained according to the Euclidean distance between individual；
    Step 3.3：Calculate the tight ness rating between candidate individual pair；
    Step 3.4：The track angle of candidate individual pair in each computing unit is calculated, candidate set is determined according to the angle；
    Step 3.5：For the candidate set of acquisition, merger is carried out according to relation transmission.
  3. 3. the method for pedestrian's relationship is judged based on track data as claimed in claim 2, which is characterized in that the step 3.3 Including：
    Step 3.3.1：The Euclidean distance between each two candidate individual is calculated, if less than distance threshold, labeled as effective distance；
    Step 3.3.2：It calculates between described two candidate individuals in the average effective distance of whole computing units, takes its inverse, It is the tight ness rating of each two candidate individual after normalization.
  4. 4. the method for pedestrian's relationship is judged based on track data as claimed in claim 2, which is characterized in that the step 3.4 Including：
    Step 3.4.1：Make the geometric locus of the candidate individual of each computing unit；
    Step 3.4.2：For each candidate individual pair, segmentation calculates the angle of line-to-line；
    Step 3.4.3：The average angle of entire computing unit is calculated, if no more than movement threshold in the same direction, labeled as effectively folder Angle；
    Step 3.4.4：Ask this to individual other units effective angle, if effectively angle sum is more than two points of unit sum One of, then it is candidate set to individual to mark this.
  5. 5. the method for pedestrian's relationship is judged based on track data as described in claim 1, which is characterized in that the step 4 is wrapped It includes：
    Step 4.1：The enhancing Hausdorff distance between pedestrian's individual in each computing unit is calculated, if no more than threshold value, is determined The computing unit is effective computing unit；
    Step 4.2：Calculating enhances the sum of Hausdorff distance in each computing unit, then divided by effective unit number, if less than threshold Being worth, then the relative position of each pedestrian's individual in the scene is denoted as the pedestrian and records coordinate in the middle position of the computing unit, otherwise, mark It is denoted as adjustment group；
    Step 4.3：Adjustment group with neighbours' candidate set of average distance minimum is merged, repeats step 4.1-4.2, if again by The group membership is then chosen at the minimum enclosed rectangle center of whole position curve of the unit as the member labeled as adjustment group Position in the scene.
  6. 6. the method for pedestrian's relationship is judged based on track data as described in claim 1, which is characterized in that the step 5 is wrapped It includes：
    Step 5.1：According to the position of pedestrian's individual in the scene, the distance between pedestrian's individual table is established；
    Step 5.2：The local density centered on each pedestrian's individual is calculated, and group distance is determined according to local density；
    Step 5.3：According to local density and component distance, the clustering cluster heart and its number are automatically determined；
    Step 5.4：Group is divided according to the close and distant distance of relationship.
  7. 7. the method for pedestrian's relationship is judged based on track data as claimed in claim 6, which is characterized in that the step 5.2 Including：
    Step 5.2.1：Setting calculates the individual amount in the range of pedestrian's individual certain radius centered on pedestrian's individual, i.e., should The local density of pedestrian's individual；
    Step 5.2.2：For individual bigger than itself local density there are 2 and above in radius, select from the individual The distance of nearest individual is as group distance；Otherwise the distance of farthest individual is selected as group distance.
  8. 8. the method for pedestrian's relationship is judged based on track data as claimed in claim 6, which is characterized in that the step 5.3 Including：
    Step 5.3.1：The local density of individual and group distance are subjected to product, and descending arranges；
    Step 5.3.2：It draws by ordinate value, serial number abscissa value of product, calculates the slope of adjacent point-to-point transmission successively simultaneously The angle of two line segments is calculated, asks for the ratio with next angle, when the ratio is not more than 1/2, is formed at this time compared with mitre Angular vertex be the last one cluster heart point, record its serial number；Or
    The step 5.4 includes：Divide the non-cluster heart individual behind two-stage merger, first intimately closed for radius merger according to intimate distance It is group, then more than intimate distance and less than or equal to the individual of personal distance, is classified as conventional relationship group, the remaining unified mark of individual Know for isolated individual.
  9. 9. a kind of system that pedestrian's relationship is judged based on track data including memory, processor and is stored on a memory simultaneously The computer program that can be run on a processor, which is characterized in that the processor realizes that right such as will when performing described program Seek the method described in 1-8.
  10. 10. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that the program is by processor The method that pedestrian's relationship is judged based on track data as described in claim 1-8 is performed during execution.
CN201711397361.4A 2017-12-21 2017-12-21 Method and system for judging pedestrian relationship based on trajectory data Expired - Fee Related CN108133185B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201711397361.4A CN108133185B (en) 2017-12-21 2017-12-21 Method and system for judging pedestrian relationship based on trajectory data

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201711397361.4A CN108133185B (en) 2017-12-21 2017-12-21 Method and system for judging pedestrian relationship based on trajectory data

Publications (2)

Publication Number Publication Date
CN108133185A true CN108133185A (en) 2018-06-08
CN108133185B CN108133185B (en) 2021-02-23

Family

ID=62392109

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201711397361.4A Expired - Fee Related CN108133185B (en) 2017-12-21 2017-12-21 Method and system for judging pedestrian relationship based on trajectory data

Country Status (1)

Country Link
CN (1) CN108133185B (en)

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109447313A (en) * 2018-09-17 2019-03-08 咪咕文化科技有限公司 A kind of determination method and device of member relation
CN109508657A (en) * 2018-10-29 2019-03-22 重庆中科云丛科技有限公司 Crowd massing analysis method, system, computer readable storage medium and equipment
CN110097074A (en) * 2019-03-19 2019-08-06 中国科学院信息工程研究所 A kind of track of vehicle compression method based on sequence similarity
CN110969483A (en) * 2019-11-29 2020-04-07 支付宝实验室(新加坡)有限公司 Method and device for identifying positions of merchants and electronic equipment
WO2020125370A1 (en) * 2018-12-21 2020-06-25 深圳云天励飞技术有限公司 Relationship analysis method and related product
CN111723616A (en) * 2019-03-20 2020-09-29 杭州海康威视系统技术有限公司 Method and device for measuring personnel relevance
CN112017209A (en) * 2020-09-07 2020-12-01 图普科技（广州）有限公司 Regional crowd trajectory determination method and device
CN112989221A (en) * 2021-02-18 2021-06-18 支付宝(杭州)信息技术有限公司 Position-based family relation analysis method and device
CN115393799A (en) * 2022-09-06 2022-11-25 中国科学技术大学 Group detection method based on pedestrian stepping synchronization effect

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101976353A (en) * 2010-10-28 2011-02-16 北京智安邦科技有限公司 Statistical method and device of low density crowd
CN107016358A (en) * 2017-03-24 2017-08-04 上海电力学院 Intermediate density scene microcommunity real-time detection method
US20170263003A1 (en) * 2011-05-23 2017-09-14 A9.Com, Inc. Tracking objects between images
CN107423480A (en) * 2017-05-19 2017-12-01 山东师范大学 A kind of crowd movement's emulation mode and system
CN107463751A (en) * 2017-08-10 2017-12-12 山东师范大学 A kind of crowd based on DBSCAN clustering algorithms by half is grouped evacuation emulation method and system

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101976353A (en) * 2010-10-28 2011-02-16 北京智安邦科技有限公司 Statistical method and device of low density crowd
US20170263003A1 (en) * 2011-05-23 2017-09-14 A9.Com, Inc. Tracking objects between images
CN107016358A (en) * 2017-03-24 2017-08-04 上海电力学院 Intermediate density scene microcommunity real-time detection method
CN107423480A (en) * 2017-05-19 2017-12-01 山东师范大学 A kind of crowd movement's emulation mode and system
CN107463751A (en) * 2017-08-10 2017-12-12 山东师范大学 A kind of crowd based on DBSCAN clustering algorithms by half is grouped evacuation emulation method and system

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
闻佳等: "实时视频中的车辆运动轨迹的提取和聚类", 《计算机工程与应用》 *

Cited By (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109447313A (en) * 2018-09-17 2019-03-08 咪咕文化科技有限公司 A kind of determination method and device of member relation
CN109508657A (en) * 2018-10-29 2019-03-22 重庆中科云丛科技有限公司 Crowd massing analysis method, system, computer readable storage medium and equipment
CN109508657B (en) * 2018-10-29 2022-04-26 重庆中科云从科技有限公司 Crowd gathering analysis method, system, computer readable storage medium and device
WO2020125370A1 (en) * 2018-12-21 2020-06-25 深圳云天励飞技术有限公司 Relationship analysis method and related product
CN110097074A (en) * 2019-03-19 2019-08-06 中国科学院信息工程研究所 A kind of track of vehicle compression method based on sequence similarity
CN111723616B (en) * 2019-03-20 2023-06-02 杭州海康威视系统技术有限公司 Personnel correlation measurement method and device
CN111723616A (en) * 2019-03-20 2020-09-29 杭州海康威视系统技术有限公司 Method and device for measuring personnel relevance
CN110969483A (en) * 2019-11-29 2020-04-07 支付宝实验室(新加坡)有限公司 Method and device for identifying positions of merchants and electronic equipment
CN110969483B (en) * 2019-11-29 2023-10-10 支付宝实验室(新加坡)有限公司 Method and device for identifying merchant position and electronic equipment
CN112017209A (en) * 2020-09-07 2020-12-01 图普科技（广州）有限公司 Regional crowd trajectory determination method and device
CN112017209B (en) * 2020-09-07 2024-05-31 图普科技（广州）有限公司 Regional crowd track determination method and device
CN112989221A (en) * 2021-02-18 2021-06-18 支付宝(杭州)信息技术有限公司 Position-based family relation analysis method and device
CN115393799A (en) * 2022-09-06 2022-11-25 中国科学技术大学 Group detection method based on pedestrian stepping synchronization effect
CN115393799B (en) * 2022-09-06 2023-11-17 中国科学技术大学 Group detection method based on pedestrian stepping synchronization effect

Also Published As

Publication number Publication date
CN108133185B (en) 2021-02-23

Similar Documents

Publication Publication Date Title
CN108133185A (en) The method and system of pedestrian's relationship is judged based on track data
Zhang et al. MCnet: Multiple context information segmentation network of no-service rail surface defects
CN107103754B (en) Road traffic condition prediction method and system
CN108664999A (en) A kind of training method and its device, computer server of disaggregated model
CN104992191B (en) The image classification method of feature and maximum confidence path based on deep learning
CN110379020A (en) A kind of laser point cloud painting methods and device based on generation confrontation network
CN106709478A (en) Pedestrian image feature classification method and system
CN109816772A (en) A kind of processing method quickly generating virtual reality scenario by CAD house type file
CN107451183B (en) Knowledge Map construction method based on text cluster thought
CN113378891A (en) Urban area relation visual analysis method based on track distribution representation
CN103440501A (en) Scene classification method based on nonparametric space judgment hidden Dirichlet model
CN109919112B (en) Method for detecting distribution and counting of flowing crowds in complex scene
CN110096961A (en) A kind of indoor scene semanteme marking method of super-pixel rank
CN110166851B (en) Video abstract generation method and device and storage medium
CN108154156A (en) Image Ensemble classifier method and device based on neural topic model
CN109978074A (en) Image aesthetic feeling and emotion joint classification method and system based on depth multi-task learning
Huang et al. Transmixer: A hybrid transformer and cnn architecture for polyp segmentation
Yuan et al. Research on simulation of 3D human animation vision technology based on an enhanced machine learning algorithm
CN116721229B (en) Method, device, equipment and storage medium for generating road isolation belt in map
Sheng et al. A faster R-CNN based high-normalization sample calibration method for dense subway passenger flow detection
Zhang et al. Prototype-based clustered federated learning for semantic segmentation of aerial images
CN102129426B (en) The method of display character relation and device
CN113240265B (en) Urban space division method based on multi-mode traffic data
CN110058945A (en) The accelerating algorithm of Large Scale Graphs parallel computation max-flow based on cutpoint splicing mechanism
CN105678346B (en) A kind of object matching search method based on space clustering

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
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20210223

Termination date: 20211221

CF01 Termination of patent right due to non-payment of annual fee