CN106355605B - Group movement consistency filter method - Google Patents
Group movement consistency filter method Download PDFInfo
- Publication number
- CN106355605B CN106355605B CN201610728267.1A CN201610728267A CN106355605B CN 106355605 B CN106355605 B CN 106355605B CN 201610728267 A CN201610728267 A CN 201610728267A CN 106355605 B CN106355605 B CN 106355605B
- Authority
- CN
- China
- Prior art keywords
- group
- movement
- individual
- consistency
- motion
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in 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
- G06T2207/30232—Surveillance
-
- 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
- G06V20/44—Event detection
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The present invention relates to a kind of group movement consistency filter methods, and initial population motion profile set of segments is obtained from intensive scene video sequence;Group Consistency motion feature is obtained according to group movement path segment set, and select group movement orientation consistency beyond given threshold as Candidate Motion group according to Group Consistency motion feature, calculate the Movement consistency direction of Candidate Motion group, and determine the consistent individual collections of group movement, the motion unit in individual collections except correlation valid interval range consistent with current individual convergent movement is rejected, and is labeled as discrete individual;Judge whether the movement velocity for rejecting the individual collections after discrete individual converges on constant, if so, obtaining filtered motion profile segment.It therefore, can be by removal inconsistent in group movement, to improve crowd surveillance performance in intensive scene.
Description
Technical field
The present invention relates to group movement image processing methods, consistent more particularly to the group movement in a kind of intensive scene
Property filter method.
Background technique
Intelligent video monitoring has great help to monitoring efficiency is improved, it, which is reduced, wants a large amount of of artificial treatment
Data, and only concentrate on some specific parts, to eliminate a large amount of unrelated data, mitigate behaviour significantly
The work load of work person, avoid due to keep a close watch on for a long time display bring it is tired and thus caused by neglect, improve
The validity of monitoring.In recent years, China's economy is grown rapidly, with the proposition of cultural power strategy, public participation social activities
Enthusiasm be continuously improved, the various crowd massing phenomenons in public place frequently occur, and the following security hidden trouble is increasingly
It is prominent.In city peak period on and off duty, crowds' congested area such as subway station, bus station, easily get congestion the accident of trampling.In order to
Public domain is comprehensively and effectively monitored, relevant departments are mounted with a large amount of monitoring camera in city each region.Prison
Control task relies primarily on duty personnel and directly observes video to complete.With the expansion of monitoring area range, Security Personnel's
Working strength also increases with it.In intensive video scene, the consistent athletic group of video based on computer vision is detected as supervising
Actively discovering for group's sexual behaviour provides new technological means in control video.
Population analysis granularity can be divided into individual, group and entire crowd.Certain methods are by proposing two consistent neighbours
Invariant is partitioned into the consistent athletic group in crowd.On the basis of initial consistent athletic group detection, certain methods are first
Learn group's conversion priori out using Markov Chain, then adjusts crowd surveillance result.Based on interaction society between group member
Characteristic is learned, certain methods carry out crowd surveillance by hierarchical clustering.Existing method consider group individual between concertedness and
Consistent characteristic is moved, still, the attribute difference between consistent athletic group, especially group movement consistency is not accounted for and describes.
The instantaneous consistent direction of motion of group, helps to analyze and understand the row between group as a kind of group movement attribute
For.Certain methods demonstrate group movement attribute consistent movement differential and effective in terms of identifying intensive scene between distinguishing group
Property.But due in intensive scene Crowds Distribute it is extensive, motion feature not necessarily table between the consistent motion unit of video capture
It is now with uniformity.The instantaneous consistent direction of motion of group is for due to motion unit tracking and video capture angle bring
Motion feature error has robustness, group movement consistency is helped to improve, to improve crowd surveillance in intensive scene
Energy.Currently, not yet there is the group movement consistency filter method optimized towards crowd surveillance performance in intensive scene.
Summary of the invention
Based on this, it is necessary to provide the group movement consistency filter method in a kind of intensive scene.
A kind of group movement consistency filter method, comprising the following steps:
Step A, initial population motion profile set of segments is obtained from intensive scene video sequence;
Step B, Group Consistency motion feature is obtained according to the group movement path segment set, and according to the group
Body consistency motion feature selects group movement orientation consistency beyond given threshold as Candidate Motion group;
Step C, the Movement consistency direction of the Candidate Motion group is calculated, and determines the consistent individual collection of group movement
It closes;
Step D, it rejects in the individual collections except correlation valid interval range consistent with current individual convergent movement
Motion unit, and be labeled as discrete individual;
Step E, judge whether the movement velocity for rejecting the individual collections after the discrete individual converges on constant, if so,
Then obtain filtered motion profile segment.
It is described in one of the embodiments, that Group Consistency movement is obtained according to the group movement path segment set
Feature, and select group movement orientation consistency beyond given threshold as candidate according to the Group Consistency motion feature
The step of athletic group includes:
Group Consistency quantization means are obtained according to the group movement path segment set, and form group movement characteristic
Quantization means;
It is indicated to select group movement consistency beyond given threshold as candidate according to the group movement property quantification
Athletic group.
It is described in one of the embodiments, that Group Consistency quantization is obtained according to the group movement path segment set
It indicates, and forms the step of group movement property quantification indicates and include:
The group movement consistency quantization means for extracting different motion Direction interval covering motion unit set, form group
Kinetic characteristic quantization means.
In one of the embodiments, further include:
Extract group's fortune of the group movement consistency quantization means of different motion Direction interval covering motion unit set
Dynamic consistency histogram, the section of the group movement consistency histogram are the section model where individual in population movement angle
It encloses, interval range is any angle section set for covering [- π, 0] ∪ [0, π].
It is described in one of the embodiments, that selection group movement consistency is indicated according to the group movement property quantification
Beyond given threshold as Candidate Motion group the step of include:
According to individual movement consistency C in all directions section of group movement consistency histogram and all directions section
The linear and nonlinear function h for covering the number N of motion unit selects group movement consistency Direction interval.
The Movement consistency direction for calculating the Candidate Motion group in one of the embodiments, and determine group
Body move consistent individual collections the step of include:
Using formula c (vi)=< vi, vdir>/||vi||·||vdir| | calculate the consistent correlation of group movement, wherein vi
It is all individual movement speed, v in current groupdirIt is group movement consistency direction.
It is described in one of the embodiments, to reject correlation consistent with current individual convergent movement in the individual collections
Motion unit except valid interval range, and the step of being labeled as discrete individual includes:
The movement consistent correlation valid interval upper bound is that individual i movement velocity and current group move in athletic group
The velocity correlation coefficint minimum value min=minimum (c (v of consistency direction speedi)) and variances sigma linear function low (v,
L)=min+ ε σ;
It is described to move consistent correlation valid interval lower bound as individual i movement velocity in athletic group and current group movement
The velocity correlation coefficint maximum value max=maximize (c (v of consistency direction speedi)) and variances sigma linear function up (v,
L)=max+ ε σ, ε is constant term;
Motion unit beyond the Movement consistency correlation valid interval upper bound and lower bound range is discrete individual.
It is unanimously related to current individual convergent movement in the rejecting individual collections in one of the embodiments,
After the step of motion unit except property valid interval range further include:
The maximal cover radius of individual collections consistent in athletic group is greater than discrete of the consistent correlation of minimum movement
The individual collections of body addition current group;
Calculate the movement velocity of current kinetic individual in population set.
Whether the constant of motion for judging the individual collections after the rejecting discrete individual in one of the embodiments,
The step of converging on constant include:
Judge whether the movement velocity of current kinetic individual in population set converges on constant φ.
If the movement velocity for rejecting the individual collections after the discrete individual in one of the embodiments, does not converge on often
Number then executes step B-D again, until the movement velocity of current kinetic individual in population set converges on constant.
Above-mentioned group movement consistency filter method obtains initial population motion profile piece from intensive scene video sequence
Duan Jihe;Group Consistency motion feature is obtained according to the group movement path segment set, and consistent according to the group
Property motion feature select group movement orientation consistency to be used as Candidate Motion group beyond given threshold, calculate described candidate transport
The Movement consistency direction of dynamic group, and determine the consistent individual collections of group movement, it rejects in the individual collections and current
Individual collections move the motion unit except consistent correlation valid interval range, and are labeled as discrete individual;Institute is rejected in judgement
Whether the movement velocity of the individual collections after stating discrete individual converges on constant, if so, obtaining filtered motion profile piece
Section.It therefore, can be by removal inconsistent in group movement, to improve crowd surveillance performance in intensive scene.
Detailed description of the invention
Fig. 1 is the flow chart of group movement consistency filter method;
Fig. 2 is the flow chart of the group movement consistency filter method of another embodiment;
Fig. 3 is the group movement state diagram before filtering;
Fig. 4 is filtered group movement state diagram.
Specific embodiment
To facilitate the understanding of the present invention, a more comprehensive description of the invention is given in the following sections with reference to the relevant attached drawings.In attached drawing
Give preferred embodiment of the invention.But the invention can be realized in many different forms, however it is not limited to herein
Described embodiment.On the contrary, purpose of providing these embodiments is keeps the understanding to the disclosure more saturating
It is thorough comprehensive.
It should be noted that it can directly on the other element when element is referred to as " being fixed on " another element
Or there may also be elements placed in the middle.When an element is considered as " connection " another element, it, which can be, is directly connected to
To another element or it may be simultaneously present centering elements.Term as used herein " vertical ", " horizontal ", " left side ",
" right side " and similar statement are for illustrative purposes only.
Unless otherwise defined, all technical and scientific terms used herein and belong to technical field of the invention
The normally understood meaning of technical staff is identical.Term as used herein in the specification of the present invention is intended merely to description tool
The purpose of the embodiment of body, it is not intended that in the limitation present invention.Term " and or " used herein includes one or more phases
Any and all combinations of the listed item of pass.
As shown in Figure 1, being the flow chart of group movement consistency filter method.
A kind of group movement consistency filter method, comprising the following steps:
Step A, initial population motion profile set of segments is obtained from intensive scene video sequence.
When being detected to the occasion that the crowd is dense, can according to the movement tendency of crowd movement walking direction crowd,
Therefore, exception whether can occur according in the current crowd of motion determination of crowd, e.g., direction safety is in the direction of falling in a swoon of crowd
Channel then can determine whether out that current occasion is likely to occur danger.If the direction of motion of crowd includes multiple, and transport in all directions
Dynamic individual amount is not in too many differences, then it is believed that current occasion is in normal condition.It is then desired to be carried out to crowd
When detection, the path segment set of group movement can be obtained from intensive scene video sequence, it in this way can be in a period of time
Group movement detected.
Step B, Group Consistency motion feature is obtained according to group movement path segment set, and according to Group Consistency
Motion feature selects group movement orientation consistency beyond given threshold as Candidate Motion group.
Step B includes:
Group Consistency quantization means are obtained according to group movement path segment set, and form group movement property quantification
It indicates.
It is indicated to select group movement consistency beyond given threshold as Candidate Motion according to group movement property quantification
Group.
Group Consistency quantization means are obtained according to group movement path segment set, and form group movement property quantification
The step of expression includes:
The group movement consistency quantization means for extracting different motion Direction interval covering motion unit set, form group
Kinetic characteristic quantization means.
It establishes and extracts group movement consistency quantization means model F (f1, f2)=α f1/β·f2It is consistent to obtain group
Property quantization means;f1Indicate group movement consistency, f2Indicate that group movement orientation consistency, α, β are weight coefficient, alpha+beta=
1,0≤α≤1,0≤β≤1;Wherein function f1, f2It is the probability entropy model and triangle cosine function about variables collection x and y, or
The linearly or nonlinearly transformation of probability entropy model and triangle cosine function;Variables collection x and y are individual movement velocity vectorsOr the polar coordinate representation v=(θ, r) or individual movement angle, θ of individual movement speed.Random entropy Yue great group fortune
Dynamic consistency is lower, and triangle cosine function Yue great group Movement consistency is higher.
Group movement consistency filter method further include:
Extract group's fortune of the group movement consistency quantization means of different motion Direction interval covering motion unit set
Dynamic consistency histogram, the section of group movement consistency histogram are the interval range where individual in population movement angle,
Interval range is any angle section set for covering [- π, 0] ∪ [0, π].E.g., including different motion Direction interval covering fortune
The group movement consistency quantization means group movement consistency histogram of dynamic individual collections;Group movement consistency histogram
Section definition is the interval range where individual in population movement angle, and interval range is any of covering [- π, 0] ∪ [0, π]
Angular interval set, angular interval set can be [- π, π], [0, π], [- π, 0], [- π, -0.5 π] ∪ [0.5 π, π], and [- 0.5
π, 0.5 π], [0,0.5 π], [0.5 π, π], [- 0.5 π,-π] or [0, -0.5 π].
Step B indicates to select group movement consistency beyond given threshold as candidate according to group movement property quantification
The step of athletic group includes:
According to individual movement consistency C in all directions section of group movement consistency histogram and all directions section
The linear and nonlinear function h for covering the number N of motion unit selects group movement consistency Direction interval.Linear function can be with
It is h=α C+ β N+ λ CN, wherein α, β, λ are weight coefficient, alpha+beta+λ=1,0≤α≤1,0≤β≤1,0≤λ≤
1。
Incorporated by reference to Fig. 3.
For example, the group movement of current occasion includes 100 individuals, the consistency quantization of this 100 individual movements is obtained
It indicates, and forms the expression of group movement property quantification.Two-dimensional coordinate system is established, individual movement direction is centered on origin to each
Quadrant extends to form vector.Therefore, in same quadrant individual corresponding to each vector be Movement consistency individual.It is assumed that
100 individuals, have 67 in first quartile, have 30 in the second quadrant, 1 in third quadrant, in the
2 of four-quadrant wait it is possible to choose 30 individuals of 67 individuals and the second quadrant of first quartile respectively as two
Select athletic group.Third quadrant and the individual of fourth quadrant are very little, therefore not having generality can not be handled.
Similarly, group movement consistency exceed given threshold, the given threshold according to the individual amount in group movement come
Setting, the object that need to follow processing have generality, that is, need most of individual in processing colony movement.
Also need after stepb it is further to candidate population handled, individual that might not be all in candidate population
The direction of motion be consistent, therefore, it is necessary to be filtered to candidate population.
Step C, the consistent correlation of movement in the Movement consistency direction of Candidate Motion group is calculated, and determines group movement
Consistent individual collections.
Step C includes:
Using formula c (vi)=< vi, vdir>/||vi||·||vdir| | or c (li)=f (li) calculate the consistent phase of group movement
Guan Xing, wherein viIt is all individual movement speed, v in current groupdirIt is group movement consistency direction;liIt is current group
Middle individual space position;Using individual movement speed in current groupOr the letter of space position l=(x, y)
Number c.
Using about individual movement speed polar coordinates v=(θ, r) in selected group movement consistency Direction interval and speed
The function o=(v, θ) of direction θ is spent, selected group movement consistency direction is calculated.Function o=(v, θ) can be movement velocity
V's and direction θ is average etc..
Step D, the fortune in individual collections except correlation valid interval range consistent with current individual convergent movement is rejected
Dynamic individual, and it is labeled as discrete individual.
Step D includes:
Moving the consistent correlation valid interval upper bound is that individual i movement velocity and current group movement are consistent in athletic group
The velocity correlation coefficint minimum value min=minimum (c (v of property direction speedi)) and variances sigma linear function low (v, l)=
min+ε·σ;
Moving consistent correlation valid interval lower bound is that individual i movement velocity and current group movement are consistent in athletic group
The velocity correlation coefficint maximum value max=maximize (c (v of property direction speedi)) and variances sigma linear function up (v, l)=
Max+ ε σ, ε are constant term;
Motion unit beyond the Movement consistency correlation valid interval upper bound and lower bound range is discrete individual.
Step E, judge whether the movement velocity for rejecting the individual collections after discrete individual converges on constant, if so, obtaining
Take filtered motion profile segment.
Movement in rejecting individual collections except correlation valid interval range consistent with current individual convergent movement
After the step of body further include:
The maximal cover radius of individual collections consistent in athletic group is greater than discrete of the consistent correlation of minimum movement
The individual collections of body addition current group;
Calculate the movement velocity of current kinetic individual in population set.
Judge that the step of whether constant of motion for rejecting the individual collections after discrete individual converges on constant includes:
Judge whether the movement velocity of current kinetic individual in population set converges on constant φ.
If the movement velocity for rejecting the individual collections after discrete individual does not converge on constant, step B-D is executed again, directly
Movement velocity to disconnected current kinetic individual in population set converges on constant.
Based on above-mentioned all embodiments, the detailed process of group movement consistency filter method are as follows:
Step 210 obtains initial population motion profile set of segments;
Step 212 obtains Group Consistency motion feature according to group movement path segment set.
Step 214 selects group movement orientation consistency beyond the work of given threshold according to Group Consistency motion feature
For Candidate Motion group.
Step 216, the Candidate Motion individual collections gone out selected by, calculate current group Movement consistency direction.
Step 218, to calculate all individuals of current group unanimously related to the movement in current group Movement consistency direction
Property, and determine the consistent individual collections of group movement.
Step 220 rejects correlation valid interval range consistent with group movement from the consistent individual collections of group movement
Except motion unit, by correlation valid interval range consistent with the movement in group movement consistency direction in discrete individual it
Interior motion unit is added to the individual collections of current group.
Step 222, the movement velocity for calculating individual collections in current group.
If the movement of individual collections converges on constant in step 224, current group, then it is assumed that the filtering of group movement consistency
It completes.
Incorporated by reference to Fig. 3 and Fig. 4.
Specifically, obtaining the group movement path segment set in video sequence, indicated with open circles or dot.According to
Group movement path segment set obtains Group Consistency motion feature, i.e., the direction of motion of each individual in acquisition group, with side
It is indicated to arrow.Selected directions arrow characterizes the roughly the same each individual in direction and is used as Candidate Motion group.If starting point is sky
Heart circle, terminal are only that each individual of arrow is the first Candidate Motion group;Starting point is open circles, and terminal is the arrow of dot
Each individual is the second Candidate Motion group.Using origin as starting point, terminal is that each individual of arrow is discrete individual.It will be each
The starting point of direction arrow is placed in the origin of two-dimensional coordinate system.Individual in same quadrant is considered the same candidate
Athletic group, therefore, the discrete individual can be added in the first Candidate Motion group.
Due to having multiple Candidate Motion individual collections, need to be for further processing to multiple Candidate Motion individual collections,
The discrete individual that the same direction of motion will be not belonging in the individual collections is rejected.So, need to calculate current group fortune
Dynamic consistency direction.All individuals of current group correlation consistent with the movement in current group Movement consistency direction is calculated,
And determine the consistent individual collections of group movement.Correlation consistent with group movement is rejected from the consistent individual collections of group movement to be had
The motion unit except interval range is imitated, correlation consistent with the movement in group movement consistency direction in discrete individual is effective
Motion unit within interval range is added to the individual collections of current group.This makes it possible to will be discrete in individual collections M
Individual, such as b1, c3 ... n2 is rejected from individual collections M.Similarly, the discrete individual in other individual collections is all made of above-mentioned
Method is rejected.Judge that various discrete individual is compared with the Movement consistency correlation valid interval between individual collections again
Compared with will be added in corresponding individual collections with regard to the discrete individual if meeting the individual collections.
After the completion of the rejecting of discrete individual in individual collections is with adding, need to judge again whether filtering is completed,
The movement velocity of individual collections in current group is calculated at this time, and judges that the movement of individual collections in current group converges on constant
Whether constant is converged on, if so, thinking that the filtering of group movement consistency is completed.If it is not, step B-D is then executed again, until
The movement velocity of current kinetic individual in population set converges on constant.
Above-mentioned group movement consistency filter method obtains initial population motion profile piece from intensive scene video sequence
Duan Jihe;Group Consistency motion feature is obtained according to group movement path segment set, and spy is moved according to Group Consistency
Sign selects group movement orientation consistency, as Candidate Motion group, to calculate the movement of Candidate Motion group beyond given threshold
Consistency direction, and determine the consistent individual collections of group movement, it rejects consistent with current individual convergent movement in individual collections
Motion unit except correlation valid interval range, and it is labeled as discrete individual;The individual collection after discrete individual is rejected in judgement
Whether the movement velocity of conjunction converges on constant, if so, obtaining filtered motion profile segment.Therefore, group can be transported
Inconsistent removal in dynamic, to improve crowd surveillance performance in intensive scene.
Each technical characteristic of embodiment described above can be combined arbitrarily, for simplicity of description, not to above-mentioned reality
It applies all possible combination of each technical characteristic in example to be all described, as long as however, the combination of these technical characteristics is not deposited
In contradiction, all should be considered as described in this specification.
The embodiments described above only express several embodiments of the present invention, and the description thereof is more specific and detailed, but simultaneously
It cannot therefore be construed as limiting the scope of the patent.It should be pointed out that coming for those of ordinary skill in the art
It says, without departing from the inventive concept of the premise, various modifications and improvements can be made, these belong to protection of the invention
Range.Therefore, the scope of protection of the patent of the invention shall be subject to the appended claims.
Claims (10)
1. a kind of group movement consistency filter method, which comprises the following steps:
Step A, initial population motion profile set of segments is obtained from intensive scene video sequence;
Step B, Group Consistency motion feature is obtained according to the group movement path segment set, and according to the group one
Cause property motion feature selects direction of motion consistency beyond the group of given threshold as Candidate Motion group;
Step C, the Movement consistency direction of the Candidate Motion group is calculated, and determines the consistent individual collections of group movement;
Step D, the fortune in the individual collections except correlation valid interval range consistent with current individual convergent movement is rejected
Dynamic individual, and it is labeled as discrete individual;
Step E, judge whether the movement velocity for rejecting the individual collections after the discrete individual converges on constant, if so, obtaining
Take filtered motion profile segment.
2. group movement consistency filter method according to claim 1, which is characterized in that described to be transported according to the group
Dynamic rail mark set of segments obtains Group Consistency motion feature, and selects the direction of motion according to the Group Consistency motion feature
Consistency includes: as the step of Candidate Motion group beyond the group of given threshold
Group Consistency quantization means are obtained according to the group movement path segment set, and form group movement property quantification
It indicates;
It is indicated to select group of the Movement consistency beyond given threshold as Candidate Motion according to the group movement property quantification
Group.
3. group movement consistency filter method according to claim 2, which is characterized in that described to be transported according to the group
Dynamic rail mark set of segments obtains Group Consistency quantization means, and forms the step of group movement property quantification indicates and include:
The group movement consistency quantization means for extracting different motion Direction interval covering motion unit set, form group movement
Property quantification indicates.
4. group movement consistency filter method according to claim 3, which is characterized in that further include:
Extract the group movement one of the group movement consistency quantization means of different motion Direction interval covering motion unit set
Cause property histogram, the section of the group movement consistency histogram are the interval range where individual in population movement angle,
Interval range is any angle section set for covering [- π, 0] ∪ [0, π].
5. group movement consistency filter method according to claim 2, which is characterized in that described to be transported according to the group
Dynamic characteristic quantization means select the Movement consistency to include: as the step of Candidate Motion group beyond the group of given threshold
It is covered according to individual movement consistency C in all directions section of group movement consistency histogram and all directions section
The linear and nonlinear function h of the number N of motion unit selects group movement consistency Direction interval.
6. group movement consistency filter method according to claim 1, which is characterized in that described to calculate the candidate fortune
The Movement consistency direction of dynamic group, and the step of determining group movement consistent individual collections includes:
Using formula c (vi)=< vi, vdir>/||vi||g||vdir| | calculate the consistent correlation of group movement, wherein viIt is current
All individual movement speed, v in groupdirIt is group movement consistency direction.
7. group movement consistency filter method according to claim 1, which is characterized in that described to reject the individual collection
Motion unit in conjunction except correlation valid interval range consistent with current individual convergent movement, and labeled as discrete individual
Step includes:
The movement consistent correlation valid interval upper bound is that individual i movement velocity and current group movement are consistent in athletic group
The velocity correlation coefficint minimum value min=minimum (c (v of property direction speedi)) and variances sigma linear function low (v, l)=
min+ε·σ;
The consistent correlation valid interval lower bound of the movement is that individual i movement velocity and current group movement be unanimously in athletic group
The velocity correlation coefficint maximum value max=maximize (c (v of property direction speedi)) and variances sigma linear function up (v, l)=
Max+ ε σ, ε are constant term;
Motion unit beyond the Movement consistency correlation valid interval upper bound and lower bound range is discrete individual.
8. group movement consistency filter method according to claim 1, which is characterized in that reject the individual described
After the step of motion unit in set except correlation valid interval range consistent with current individual convergent movement further include:
The discrete individual that the maximal cover radius of individual collections consistent in athletic group is greater than the consistent correlation of minimum movement is added
Enter the individual collections of current group;
Calculate the movement velocity of current kinetic individual in population set.
9. group movement consistency filter method according to claim 8, which is characterized in that the judgement reject it is described from
The step of whether constant of motion of individual collections after dissipating individual converges on constant include:
Judge whether the movement velocity of current kinetic individual in population set converges on constant φ.
10. group movement consistency filter method according to claim 1, which is characterized in that if rejecting described discrete
The movement velocity of individual collections after body does not converge on constant, then executes step B-D again, until current kinetic individual in population
The movement velocity of set converges on constant.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610728267.1A CN106355605B (en) | 2016-08-25 | 2016-08-25 | Group movement consistency filter method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610728267.1A CN106355605B (en) | 2016-08-25 | 2016-08-25 | Group movement consistency filter method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106355605A CN106355605A (en) | 2017-01-25 |
CN106355605B true CN106355605B (en) | 2018-12-28 |
Family
ID=57854152
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610728267.1A Active CN106355605B (en) | 2016-08-25 | 2016-08-25 | Group movement consistency filter method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106355605B (en) |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107016358B (en) * | 2017-03-24 | 2020-03-20 | 上海电力学院 | Real-time detection method for small group in medium-density scene |
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 |
CN114520024B (en) * | 2022-01-17 | 2024-03-22 | 浙江天科高新技术发展有限公司 | Sequence combination method based on k-mer |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103593656A (en) * | 2013-11-22 | 2014-02-19 | 四川大学 | Locating method for recognizing locomotion patterns of crowds in videos based on local binary pattern |
CN103839065A (en) * | 2014-02-14 | 2014-06-04 | 南京航空航天大学 | Extraction method for dynamic crowd gathering characteristics |
CN106157325A (en) * | 2015-04-07 | 2016-11-23 | 中国科学院深圳先进技术研究院 | Group abnormality behavioral value method and system |
CN106157326A (en) * | 2015-04-07 | 2016-11-23 | 中国科学院深圳先进技术研究院 | Group abnormality behavioral value method and system |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8462987B2 (en) * | 2009-06-23 | 2013-06-11 | Ut-Battelle, Llc | Detecting multiple moving objects in crowded environments with coherent motion regions |
-
2016
- 2016-08-25 CN CN201610728267.1A patent/CN106355605B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103593656A (en) * | 2013-11-22 | 2014-02-19 | 四川大学 | Locating method for recognizing locomotion patterns of crowds in videos based on local binary pattern |
CN103839065A (en) * | 2014-02-14 | 2014-06-04 | 南京航空航天大学 | Extraction method for dynamic crowd gathering characteristics |
CN106157325A (en) * | 2015-04-07 | 2016-11-23 | 中国科学院深圳先进技术研究院 | Group abnormality behavioral value method and system |
CN106157326A (en) * | 2015-04-07 | 2016-11-23 | 中国科学院深圳先进技术研究院 | Group abnormality behavioral value method and system |
Non-Patent Citations (1)
Title |
---|
《Coherent Filtering:Detecting Coherent Motions from Crowd Clutters》;Bolei Zhou et al.;《Proceeding of the IEEE Conference on European Conference on Computer Vision》;20121231;857-871 * |
Also Published As
Publication number | Publication date |
---|---|
CN106355605A (en) | 2017-01-25 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103839065B (en) | Extraction method for dynamic crowd gathering characteristics | |
CN103824070B (en) | A kind of rapid pedestrian detection method based on computer vision | |
EP2864930B1 (en) | Self learning face recognition using depth based tracking for database generation and update | |
CN103198493B (en) | A kind ofly to merge and the method for tracking target of on-line study based on multiple features self-adaptation | |
CN107230267B (en) | Intelligence In Baogang Kindergarten based on face recognition algorithms is registered method | |
CN102622604B (en) | Multi-angle human face detecting method based on weighting of deformable components | |
CN103605983B (en) | Remnant detection and tracking method | |
CN105205455A (en) | Liveness detection method and system for face recognition on mobile platform | |
CN106355605B (en) | Group movement consistency filter method | |
Cao et al. | Abnormal crowd motion analysis | |
CN106339657B (en) | Crop straw burning monitoring method based on monitor video, device | |
CN104091171A (en) | Vehicle-mounted far infrared pedestrian detection system and method based on local features | |
CN105022999A (en) | Man code company real-time acquisition system | |
KR20160109761A (en) | Method and System for Recognition/Tracking Construction Equipment and Workers Using Construction-Site-Customized Image Processing | |
CN103955949A (en) | Moving target detection method based on Mean-shift algorithm | |
Karpagavalli et al. | Estimating the density of the people and counting the number of people in a crowd environment for human safety | |
CN103714697A (en) | Method for identifying and tracking criminal's vehicle | |
CN103824280A (en) | Typhoon center extraction method | |
Janku et al. | Fire detection in video stream by using simple artificial neural network | |
CN107392089A (en) | A kind of eyebrow movement detection method and device and vivo identification method and system | |
CN107358155A (en) | A kind of funny face motion detection method and device and vivo identification method and system | |
CN107358154A (en) | A kind of head movement detection method and device and vivo identification method and system | |
CN104463104A (en) | Fast detecting method and device for static vehicle target | |
CN107368777A (en) | A kind of smile motion detection method and device and vivo identification method and system | |
Miao et al. | Abnormal behavior learning based on edge computing toward a crowd monitoring system |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | 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 |